This is an Open Access document downloaded from ORCA, Cardiff University's institutional repository: http://orca.cf.ac.uk/119344/

This is the author’s version of a work that was submitted to / accepted for publication.

Citation for final published version:

Troll, V, Weis, F, Jonsson, E, Andersson, U, Majidi, S, Hogdahl, K, Harris, C, Millet, Marc-Alban, Chinnasamy, Sakthi, Kooijman, E and Nilsson, K 2019. Global Fe-O isotope correlation reveals magmatic origin of Kiruna-type apatite-iron-oxide ores. Nature Communications 10 , 1712. 10.1038/s41467-019-09244-4 file

Publishers page: https://doi.org/10.1038/s41467-019-09244-4

Please note: Changes made as a result of publishing processes such as copy-editing, formatting and page numbers may not be reflected in this version. For the definitive version of this publication, please refer to the published source. You are advised to consult the publisher’s version if you wish to cite this paper.

This version is being made available in accordance with publisher policies. See http://orca.cf.ac.uk/policies.html for usage policies. Copyright and moral rights for publications made available in ORCA are retained by the copyright holders. 1 Global Fe-O isotope correlation reveals magmatic origin of Kiruna-type

2 apatite-iron-oxide ores

3 1 1,2 1,3 1,4 5 4 Valentin R. Troll , Franz Weis , Erik Jonsson , Ulf B. Andersson , Seyed Afshin Majidi , Karin 1,6 7 8 1,9 2 5 Högdahl , Chris Harris , Marc-Alban Millet , Sakthi Saravanan Chinnasamy , Ellen Kooijman , 3, 10 6 Katarina P. Nilsson

7 8 1Section for Mineralogy, Petrology and Tectonics, Department of Earth Sciences, Uppsala University, Villavägen 16, 9 75236 Uppsala, Sweden 10 2Swedish Museum of Natural History, Dept. of Geosciences, Frescativägen 40, 114 18 Stockholm, Sweden; 11 3Department of Mineral Resources, Geological Survey of Sweden, Villavägen 18, Box 670, 75128 Uppsala, Sweden 12 4Luossavaara-Kiirunavaara AB, Research & Development, FK9, 981 86 Kiruna, Sweden 13 5Geological Survey of Iran, Meraj St, Azadi Sq, 138783-5841 Tehran, Iran 14 6Geology and Mineralogy, Åbo Akademi University, Domkyrkotorget 1, 20500 Turku, Finland 15 7Department of Geological Sciences, University of Cape Town, Rondebosch 7701, Republic of South Africa 16 8School of Earth and Ocean Sciences, Cardiff University, Park Place, Cardiff, CF10 3AT, United Kingdom 17 9National Institute of Technology Rourkela, Department of Earth & Atmospheric Sciences, NIT Rourkela, 769008 18 Odisha, India 19 10 Swedish Ministry of Enterprise and Innovation, Division for Business, Mäster Samuelsgatan 70, 10333 Stockholm, 20 Sweden 21 Corresponding author: [email protected]

22

23 Two sentence summary

24 The origin of giant Kiruna-type iron ores has been debated for nearly 100 years. This study employs

25 extensive stable isotope data from Kiruna-type ores worldwide and magmatic and hydrothermal

26 reference materials to show that iconic Kiruna-type ores originate primarily from ortho-magmatic

27 processes.

28

29 Summary paragraph

30 Kiruna-type apatite-iron-oxide ores are key iron sources for modern industry, yet their origin

31 remains controversial. Diverse ore-forming processes have been discussed, comprising low-

32 temperature hydrothermal processes versus a high-temperature origin from magma or magmatic

33 fluids. We present an extensive set of new and combined iron and oxygen isotope data from

34 magnetite of Kiruna-type ores from Sweden, Chile and Iran, and compare them with new global

35 reference data from layered intrusions, active volcanic provinces, and established low-temperature

Troll et al. NCOMMS-18-08734B 1 36 and hydrothermal iron ores. We show that approximately 80 % of the magnetite from the

56 18 37 investigated Kiruna-type ores exhibit δ Fe and δ O ratios that overlap with the volcanic and

38 plutonic reference materials (> 800 °C), whereas ~20 %, mainly vein-hosted and disseminated

39 magnetite, match the low-temperature reference samples (≤ 400 °C). Thus, Kiruna-type ores are

40 dominantly magmatic in origin, but may contain late-stage hydrothermal magnetite populations that

41 can locally overprint primary high-temperature magmatic signatures.

42

43 Keywords: Kiruna-type apatite-iron-oxide ores, magmatic vs. hydrothermal origin, magnetite.

44 iron isotopes, oxygen isotopes

45

46 Statistics and format:

47 Abstract: 149 Words of 150

48 Main Manuscript: 4284 words + 713 for Figure captions

49 Introduction: 481 words

50 Acknowledgements, Author Contributions and Methods: 1021 words

51 References: 70 out of 70 for Nature Communications

52 Figures: 6 plus 2 in Appendix

53

54

55 Apatite-iron oxide ore is by far the biggest source of iron in Europe and one of the main iron

1 56 sources worldwide . In Europe, these magnetite-dominated ores have traditionally been sourced

57 from two principal regions, the Bergslagen ore province in south central Sweden and the Kiruna-

2 58 Malmberget region in northern Sweden (Fig.1) . The apatite-iron oxide ores from these localities

59 are internationally renowned and similar ores elsewhere are usually referred to as being of Kiruna-

3–5 6 60 type . While the Grängesberg and Kiruna deposits are Palaeoproterozoic in age , similar apatite-

61 iron oxide deposits along the American Cordilleras are much younger and range in age from

7–9 62 Jurassic to Neogene, like the Pliocene El Laco deposit in Chile . Together with the occurrences of

Troll et al. NCOMMS-18-08734B 2 10– 63 Paleozoic apatite-iron oxide ores in Turkey, Iran, and China, and Triassic examples from Korea

13 64 , apatite-iron oxide ores have repeatedly formed across the globe and throughout geological time.

65 The origin of the Kiruna-type apatite-iron-oxide ores remains ambiguous, however, despite a

66 long history of study and a concurrently intense scientific debate. Several fundamentally different

67 modes of formation have been proposed. Today two broad schools of thought prevail, represented

68 by either direct magmatic formation processes, such as segregation or crystallization, or by

69 hydrothermal replacement processes, including hydrothermal precipitation in the sense of iron-

3,4,14–31 70 oxide-copper-gold (IOCG) deposits . Specifically, the discussion revolves around a direct

71 magmatic origin (ortho-magmatic) from volatile- and Fe-P-rich magmas or high-temperature

1,5,18,31 72 magmatic fluids , versus a purely hydrothermal one, where circulating, metal-rich fluids

73 replace original host rock mineralogy with apatite-iron-oxide mineralization at medium to low

4,19,28–30,32 74 temperature . An ortho-magmatic origin is generally understood to be either formation by

75 direct crystallization from a magma or from high-temperature magmatic fluids (e.g. ≥ 800 °C), or

76 via high-temperature liquid immiscibility and physical separation of an iron oxide-dominated melt

77 from a silicate-dominated magma, where the former may subsequently crystallize as a separate

17,18,33–38 78 body . Hydrothermal processes, in turn, encompass transport and precipitation, including

79 replacement-type reactions, by means of aqueous fluids at more moderate to low temperatures

27,29,30,32 80 (typically ≤ 400 °C) . Both the magmatic and the hydrothermal hypotheses are supported in

81 part by field observations, textural relationships, and mineral chemistry, however, petrological field

82 evidence and chemical trends of major and trace elements have frequently been interpreted in

5,14,16–22,28–32 83 different ways . Moreover, many of previous investigations have focused on one case

84 study only and frequently present a range of various data, with individual data sets often being

85 relatively restricted in respect to data volume. What has so far been missing is a broad and decisive

86 geochemical approach to distinguish between these two rival formation hypotheses on an across-

87 deposit scale. To date no systematic stable isotope study employing several distinct Kiruna–type

88 apatite iron oxide ore deposits is available and, importantly, no systematic comparison with

Troll et al. NCOMMS-18-08734B 3 89 accepted magmatic and hydrothermal rock and ore suites has previously been presented in the

90 literature.

91

92 Results

93

94 Scientific rationale and sample selection

95 Here we use the isotopes of iron and oxygen, the two essential elements in magnetite (Fe3O4), on

96 magnetite samples from four major Kiruna-type ore provinces Magnetite is the main iron-bearing

97 component in Kiruna-type deposits and our approach therefore utilizes highly reliable major

98 elements as petrogenetic tracers, as opposed to e.g., traditional minor element approaches that rely

99 on low-concentration constituents in these ores and their host rocks. We present an extensive set of

100 new Fe and O isotope data of magnetite from a suite of world-class Kiruna-type ores from three

101 continents, represented by Sweden, Chile, and Iran and complement these data by a large suite of

102 new comparative data on accepted magmatic and hydrothermal reference samples (54 new Fe-O

103 coupled isotope ratios and another 12 individual ratios, totaling 120 new individual isotope ratios).

104 In addition, the four different regions of Kiruna-type deposits investigated are separated in space

105 and time, as are the extensive suite of volcanic, plutonic and low-temperature reference materials

106 that we employ to define the endmember processes reflected in our ore deposit data (see Fig.1 and

107 Supplementary Table 1). We use these data to address whether Kiruna-type apatite-iron oxide ores

108 form primarily through direct magmatic processes (magma and magmatic fluids) at high

109 temperatures (≥ 800°C), or alternatively, through precipitation from hydrothermal fluids at

110 considerably lower temperatures (≤ 400°C). The comparative aspect of our work is a particular

111 strength and to the best of our knowledge, no other systematic study with such global coverage has

112 been performed to date. In addition, we offer the first substantial data sets for coupled Fe and O

113 isotopes for the world famous Kiruna, Grängesberg, and Bafq deposits, plus new comparative data

5,14,17,18,32 114 for El Laco that are consistent with published data for this deposit . The study’s global

Troll et al. NCOMMS-18-08734B 4 115 extent, its systematic approach, its large data volume, and most crucially, the global Fe-O isotope

116 correlation permits a decisive conclusion, which allows us to advance our understanding of the

117 relationship between magmatic and hydrothermal processes in the genesis of Kiruna-type iron-

118 oxide ore deposits. The use of a broad set of Fe-O isotope data is especially useful as it establishes

119 not only genetic similarities between different deposits but also provides constrains on the

1,5,39 120 formation temperatures and allows us to identify a high- versus a low-temperature origin of

121 individual ore deposits.

122 To facilitate a widely applicable comparison of apatite-iron-oxide ores analyses of iron and

123 oxygen isotope ratios of magnetite from massive apatite-iron-oxide ores from the Kiruna Mining

124 District in northern Sweden (n=14), the Grängesberg Mining District and the nearby Blötberget ore

125 body in central Sweden (n=16), the El Laco district in Chile (n=6) and the Bafq Mining District in

126 central Iran (n=6) (Figs.1 & 2) were performed. These ore provinces and regions are briefly

127 introduced below, and full details are given in Supplementary Note 1.

128 The study of apatite-iron-oxide mineralizations began at the iconic deposits of the Kiruna

3 129 Mining District . The present Kiruna mine at Kirunavaara, is the largest deposit of its type, and is

130 the main supplier of iron ore in Europe. This deposit alone represent a pre-mining reserve of ≥ 2

131 billion tons of high grade ore. The ores in Kiruna and similar Palaeoproterozoic deposits in the

132 district (Supplementary Table 1), are dominated by magnetite and contain between 50-70 % Fe

133 with a P content of, commonly, up to 2 wt. % that is mainly hosted by fluorapatite and subordinate

25 134 monazite-(Ce) . Additional gangue minerals that can occur in small proportions are Mg-rich

135 actinolite, phlogopite, chlorite, titanite, talc, feldspar, quartz, carbonates, sulphides, sulphates and

3,5,25 136 clays .

137 The apatite-iron-oxide ores of the Grängesberg and the nearby Blötberget deposits represent

138 the largest iron ore accumulation in the Bergslagen ore province, a classic mining region in central

40 139 Sweden considered to represent a Palaeoproterozoic continental rift or back arc basin . A historic

140 production of 156 Mt of ore, averaging 60 % Fe and 0.81 % P, is documented from Grängesberg. In

Troll et al. NCOMMS-18-08734B 5 141 addition to iron oxides (dominantly magnetite), the presence of phosphates such as fluorapatite,

142 monazite-(Ce), and xenotime-(Y), together with REE-silicates constitute a potentially significant

1,41 143 resource .

144 The El Laco apatite-iron-oxide ore deposit at the Pico Laco volcanic complex in northern

145 Chile consists of seven individual ore bodies which together comprise ~500 Mt of mainly

5,18,26,30 146 magnetite-dominated ore with an average grade of 60 % Fe . The Plio- to Pleistocene El

147 Laco deposit is part of the young apatite-iron-oxide mineralizations that characterize the eastern

148 high Andes and is separate from the Cretaceous apatite-iron-oxide ores of the so-called Chilean Iron

7,14,17,18 149 Belt .

150 The Bafq Mining District in Central Iran comprises 34 documented iron ore mineralizations

11,42 151 with a total reserve of ~2 billion tons of iron ore with grades between 53 and 65 % . The apatite-

152 iron oxide ores of the Bafq region are coeval with their early Cambrian andesitic and rhyolitic host

43 153 rocks that formed in a volcanic arc setting . Samples from the Bafq Mining District were collected

42 154 from the Sechahun, Lakkeh Siah, Chadormalu and Esfordi deposits .

155

156 Analyzed plutonic reference samples (Fig. 1) include magnetite from the layered igneous

157 intrusion of Panzhihua in China (n=2), the Bushveld igneous complex in South Africa (n=1), and

158 the layered intrusions of Taberg (n=1), Ulvön (n=1), and Ruoutevare (n=1) in Sweden and an iron-

159 rich gabbro nodule from Iceland (n=1). Samples representative of magmatic magnetites of volcanic

160 derivation were chosen from basalts and dolerite from the Canary Islands (n=3), recent basaltic

161 andesites from Indonesia (n=6), dacites from New Zealand (n=2) and a hypabyssal dolerite from

162 Cyprus (n=1) (Supplementary Table 1). Reference samples for low-temperature or hydrothermal

163 iron ore deposits include magnetite from the polymetallic magnetite-skarn deposit at Dannemora

164 (n=4), the banded iron formation at Striberg (n=1) and the marble-hosted iron oxide deposit at

165 Björnberget (n=1), all situated in Bergslagen, Central Sweden (Supplementary Table 1,

166 Supplementary Note 1).

Troll et al. NCOMMS-18-08734B 6 167

168 Our interpretations are based on the combination of new iron (n=63) and oxygen (n=57) isotope

169 ratios combined with literature data for magnetite from apatite-iron-oxide ores and available

1,5,14,31,32,44–48 170 volcanic, plutonic and the low-temperature or hydrothermal reference materials .

171 Notably, the literature data for low-temperature or hydrothermal magnetite include a sample from

172 the north-American Mineville apatite-iron-oxide deposit, which has been extensively overprinted by

14,49 173 later hydrothermal processes .

174

175 Iron and oxygen isotope results.

56 176 Magnetite from massive apatite-iron-oxide ores from Kiruna have a relatively restricted δ Fe

177 range of +0.12 to +0.41 ‰ (n=11) (Supplementary Table 1, Fig. 3). The Grängesberg and

56 178 Blötberget magnetite samples show δ Fe-values mainly between +0.11 and +0.40 ‰ (n=16).

179 However, one sample from Grängesberg shows an exceptionally high value of +1.0 ‰. Apatite-

56 180 iron-oxide ores from El Laco yield δ Fe-values between +0.24 and +0.36 ‰ (n=6). Magnetite from

181 Bafq have a range in the +0.20 to +0.32 ‰ interval (n=6).

56 182 Magnetite samples from the plutonic and volcanic reference suites show δ Fe-values from +0.11

183 to +0.61 ‰ (n=5) and from +0.06 to +0.46 ‰ (n=13), respectively, consistent with iron isotope

39,46 184 values for magmatic rock suites elsewhere (e.g. Fig. 3) .

185 The low-temperature deposit group, i.e. the Dannemora iron oxide skarn samples and the

56 186 Björnberget and the Striberg samples, on the other side, show relatively low δ Fe-values that range

187 from -0.57 to +0.01 ‰. The magnetite compositions in the low-temperature group form a separate

188 group (Fig. 3) that does not overlap with the reported range of igneous magnetites (+0.06 to +0.49

39,46 189 ‰) , but with low-temperature hydrothermal samples from elsewhere (e.g., Mineville, USA and

14,45,49 190 Xinqiao, China) .

191

Troll et al. NCOMMS-18-08734B 7 18 192 Magnetite separates from the Kiruna Mining District range in δ O value from -1.0 ‰ to

193 +4.1 ‰ (n=14), and those from Grängesberg and Blötberget are between -1.1 to +2.8 ‰ (n=16).

18 194 (Supplementary Table 1, Fig. 4). Magnetite from El Laco shows a large range in δ O values from

195 -4.3 to +4.4 ‰ (n=6), whereas magnetite samples from Bafq give a smaller range of +0.6 to +3.4 ‰

196 (n=6).

197 Magnetites from the plutonic reference samples (Panzhihua, Bushveld, Taberg, Ulvön,

198 Ruoutevare, Iceland gabbro bomb; n=7), and from recent volcanic provinces (New Zealand,

18 199 Indonesia and Tenerife, n=3) show exclusively positive δ O-values between +1.8 and +4.8 ‰ and

200 +3.7 to +3.9 ‰ respectively, which is within or near the commonly accepted range of igneous

18 50 201 magnetites (δ O = +1.0 to +4.0 ‰) . The low-temperature and hydrothermal reference ore

18 202 samples (e.g. Dannemora, Björnberget, Striberg), have low δ O values (-1.2 to -0.4 ‰; n=5) with

18 203 one exception; a skarn sample from Dannemora that has a δ O-value of +2.1 ‰ (Supplementary

204 Table 1). This particular sample, however, comes from a part of the deposit (Konstäng) which itself

205 is reported to be geochemically anomalous with respect to the deposit as a whole (see

206 Supplementary Note 1).

207

208 Comparing the oxygen and iron isotope data of magnetite samples from Kiruna, Grängesberg,

209 El Laco and Bafq, we find that they overlap with the magnetite data from the plutonic and recent

210 volcanic reference samples. Recognized low-temperature or hydrothermal deposits, such as

211 Striberg, Björnberget, and Dannemora record magnetite isotope values that, in turn, differ distinctly

212 in their Fe and O isotope signatures from magmatic values (Figs. 3 & 4). We note that the oxygen

213 isotope data from two vein and disseminated (Ve-Di) magnetite samples from Grängesberg, three

214 samples from Kiruna, as well as magnetite from two samples from El Laco overlap with the low-

215 temperature and hydrothermal reference group. However, these samples still show Fe isotope

216 signatures that are similar to our magmatic reference suite and are hence assumed to reflect

217 originally igneous sources.

Troll et al. NCOMMS-18-08734B 8 218

219 The compositional overlap between Kiruna-type magnetite and the plutonic and volcanic

220 reference suite for Fe and O isotopes is consistent with an ortho-magmatic (magma or highest-

221 temperature magmatic-fluids) origin for the Kiruna-type apatite-iron-oxide samples in this study.

222 Low-temperature processes are reflected in a small number of the Kiruna-type ore samples (n=7)

223 represented by vein- and disseminated-type magnetite samples. The exceptionally high Fe-value for

224 one Grängesberg sample, in turn, is in agreement with the “ultra-magmatic” Fe isotope composition

51 56 225 recorded in magnetite from the Bushveld complex (Fig. 3) . In contrast, the lower δ Fe- and

18 226 δ O-values in Figs. 3 and 4 then either reflect the lower end of the magmatic temperature range, or

227 secondary effects, such as alteration, as well as leaching and subsequent re-precipitation at

228 temperatures below 400 °C which postdates an initial high-temperature (magmatic) stage of

229 formation (Fig. 5). Therefore, the oxygen and iron isotope data for the massive apatite-iron oxide

230 magnetites indicate an originally high-temperature magmatic signature that, most clearly for oxygen

231 isotopes, transitions to lower temperature values indicating a gradual cooling trend. One critical

232 issue, especially for the Palaeoproterozoic Swedish deposits, which have gone through variable

233 grades of metamorphism, is that post-depositional processes (e.g., fluid overprint, re-heating) might

234 have affected the primary isotope composition of the ore and cannot be entirely excluded. Yet, since

235 the same trends in isotope signatures are observed for both the older and younger, less geologically

236 overprinted deposits, we argue that post-depositional changes in isotope composition was negligible

237 in respect to the Fe-O isotope chemistry of our magnetite samples. This is particularly the case for

238 the massive magnetite ores, where this overall chemically inert and refractory mineral would

1 239 provide local buffering with regards to post-depositional re-equilibration .

240

241 Discussion

242 Based on fluid and melt inclusion studies and isotope compositions of mineral pairs, high

243 temperatures of ore formation have been proposed for various apatite-iron oxide ores. For instance,

Troll et al. NCOMMS-18-08734B 9 244 temperature determinations for equilibrium magnetite–quartz pairs and magnetite–pyroxene pairs

1,5 245 from Kiruna, El Laco and Grängesberg yield temperatures that consistently exceed 600 °C . Such

246 crystallization temperatures are further supported by e.g. the occurrence of high-temperature

14,34,52 247 actinolite in e.g. the Los Colorados and Kiruna deposits by Ti exsolution textures in

52 53 248 magnetite from Kiruna , and by Zr in titanite studies at Kiruna that suggest 750-800°C . These

249 temperature determinations provide independent support for a high-temperature (ortho-magmatic)

250 origin of these ore assemblages. Using these temperatures as a reference point and employing

251 appropriate equilibrium fractionation factors, we modeled the isotope compositions of respective

252 equilibrium source to test the mineralization conditions of our sample suite (Supplementary Table

253 2 and Supplementary Fig. 2). We applied available fractionation factors between magnetite and

254 andesite/dacite magma (T ~1000 °C) as well as between an equilibrium aqueous magmatic fluid

255 phase at temperatures between 600 and 800 °C for both iron and oxygen isotopes (see

256 Supplementary Tables 2, 3, 4, 5).

257 For oxygen isotopes, the calculations indicate that magnetite samples from apatite-iron

18 258 oxide ores with δ O ≥ 0, corresponding to over 80 % of our sample set, reflect equilibrium with

18 18 259 either an intermediate magma (δ O of +5.7 to +8.7 ‰) or a high-temperature magmatic fluid (δ O

260 of +5.2 to +9.6 ‰) (Supplementary Tables 2, 3, 4, 5, Supplementary Fig. 2). Although

261 equilibrium with ortho-magmatic, high-temperature sources are found for most of our samples, it

262 naturally remains difficult to distinguish between a magma or a very high-temperature,

263 magmatically-derived aqueous fluid as the initial magnetite source. This realization is highlighted

264 by the fact that our oxygen isotope values from Kiruna-type magnetite samples overlap with oxygen

14 265 isotopes from the Granisle porphyry copper deposit presented in Bilenker et al. , which is

266 proposed to have formed entirely from expelled ortho-magmatic fluids. For the same samples, iron

267 isotope equilibrium source calculations yield values that correspond to magmas and modelled

56 268 magmatic fluids with δ Fe values from +0.08 to +0.38 ‰, and -0.13 to +0.17 ‰, respectively,

269 which plot partly above the reported array of intermediate magmas and magmatic waters

Troll et al. NCOMMS-18-08734B 10 39,54 270 (Supplementary Fig. 2) . This suggests that the metal sources of the samples that exceed the

56 271 range were likely enriched in the heavy iron isotope ( Fe) already at the time of magnetite

39,46 56 272 formation , which we refer to as ultra-magmatic. Such Fe enrichment may be caused by

39 273 magmatic degassing, which is common in many volcanic systems , or more likely represents the

55 274 result of iron oxide-enriched melts acting as an iron sink . Notably, this enrichment is also seen in

275 some samples of our magmatic (plutonic and volcanic) reference suite (Supplementary Table 4,

276 Supplementary Fig. 2), as well as in several Kiruna-type apatite-iron-oxide derived magnetite

14,48 277 samples in other studies (Supplementary Table 5, Supplementary Fig. 2) . Magma degassing

39 56 278 may preferentially remove the lighter Fe isotopes , increasing the δ Fe in the melt. Alternatively,

279 the iron-sink scenario is possibly caused by the tendency for magnetite to incorporate the heavy Fe

280 isotopes over the lighter ones either during crystallization or during silicate-metal immiscibility.

281 This could lead to ultra-magmatic signals following prolonged crystallization of silicate phases

282 from a basaltic magma. Specifically, the heavier iron isotope that partitions into the melt due to

283 removal of early Fe-fractionating minerals, such as olivine and pyroxene, will deplete the melt in

2+ 3+ 46 284 the isotopically lighter Fe and will leave Fe preferentially in the residual magma . When

285 magnetite becomes the dominant iron-bearing phase later in the crystallization sequence it will

46 3+ 286 consequently reflect the isotopically heavy melt signature . Finally, the high Fe /Fetot and the

3+ 287 strong bonding in the tetrahedral site for Fe in magnetite make it a highly suitable host for the

46 288 heavier iron isotope . The combined effects of magma degassing, prolonged fractional

289 crystallization leading to more andesitic to dacitic melts, and the preference of magnetite for the

290 heavy iron isotope are the likely reasons for the ultra-magmatic signature in several magnetite

291 samples from the plutonic-volcanic reference suite as well as from some apatite-iron-oxide ore

292 samples.

293 The key concepts proposed by the magmatic school of thought are formation of Kiruna-type ores

294 by either liquid immiscibility or separation of magnetite cumulates (by sinking or flotation/frothing)

5,17,18,31,38,48 295 from a silicate melt . To evaluate which magmatic process is dominantly responsible for

Troll et al. NCOMMS-18-08734B 11 296 the formation of the massive magnetite bodies has proven difficult and while some petrological and

18,38 297 experimental studies favour the concept of liquid immiscibility , other workers suggest

14,31 298 cumulate-type processes . Unfortunately, the currently available experiments that support liquid

36,38 299 immiscibility are not truly representative of nature (e.g. 40 wt. % P2O5 in a starting melt) . Such

300 a composition contrasts with the host rocks to actual Kiruna-type deposits at Bafq, El Laco,

301 Grängesberg and Kiruna. Using our new data to assess formation mechanisms, we can employ the

35,56 302 isotope fractionation between immiscible Fe-rich melts and their silicate counterparts .

303 Assuming a hydrous system, the maximum fractionation for oxygen isotopes between an iron-rich

35 304 melt and a silicate melt is 0.8 ‰ . Testing if our massive magnetite samples are representative of

305 an Fe-rich melt that formed from immiscibility should then produce associated equilibrium silicate

18 306 melts with δ O between -3.5 and +5.2 ‰. These values are out of the range of common

57,58 307 intermediate igneous rocks . Testing the same approach for the plutonic reference material (n=7)

18 308 produces two equilibrium melts (δ O =+5.4 and +5.6 ‰) that both match a basaltic igneous

309 composition (MORB = +5.7 ±0.4 ‰). The formation of Kiruna-type ores by liquid immiscibility is

310 therefore not fully aligned with our data. Although this would, at first glance, favour cumulate

311 processes over liquid immiscibility, there is probably uncertainty as to the natural fractionation of

18 312 δ O during liquid immiscibility and, to date, little information is available for Fe–isotopes in such

313 situations. This leads us to encourage further tests in order to verify which of these two magmatic

314 processes is more dominant in the formation of apatite-iron oxide systems. Future developments in

315 the field of in-situ analysis of Fe-O isotope composition may hopefully help resolve small features

31 316 such as thin, Ti-poor outer alteration rims as reported by e.g., Knipping et al. on magnetite

317 samples from the Los Colorados apatite-iron-oxide deposit that these authors interpret to have

318 bearing on the ore formation process. However, such textures cannot yet be analyzed for Fe and O

59 319 isotopes in situ , and in respect to our whole-grain results, such volumetrically small features

320 would have a minimal effect on the bulk isotope signature of our samples.

321

Troll et al. NCOMMS-18-08734B 12 322 In contrast to the massive ore samples, the vein and disseminated magnetites associated with

18 323 apatite-iron oxide ores and low-δ O massive magnetite ores from Kiruna, Grängesberg and El Laco

18 324 (i.e. δ Omgt < 0 ‰, n=7) are not in equilibrium with recognized magmatic sources at the previously

325 established temperatures. For these samples equilibrium with magmatic sources would only be

326 obtained at temperatures below 400 °C (Supplementary Table 6). Notably, Fe-P-rich magmas can

34 327 only be liquid down to ~600 °C implying that these magnetite samples cannot have formed

328 directly from a magma. Moreover, the O isotope signatures in these magnetites overlap with those

329 of our low-temperature and hydrothermal reference group and they can either be explained by a

330 cooling magmatic fluid, or by a low-temperature hydrothermal system with external fluid influx.

18 331 For El Laco, disequilibrium between high-temperature magmatic sources and a sub-set of low-δ O

5,32 332 samples has previously been discussed and is attributed to late-stage or secondary processes .

18 333 The low-δ O magnetite samples from El Laco are also associated with considerable amounts of

18 334 hematite that probably formed as a result of oxidation of magnetite by low-δ O, possibly meteoric-

5,32 335 dominated, hydrothermal fluids at temperatures of ≤ 150 °C . Meteoric fluids, particularly from

336 higher altitudes such as the Andes, could cause such a negative shift in oxygen isotopes. However,

337 these fluids would be Fe poor and may thus have only limited effect on iron isotope composition.

338 For the isotope analysis great care was taken to avoid any direct hematite contamination of the

339 samples, yet minor hematite formation along fractures in discrete magnetite grains is seen in some

5,14,17,18,32 340 El Laco samples and may in part explain the larger spread in oxygen isotope data . Our

341 iron isotope data, on the other hand, confirm a magmatic isotope signal throughout, i.e. even in the

18 18 56 342 low-δ O samples. The low-δ O but magmatic δ Fe magnetite compositions at El Laco are thus

343 likely to represent overprint, remobilization and reprecipitation of an originally magmatic iron and

344 oxygen signal by hydrothermal fluids that strongly affected oxygen isotopes in the magnetite, but

14,31 345 had little effect on the iron isotopes . The fluids affecting the ores may either have been derived

346 from the cooling magmatic system or from an external fluid contribution during the evolution of the

347 mineralization. Remarkably, hydrothermal overprint in the Kiruna Mining District in Sweden, for

Troll et al. NCOMMS-18-08734B 13 52 348 instance, has now been suggested to post-date ore formation with up to 250 My . If this is correct,

349 it implies that some non-magmatic (i.e. hydrothermal) isotope signals in apatite-iron-oxide ores may

350 be entirely unrelated to the original mode of formation.

351

352 The Fe-O isotope data obtained on magnetite samples from apatite-iron oxide ores from

353 Sweden, Chile and Iran are thus broadly consistent with the few available Fe-O isotope values from

354 a) the literature (apatite-iron-oxide ores from USA and the Chilean Iron belt), and b) the volcanic

355 and plutonic reference data in this study from various layered igneous intrusions (from China, South

356 Africa and Sweden) and from recent volcanic provinces (Indonesia, Canary Islands, New Zealand,

357 and Iceland). Moreover, the iron and oxygen isotopes of magnetite samples from apatite-iron oxide

358 ores differ, for most samples, from magnetites produced by low temperature or hydrothermal

359 processes (≤ 400°C). The iron and oxygen isotope data together with the calculated equilibrium

360 sources are therefore in agreement with a predominantly (ortho-)magmatic origin (magma and high

361 temperature magmatic fluids) for the investigated Kiruna-type deposits, rather than of a low-

362 temperature hydrothermal one. While our data are very well suited to distinguish these broad

363 formation conditions, they are not ideally suited to resolve the precise formation process and agent,

364 i.e we cannot distinguish magma versus high-temperature fluid or liquid immiscibility processes

365 versus magnetite accumulation (Supplementary Fig. 2). Accepting an essentially magmatic nature

366 of these deposits, a local hydrothermal overprint and replacement within a volcanic to sub-volcanic

367 system would naturally be expected (Fig. 6), and will involve localized late-stage or secondary

1,18,32,41 368 hydrothermal alteration and overprint . Hydrothermal alteration might locally be pronounced

60,61 369 and may seem pervasive in places, as is known from many volcanic provinces . This by-product

370 of otherwise ortho-magmatic formation processes must, however, not be confused with the main

371 source signal of most Kiruna-type magnetite samples revealed in our study (Figs. 5 & 6).

372 Hydrothermally-formed magnetite is isotopically distinct and appears subordinate in our

373 sample suite. The bulk of the Kiruna-type ore investigated thus formed in sub-volcanic

Troll et al. NCOMMS-18-08734B 14 374 environments under essentially high-temperature magmatic conditions, in line with the magmatic

3,5,14,17,18,31,62 375 school of thought . Precipitation of magnetite was either from iron-oxide-saturated

376 intermediate magmas or from immiscible Fe-rich melts that separated from broadly andesitic to

377 dacitic parent magmas and subsequent physical (either gravity or gas-driven) magnetite segregation

18,24,31,35,63 378 to form massive magnetite melts and mushes . Kiruna-type apatite-iron-oxide ores are

379 hence dominantly a magmatic phenomenon and they presumably continue to form in active arc- and

380 back-arc type sub-volcanic environments up to the present day. Our combined isotope data and

381 calculations represent a significant advance in the understanding of Kiruna-type ore deposits and

382 over-rules most arguments for a completely hydrothermal mode of formation. Moreover, we

383 provide a reference system for Fe-O isotopes in Kiruna-type ores against which future research can

384 test genetic concepts for low- versus high-temperature origin of as yet underexplored Kiruna-type

385 deposits.

386

387 Acknowledgements

388 We thank Prof. N. Arndt (Université Joseph Fourier, Grenoble), Prof. J. Gamble (Victoria

389 University, Wellington), Dr. J. O. Nyström (Swedish Museum of Natural History, Stockholm),

390 Gunnar Rauséus (Dannemora Mineral AB), and Prof. C. J. Stillman (Trinity College Dublin) for

391 donating samples for this study. We further thank Dr. J. Omrani, Dr. A. Houshmandzadeh, Dr. M.

392 Kargarrazi and the Geological Survey of Iran for their help and support during field work and data

393 interpretation, as well as the staff at the Geological Survey of Sweden (SGU) Mineral Office in

394 Malå, for helpful assistance during drill core sampling. We thank Joel Baker for his help during

395 isotope analyses at the Victoria University in Wellington and Harri Geiger and Weian Sun for help

396 during the microprobe sessions. Generous funding from the SGU, the Swedish Research Council

397 (VR) and Uppsala University (UU) is gratefully acknowledged. This is Vegacenter contribution

398 #012.

399

Troll et al. NCOMMS-18-08734B 15 400 Author contributions

401 VRT, EJ and KH conceived the study. Field work was carried out by EJ, KH, KPN, SAM,

402 UBA and VRT. Sample preparation was carried out by FW and SS. Isotope analyses were

403 performed by CH, EK, FW and MAM. Modelling was performed by FW, VRT and CH and

404 illustrations were prepared by FW and VRT. The manuscript was written by VRT and FW with

405 contributions from all co-authors.

406

407 Additional information

408 Supplementary Information accompanies this paper.

409

410 Competing interests: The authors declare no competing interests.

Troll et al. NCOMMS-18-08734B 16 411 Methods

412 Sampling. Samples from the Kiruna Mining District (n=11) come from the type locality for Kiruna-

413 type apatite-iron-oxide ores at Kiirunavaara, Sweden, as well as from other apatite-iron-oxide

414 deposits in the district, including magnetite dykes in the footwall of the smaller deposit at

415 Luossavaara. as well as massive ore from the Mertainen and Rektorn deposits. Mineralised samples

416 from the Grängesberg Mining District (GMD) were collected from three drillcores (DC), DC 690

417 (n=7), DC 717 (n=3) and DC 575 (n=3) that transect the deposit with a shallow plunge (< 20 °) and

418 were drilled at 650 m (n=2) and 570 m below the surface respectively. In addition to massive

419 apatite-iron-oxide samples, two Grängesberg samples were selected from magnetite veins and

420 disseminations in the host rocks. One sample was collected from the smaller Blötberget apatite-

421 iron-oxide deposit within the greater Grängesberg area and another from the nearby but contrasting

422 marble-hosted hydrothermal/low-temperature iron oxide deposit at Björnberget (n=1). Samples

423 from El Laco (n=6) were sampled at the surface and come from Laco Sur, the southern deposit in

424 the area. To obtain a meaningful and widely applicable comparison of the apatite-iron-oxide ores

425 with hydrothermal and magmatic reference samples, oxygen and iron isotope values were also

426 determined on magnetite from massive ore from the iron oxide-polymetallic skarn deposit at

427 Dannemora in Sweden (n=4), the banded iron formation at Striberg in Sweden (n=1), the layered

428 igneous intrusion of Panzhihua in China (n=2), the Bushveld igneous complex in South Africa

429 (n=1), the Swedish layered igneous intrusions of Taberg (n=1), Ulvön (n=1) and Ruoutevare (n=1),

430 and from a gabbro bomb from Skjaldbreiður in Iceland (n=1). Samples representative of recent

431 igneous magnetites were chosen from basaltic andesites from Indonesia (n=6), basalts and dolerite

432 from the Canary Islands (n=3), dacites from New Zealand (n=2) and a dolerite from Troodos massif

433 in Cyprus (n=1). An overview of the samples used in this study and details on their mineral

434 assemblage and provenance is given in Supplementary Table 1. An outline of the geological

435 setting of our sample suite is given in the Supplementary Information

Troll et al. NCOMMS-18-08734B 17 436 Iron isotope analysis. Analysis of the individual magnetite samples for Fe isotopes was dominantly

437 carried out at the Victoria University in Wellington, New Zealand (n=47). The crystals were

438 digested and chemically purified with concentrated HF and HNO3 acid and the analysis was then

57 58 439 done using a Fe– Fe double spike and a Nu Plasma MC-ICP-MS (Multicollector-Inductively

440 Coupled Plasma Mass Spectrometer). As a standard the international IRMM-014 CRM material

64 441 was used. Full details on the Fe isotope analysis are given in Millet et al. . All iron isotope data

56 56 54 442 were recorded as δ Fe, which is the deviation of Fe/ Fe relative to the IRMM-014 CRM standard

443 material. The average 2σ error during iron isotope analysis was 0.03 ‰.

444 A set of magnetite samples (n=11) was analysed for iron-isotopes by ALS Scandinavia Ltd.

445 in Luleå, Sweden. The magnetite samples were prepared for analysis by microwave-assisted

65 446 digestion in a HNO3+HCl+HF mixture according to the method described in Ingri et al. . The

447 isotope analysis was then carried out with a Thermo Scientific Neptune MC-ICP mass spectrometer.

56 448 The δ Fe-values were calculated with relation to the IRMM-014 CRM standard. The 2σ error was

449 calculated from two independent consecutive measurements and was on average also about 0.03‰.

450 Six further magnetite samples were analyzed at the Vegacenter at the Swedish Museum of Natural

451 History in Stockholm. The crystals were digested and chemically purified using concentrated HF

66 64 452 and HNO3 and 10M HCl acid following the procedures of Borrok et al. and Millet et al. . The

453 samples were diluted with 0.3 M HNO3 to a concentration of 2-3 ppm before measurement. The Fe

454 isotope analyses were performed on a Nu Plasma II HR-MC-ICP-MS in pseudo-high-resolution

455 mode to resolve interfering species. The samples were corrected for mass bias using the standard-

456 sample bracketing technique, normalizing to the IRMM-014 standard. The average 2σ external

56 457 reproducibility for the samples was 0.06 ‰ for δ Fe.

458

459 Oxygen isotope analysis. The analysis for oxygen isotopes was carried out at the University of

460 Cape Town (South Africa) using a Finnigan DeltaXP dual inlet gas source mass spectrometer

67 461 (n=59). For the oxygen analysis the magnetite samples were prepared by laser fluorination ,

Troll et al. NCOMMS-18-08734B 18 462 whereby they were reacted with 10 kPa of BrF5, and the purified O2 was collected onto a 5 Å

463 molecular sieve in a glass storage bottle. As a reference and calibration standard Monastery garnet

68 18 464 was used . All oxygen data were recorded in the usual δ O notation relative to SMOW where

18 18 16 465 δ O = (Rsample/Rstandard -1)*1000, and R = the measured ratio O/ O. All oxygen isotope data was

466 obtained with a 2σ error of ≤0.2 ‰.

467

468 Data availability: The authors declare that all relevant data are available within the article and its

469 supplementary information files.

470

471

Troll et al. NCOMMS-18-08734B 19 472 References

473 1. Jonsson, E. et al. Magmatic origin of giant ‘Kiruna-type’ apatite-iron-oxide ores in Central

474 Sweden. Sci. Rep. 3, (2013).

475 2. Williams, P. & Barton, M. Iron oxide copper-gold deposits: Geology, space-time

476 distribution, and possible modes of origin. Econ. Geol. 100th Anniversary Volume, 371-

477 405 (2005). doi:10.5382/AV100.13

478 3. Geijer, P. Igneous rocks and iron ores of Kiirunavaara, Luossavaara and Tuolluvaaratle.

479 Geologie des Kirunagebietes 2 (1910).

480 4. Hitzman, M. W., Oreskes, N. & Einaudi, M. T. Geological Characteristics and Tectonic

481 Setting of Proterozoic Iron-Oxide (Cu-U-Au-Ree) Deposits. Precambrian Res. 58, 241–287

482 (1992).

483 5. Nyström, J. O., Billström, K., Henríquez, F., Fallick, A. E. & Naslund, H. R. Oxygen isotope

484 composition of magnetite in iron ores of the Kiruna type in Chile and Sweden. GFF 130,

485 177–188 (2008).

486 6. Weihed, P. et al. 8: Precambrian geodynamics and ore formation: The Fennoscandian Shield.

487 Ore Geol. Rev. 27, 273–322 (2005).

488 7. Oyarzún, J. Frutos, J. Tectonic and petrological frame of the Cretaceous iron deposits of

489 north Chile. Min. Geol. 34, 21–31 (1984).

490 8. Lyons, J. I. Volcanogenic iron oxide deposits, Cerro de Mercado and vicinity, Durango,

491 Mexico. Econ. Geol. 83, 1886–1906 (1988).

492 9. Naslund, H. R., Henríquez, F., Nyström, J. O., Vivallo, W. & Dobbs, F. M. Magmatic iron

493 ores and associated mineralisation: examples from the chilean high andes and coastal field

494 Relationships at El Laco. Porter, TM, Hydrothermal Iron Oxide Copper-Gold Relat. Depos.

495 A Glob. Perspect. 2, 207-226 (2002).

496 10. Helvaci, C. Apatite-rich iron deposits of the Avnik (Bingol) region, southeastern Turkey.

497 Economic Geology 79, 354–371 (1984).

Troll et al. NCOMMS-18-08734B 20 498 11. Mücke, A. & Younessi, R. Magnetite-apatite deposits (Kiruna-type) along the Sanandaj-

499 Sirjan zone and in the Bafq area, Iran, associated with ultramafic and calcalkaline rocks and

500 carbonatites. Mineral. Petrol. 50, 219–244 (1994).

501 12. Zhang, Z. et al. Spatio-temporal distribution and tectonic settings of the major iron deposits

502 in China: An overview. Ore Geol. Rev. 57, 247-263 (2014).

503 doi:10.1016/j.oregeorev.2013.08.021

504 13. Seo, J., Choi, S. G., Kim, D. W., Park, J. W. & Oh, C. W. A new genetic model for the

505 Triassic Yangyang iron-oxide-apatite deposit, South Korea: Constraints from in situ U-Pb

506 and trace element analyses of accessory minerals. Ore Geol. Rev. 70, 110–135 (2015).

507 14. Bilenker, L. D. et al. Fe-O stable isotope pairs elucidate a high-temperature origin of Chilean

508 iron oxide-apatite deposits. Geochim. Cosmochim. Acta 177, 94–104 (2016).

509 15. Hofstra, A. H. et al. Mineral thermometry and fluid inclusion studies of the pea ridge iron

510 oxide-apatite-rare earth element deposit, Mesoproterozoic St. Francois Mountains Terrane,

511 Southeast Missouri, USA. Econ. Geol. 111, 1985-2016 (2016).

512 doi:10.2113/econgeo.111.8.1985

513 16. Montreuil, J. F., Corriveau, L., Potter, E. G. & De Toni, A. F. On the relationship between

514 alteration facies and metal endowment of iron oxide-alkali-altered systems, Southern Great

515 Bear Magmatic Zone (Canada). Econ. Geol. 111, 2139-2168 (2016).

516 doi:10.2113/econgeo.111.8.2139

517 17. Tornos, F., Velasco, F. & Hanchar, J. M. Iron-rich melts, magmatic magnetite, and

518 superheated hydrothermal systems: The El Laco deposit, Chile. Geology 44, 427-430 (2016).

519 doi:10.1130/G37705.1

520 18. Tornos, F., Velasco, F. & Hanchar, J. M. The Magmatic to Magmatic-Hydrothermal

521 Evolution of the El Laco Deposit (Chile) and Its Implications for the Genesis of Magnetite-

522 Apatite Deposits. Soc. Econ. Geol. 112, 1595–1628 (2017).

523 19. Westhues, A., Hanchar, J. M., Whitehouse, M. J. & Martinsson, O. New constraints on the

Troll et al. NCOMMS-18-08734B 21 524 timing of host-rock emplacement, hydrothermal alteration, and iron oxide-apatite

525 mineralization in the Kiruna District, Norrbotten, Sweden. Econ. Geol. 111, 1595-1618

526 (2016). doi:10.2113/econgeo.111.7.1595

527 20. Westhues, A. et al. Tracing the fluid evolution of the Kiruna iron oxide apatite deposits using

528 zircon, monazite, and whole rock trace elements and isotopic studies. Chem. Geol. 466, 303-

529 322 (2017). doi:10.1016/j.chemgeo.2017.06.020

530 21. Westhues, A., Hanchar, J. M., LeMessurier, M. J. & Whitehouse, M. J. Evidence for

531 hydrothermal alteration and source regions for the Kiruna iron oxide-apatite ore (northern

532 Sweden) from zircon Hf and O isotopes. Geology 45, 571-574 (2017). doi:10.1130/G38894.1

533 22. Broughm, S. G., Hanchar, J. M., Tornos, F., Westhues, A. & Attersley, S. Mineral chemistry

534 of magnetite from magnetite-apatite mineralization and their host rocks: examples from

535 Kiruna, Sweden, and El Laco, Chile. Miner. Depos. 52, 1223–1244 (2017).

536 23. He, X. F., Santosh, M., Tsunogae, T. & Malaviarachchi, S. P. K. Magnetite-apatite deposit

537 from Sri Lanka: Implications on Kiruna-type mineralization associated with ultramafic

538 intrusion and mantle metasomatism. Am. Mineral. 103, 26-38 (2018). doi:10.2138/am-2018-

539 6254

540 24. Frietsch, R. & Geology, E. On the Magmatie Origin of Iron Ores o , f the Kiruna Type. Econ.

541 Geol. 73, 478–485 (1978).

542 25. Parak, T. Kiruna iron ores are notintrusive-magmatic ores of the kiruna type Economic

543 Geology 70, 1242–1258 (1975).

544 26. Nystrom, J. O. & Henriquez, F. Magmatic features of iron ores of the Kiruna type in Chile

545 and Sweden: ore textures and magnetite geochemistry. Econ. Geol. (1994).

546 doi:10.2113/gsecongeo.90.2.473

547 27. Sillitoe, R. H. & Burrows, D. R. New field evidence bearing on the origin of the El Laco

548 magnetite deposit, Northern Chile. Econ. Geol. 97, 1101–1109 (2002).

549 28. Smith, M. P., Storey, C. D., Jeffries, T. E. & Ryan, C. In situ U-Pb and trace element analysis

Troll et al. NCOMMS-18-08734B 22 550 of accessory minerals in the Kiruna District, Norrbotten, Sweden: New constraints on the

551 timing and origin of mineralization. J. Petrol. (2009). doi:10.1093/petrology/egp069

552 29. Smith, M. P., Gleeson, S. A. & Yardley, B. W. D. Hydrothermal fluid evolution and metal

553 transport in the Kiruna District, Sweden: Contrasting metal behaviour in aqueous and

554 aqueous-carbonic brines. Geochim. Cosmochim. Acta (2013). doi:10.1016/j.gca.2012.10.015

555 30. Dare, S. A. S., Barnes, S. J. & Beaudoin, G. Did the massive magnetite “lava flows” of El

556 Laco (Chile) form by magmatic or hydrothermal processes? New constraints from magnetite

557 composition by LA-ICP-MS. Miner. Depos. 50, 607–617 (2014).

558 31. Knipping, J. L. et al. Giant Kiruna-type deposits form by efficient flotation of magmatic

559 magnetite suspensions. Geology 43, 591–594 (2015).

560 32. Rhodes, a L. & Oreskes, N. Oxygen isotope composition of magnetite deposits at El Laco,

561 Chile: Evidence of formation from isotopically heavy fluids. Econ. Geol. Spec. Publ. No. 7

562 Geol. Ore Depos. Cent. Andes 333–351 (1999).

563 33. Naslund, H. R. The effect of oxygen fugacity on liquid immiscibility in iron- bearing silicate

564 melts. Am. J. Sci. 283, 1034-1059 (1983). doi:10.2475/ajs.283.10.1034

565 34. Lledo, H. I. & Jenkins, D. M. Experimental investigation of the upper thermal stability of

566 Mg-rich actinolite; implications for kiruna-type iron deposits. J. Petrol. 49, 225-238 (2008).

567 doi:10.1093/petrology/egm078

568 35. Lester, G. W., Clark, A. H., Kyser, T. K. & Naslund, H. R. Experiments on liquid

569 immiscibility in silicate melts with H2O, P, S, F and Cl: Implications for natural magmas.

570 Contrib. to Mineral. Petrol. 166, 329-349 (2013). doi:10.1007/s00410-013-0878-1

571 36. Mungall, J. E. Immiscible silicate and Fe-P oxide melts preserved in unconsolidated tephra at

572 El Laco volcano, Chile. 46, 255–258 (2018).

573 37. Hou, T. et al. Experimental study of liquid immiscibility in the Kiruna-type Vergenoeg iron–

574 fluorine deposit, South Africa. Geochim. Cosmochim. Acta 203, 303–322 (2017).

575 38. Hou, T. et al. Immiscible hydrous Fe-Ca-P melt and the origin of iron oxide-apatite ore

Troll et al. NCOMMS-18-08734B 23 576 deposits. Nat. Commun. 9, 1–8 (2018).

577 39. Heimann, A., Beard, B. L. & Johnson, C. M. The role of volatile exsolution and sub-solidus

578 fluid/rock interactions in producing high 56Fe/54Fe ratios in siliceous igneous rocks.

579 Geochim. Cosmochim. Acta 72, 4379–4396 (2008).

580 40. Allen, R. L., Lundström, I., Ripa, M., Simeonov, A. & Christofferson, H. Facies analysis of a

581 1.9 Ga, continental margin, back-arc, felsic caldera province with diverse Zn-Pb-Ag-(Cu-Au)

582 sulfide and Fe oxide deposits, Bergslagen Region, Sweden. Econ. Geol. 91, 979–1006

583 (1996).

584 41. Jonsson, E., Harlov, D. E., Majka, J., Högdahl, K. & Persson-Nilsson, K. Fluorapatite-

585 monazite-allanite relations in the Grängesberg apatite-iron oxide ore district, Bergslagen,

586 Sweden. Am. Mineral. 101, 1769-1782 (2016). doi:10.2138/am-2016-5655

587 42. Mokhtari, M. A. A., Zadeh, G. H. & Emami, M. H. Genesis of iron-apatite ores in Posht-e-

588 Badam Block (Central Iran) using REE geochemistry. J. Earth Syst. Sci. 122, 795–807

589 (2013).

590 43. Ramezani, J. & Tucker, R. D. The Saghand Region, Central Iran: U-Pb geochronology,

591 petrogenesis and implications for Gondwana tectonics. Am. J. Sci. 303, 622–665 (2003).

592 44. Barnes, J. D., Paulick, H., Sharp, Z. D., Bach, W. & Beaudoin, G. Stable isotope (δ18O, δD,

593 δ37Cl) evidence for multiple fluid histories in mid-Atlantic abyssal peridotites (ODP Leg

594 209). Lithos 110,83-94 (2009). doi:10.1016/j.lithos.2008.12.004

595 45. Wang, Y., Zhu, X. kun, Mao, J. wen, Li, Z. hong & Cheng, Y. bo. Iron isotope fractionation

596 during skarn-type metallogeny: A case study of Xinqiao Cu-S-Fe-Au deposit in the Middle-

597 Lower Yangtze valley. Ore Geol. Rev. 43, 194–202 (2011).

598 46. Sossi, P. A., Foden, J. D. & Halverson, G. P. Redox-controlled iron isotope fractionation

599 during magmatic differentiation: An example from the Red Hill intrusion, S. Tasmania.

600 Contrib. to Mineral. Petrol. 164, 757–772 (2012).

601 47. Dziony, W. et al. In-situ Fe isotope ratio determination in Fe-Ti oxides and sulfides from

Troll et al. NCOMMS-18-08734B 24 602 drilled gabbros and basalt from the IODP Hole 1256D in the eastern equatorial Pacific.

603 Chem. Geol. 363, 101–113 (2014).

604 48. Childress, T. M., Simon, A. C., Day, W. C., Lundstrom, C. C. & Bindeman, I. N. Iron and

605 oxygen isotope signatures of the pea ridge and pilot knob magnetite-apatite deposits,

606 Southeast Missouri, USA. Econ. Geol. 111, 2033–2044 (2016).

607 49. Valley, P. M., Hanchar, J. M. & Whitehouse, M. J. New insights on the evolution of the

608 Lyon Mountain Granite and associated Kiruna-type magnetite-apatite deposits, Adirondack

609 Mountains, New York State. Geosphere 7, 357-389 (2011). doi:10.1130/GES00624.1

610 50. H. P. Taylor. Oxygen isotope studies of hydrothermal mineral deposits. in Geochemistry of

611 Hydrothermal Ore Deposits (ed. Barnes, H. L.) 109–142 (Holt, Rinehart and Winston Inc.,

612 1967).

613 51. Bilenker, L. D., VanTongeren, J. A., Lundstrom, C. C. & Simon, A. C. Iron isotopic

614 evolution during fractional crystallization of the uppermost Bushveld Complex layered mafic

615 intrusion. Geochem. Geophys. 18, 956-972 (2017). doi:10.1002/2016GC006660

616 52. Aupers K. Gangue mineralogy and deportment of deleterious elements in the iron ore of the

617 Kiirunavaara deposit, Sweden, M.Sc. thesis, TU Bergakademie Freiberg, 167 p (2014).

618 53. Gray, W. Genesis of the Kiirunavaara nodular porphyry and Kiruna-type iron deposits: clues

619 from titanite geothermometry and magnetite trace element compositions. Bachelor thesis,

620 University of St. Andrews, 54 p (2016).

621 54. Dauphas, N. et al. Magma redox and structural controls on iron isotope variations in Earth’s

622 mantle and crust. Earth Planet. Sci. Lett. 398, 127–140 (2014).

623 55. Telus, M. et al. Iron, zinc, magnesium and uranium isotopic fractionation during continental

624 crust differentiation: The tale from migmatites, granitoids, and pegmatites. Geochim.

625 Cosmochim. Acta 97, 247–265 (2012).

626 56. Kyser, T. K., Lesher, C. E. & Walker, D. The effects of liquid immiscibility and thermal

627 diffusion on oxygen isotopes in silicate liquids. Contrib. to Mineral. Petrol. 133, 373-381

Troll et al. NCOMMS-18-08734B 25 628 (1998). doi:10.1007/s004100050459

629 57. Jr, H. P. T. The oxygen isotope geochemistry of igneous rocks. Contrib. to Mineral. Petrol.

630 19, 1–71 (1968).

631 58. Bindeman, I. Oxygen Isotopes in Mantle and Crustal Magmas as Revealed by Single Crystal

632 Analysis. Rev. Mineral. Geochemistry 69, 445–478 (2008).

633 59. Huberty, J. M. et al. Crystal orientation effects in δ18O for magnetite and hematite by SIMS.

634 Chem. Geol. 276, 269–283 (2010).

635 60. Donoghue, E. et al. Low-temperature hydrothermal alteration of intra-caldera tuffs, Miocene

636 Tejeda caldera, Gran Canaria, Canary Islands. J. Volcanol. Geotherm. Res. 176, 551-564

637 (2008). doi:10.1016/j.jvolgeores.2008.05.002

638 61. Donoghue, E., Troll, V. R. & Harris, C. Fluid-rock interaction in the miocene, post-caldera,

639 Tejeda Intrusive Complex, Gran Canaria (Canary Islands): Insights from mineralogy, and O-

640 and H-isotope geochemistry. J. Petrol. 51, 2149–2176 (2010).

641 62. Henriquez, F. & Nystrom, J. O. Magnetite bombs at El Laco volcano, Chile. GFF 120, 269–

642 271 (1998).

643 63. O’Driscoll, B., Stevenson, C. T. E. & Troll, V. R. Mineral lamination development in layered

644 gabbros of the british palaeogene igneous province: A combined anisotropy of magnetic

645 susceptibility, quantitative textural and mineral chemistry study. J. Petrol. 49, 1187–1221

646 (2008).

647 64. Millet, M. A., Baker, J. A. & Payne, C. E. Ultra-precise stable Fe isotope measurements by

648 high resolution multiple-collector inductively coupled plasma mass spectrometry with a Fe-

649 57-Fe-58 double spike. Chem. Geol. 304, 18–25 (2012).

650 65. Ingri, J. et al. Iron isotope fractionation in river colloidal matter. Earth Planet. Sci. Lett. 245,

651 792–798 (2006).

652 66. Borrok, D. et al. Separation of copper, iron, and zinc from complex aqueous solutions for

653 isotopic measurement. Chem. Geol. 242, 400–414 (2007).

Troll et al. NCOMMS-18-08734B 26 654 67. Harris, C. & Vogeli, J. Oxygen isotope composition of garnet in the Peninsula Granite, Cape

655 Granite Suite, South Africa: Constraints on melting and emplacement mechanisms. South

656 African J. Geol. (2010). doi:10.2113/gssajg.113.4.401

657 68. Harris, C., Smith, S. H. & le Roex, A. P. Oxygen isotope composition of phenocrysts from

658 Tristan da Cunha and Gough Island lavas: variation with fractional crystallization and

659 evidence for assimilation. Contrib. to Mineral. Petrol. 138, 164–175 (2000).

660 69. Anbar, A. D. Iron stable isotopes: Beyond biosignatures. Earth Planet. Sci. Lett. 217, 223–

661 236 (2004).

662 70. Severmann, S. & Anbar, A. D. Reconstructing paleoredox conditions through a multitracer

663 approach: The key to the past is the present. Elements 5, 359–364 (2009).

Troll et al. NCOMMS-18-08734B 27 664 Figures

665 Fig.1

666

667 Figure 1. Sample overview map. a) Global map showing the different locations of origin for

668 apatite-iron-oxide ore and reference samples. b) A close up view of the main part of the

669 Fennoscandian Shield showing the sample locations for magnetites from Sweden.

670

671

Troll et al. NCOMMS-18-08734B 28 672 Fig.2

673

674 Figure 2. Images of apatite-iron oxide ores in this study. False-colour BSE images of massive

675 magnetite ore samples from a) and b) Kiruna, c) and d) Grängesberg e) and f) Bafq and d) and h) El

676 Laco. Kiruna, Grängesberg and Bafq magnetite samples are homogeneous and commonly lack

677 zonation or signs of alteration. El Laco (g and h), is exceptional in this respect as for some samples

678 intra-crystal zonation is observed. As a supplement, regular greyscale BSE images for these

679 samples are provided in Supplementary Fig. 1.

680

681

Troll et al. NCOMMS-18-08734B 29 682 Fig.3

683

684 Figure 3. Iron isotope results from this study. Shown is the distribution of iron isotopes in

685 magnetites from the Kiruna and Grängesberg districts, El Laco, and the Bafq district together with

14,45,47,48 686 available literature data . Reference fields for common hydrothermal and magmatic

39,45–47,51,69,70 687 magnetites are shown for comparison . Magnetites from apatite-iron oxide ores show a

688 clear distinction from low-temperature or hydrothermal magnetites and overlap with the layered

689 intrusions and volcanic reference magnetites (i.e. in the magmatic reference field). Data from Wang

45 690 et al. show the effects of a progressive transgression from ortho-magmatic processes to

56 691 hydrothermal fluid evolution from originally higher to lower δ Fe values and an originally

692 magmatic fluid may thus evolve into a hydrothermal fluid. One hydrothermal sample from the

56 18 693 highly altered, remobilized and recrystallized Mineville deposit in the USA (δ Fe = -0.92‰, δ O

Troll et al. NCOMMS-18-08734B 30 14 694 = -0.79) is not shown for simplification. Ve-Di samples represent vein and disseminated

695 magnetites.

696

697 Fig.4

698

699 Figure 4. Oxygen isotope results from this study. Oxygen isotopes of magnetite samples from

14 48 700 Kiruna, Grängesberg, El Laco, the Chilean Iron Belt , the Pea Ridge and Pilot Knob deposits ,

701 and the Bafq district, are compared to reference samples from layered igneous intrusions, recent

702 volcanic magnetites, and low-temperature or hydrothermal ore deposits. The range of typical

50 703 igneous magnetite is outlined in the reference box . The majority of magnetite samples from

18 704 apatite-iron oxide ores plot within the reference field for common magmatic δ O-values and

705 overlap with magnetite values from recent volcanic rocks and layered intrusions. The low-

706 temperature or hydrothermal reference suite, together with low-temperature magnetite literature

Troll et al. NCOMMS-18-08734B 31 14,44 707 data , plot dominantly to the left of the magmatic magnetite field, with only one exception, a

708 magnetite from the Fe-skarn deposit at Dannemora. This particular outlier comes from a part of the

709 deposit (Konstäng) which itself represents a geochemical anomaly within the Dannemora deposit.

5,14,17,18,32 710 Our values for El Laco overlap with results from previous studies .

711

712 Fig.5

713

714 Figure 5. Distribution of Fe and O isotope values of magnetite samples used in this study. The

715 various magnetite samples can be divided into three groups according to their Fe-O isotope

716 composition; i) high-temperature magmatic magnetites, ii) hydrothermal magnetite samples and iii)

Troll et al. NCOMMS-18-08734B 32 717 low-temperature magnetite samples. Most of the magnetite compositions of the apatite-iron-oxide

718 ores in this study lie within, or near, the reference field for igneous magnetite, and overlap with the

719 plutonic and volcanic magnetite samples analysed as reference suite. See also Supplementary Note

720 1 and Supplementary Fig. 2 for a detailed assessment of temperature-dependent equilibrium

721 compositions. Reference field for common igneous and hydrothermal magnetites are based on

14,39,44,46,47,50,51 722 literature data .

723

724 Fig.6

725

726 Figure 6. Schematic representation of magmatic stages for Kiruna-type apatite-iron-oxide

727 ores from this and other studies, and from the analyzed reference materials. Stages II and III

728 comprise ortho-magmatic ore formation: with decreasing temperature and on-going crystallization

729 in the melt, the volatile/fluid pressure will increase and magmatic fluids are being expelled into the

730 surrounding rocks. Below ~600 °C (towards the end of stage III), the magmatic-derived volatile

731 pressure may begin to decrease, allowing progressively more of available external fluids into the

732 system that initiate hydrothermal activity (< 400 °C). Massive apatite-iron oxide ores appear to

733 commence crystallization in the ortho-magmatic stages (Stages II and III), whereas vein and

734 disseminated magnetites formed mainly during Stage IV (hydrothermal precipitation and

Troll et al. NCOMMS-18-08734B 33 735 replacement). This implies that the commonly observed hydrothermal signals in apatite-iron oxide

736 ores are late-stage products that are results of syn- to post-magmatic hydrothermal processes active

737 during the cooling of the volcanic system, or in some cases possibly reprecipitation during later

738 overprints.

Troll et al. NCOMMS-18-08734B 34 739 Supplementary Information

740

741 Supplementary Note 1: Sampling sites

742 1. Sampling locations for apatite-iron-oxide ores

743 1.1 Kiruna district, northernmost Sweden: The Kiirunavaara, Luossavaara and Mertainen

744 deposits in the Kiruna Mining District in Lappland, northern Sweden are hosted by trachyandesites

3,19,71–74 745 and rhyodacitic ignimbrites and tuffs with an age of 1.89 to 1.87 Ga . These volcanic rocks

75 746 are underlain by older greenstones which are supposed to have formed in an extensional setting .

747 Deformation in the area is generally non-penetrative, dominated by local shearing and brittle

748 tectonics and while regional metamorphism in greenschist facies have been invoked an even lower

76,77 749 overprint have been suggested at Kiirunavaara and Luossavaara . The apatite-iron oxide ores in

750 the Kiruna district are dominated by magnetite, with iron contents of 50-70 % and up to c. 20 %

25 751 apatite . The deposits of the Kiruna district hold pre-mining reserves of more than 2 billion tons of

5 752 ore . The ore bodies have been interpreted to be just like the El Laco deposit of primarily magmatic

753 origin on the basis of geochemical and textural observations including nodular ore textures and

5,24,26,78–81 754 oxygen isotopes .

755

756 1.2 Grängesberg/Blötberget, Bergslagen, Central Sweden: The apatite-iron-oxide deposit at

757 Grängesberg and the smaller Blötberget deposit are dominated by magnetite with subordinate

758 hematite and additionally both oxides occur as associated veins and disseminations in the

1,82 759 immediate host rocks . In the massive ores, bands of fine-grained fluorapatite with associated

41 760 REE phosphates as well as variable amounts of silicates are characteristic . The main deposit at

761 Grängesberg, the so called “Export Field” consists of iron oxide ores in the ratio of approximately

762 80 % magnetite and 20 % hematite. Hematite-dominated parts occur mostly in the structural

763 footwall and in the vicinity of crosscutting pegmatite dykes. Alteration zones in the host rocks right

764 next to the mineralisation comprise disseminated and discrete phyllosilicate (biotite, chlorite) and

Troll et al. NCOMMS-18-08734B 35 765 amphibole-rich assemblages (so-called sköl) with variable amounts of iron oxides and fluorapatite

83 766 . The mineralisation is stratiform and dips between 50° and 70° towards the south-east and could

84,85 767 be followed for more than 900 m at the surface where its width ranged between 50 and 100 m .

768 The ore is hosted by metavolcanic rocks of andesitic to dacitic compositions belonging to the c.

40,86 769 1.91-1.87 Ga volcano-sedimentary succession of Bergslagen .

770

771 1.3 El Laco, Pico Laco, Northern Chile: The apatite-iron oxide ores of El Laco are situated in

5 772 Northern Chile at the flank of the Pliocence Pico Laco volcanic complex . The area around El Laco

2 773 hosts seven different deposits distributed over 30 km with a total amount of 500 Mt of high grade

5 774 (~60% Fe) ore . The ore consists mainly of magnetite, however, hematite is also present as an

775 oxidation product. Fission track dating of apatite crystals within the El Laco ore gave an age

87 776 estimation of 2.1± 0.1 Ma . The host rocks of the deposit consist of typical subduction zone

777 andesites and dacites which have been hydrothermally altered with alteration increasing at depth

5,88,89 5,88 778 . Although hydrothermal activity was associated with the ore formation , the ore at El Laco

779 is supposed to have formed by dominantly magmatic processes, involving an iron oxide-rich

62 780 magma and magmatic fluids . Textural analysis indicates that the ore resembles intrusive and

781 extrusive magmatic activity, such as lava flows, pyroclastics and dykes, where the lava flow

5,88 782 deposits are notably dominated by hematite . A magmatic origin is also supported by features of

783 iron oxide (magnetite) lava bombs, aa and pahoehoe lavas as well as vesicle-like cavities in

5,62 784 addition to geochemical data from oxygen isotope analysis . However, a hydrothermal origin is

32 785 put forward on the basis of oxygen isotopes arguing that the surprising isotopic homogeneity of

786 magnetite samples at El Laco (~ +4.0 ‰) could not result from magmatic processes as a wider

787 range of values would be expected due to magma cooling and hydrothermal processes associated

788 with volcanic activity. Instead, these authors propose an origin by hydrothermal replacement with

18 27,30,32 789 some high-δ O hydrothermal fluid as transport agent possibly also in some sort of evaporitic

Troll et al. NCOMMS-18-08734B 36 790 pan. The samples for this study represent massive magnetite ore from the Laco Sur apatite-iron

791 oxide ore deposit at El Laco.

792

793 1.4 Bafq, Central Iran:

794 The Bafq-Saghand metallogenic zone is located in the Kashmar-Kerman Tectonic Zone (KKTZ)

795 in Central Iran and comprises about 34 recorded iron ore mineralizations with nearly ~1500 Mt ore

90,91 796 with an average grade of 55% Fe . Among these deposits, larger apatite-iron-oxide ores that are

797 currently mined include Chadormalu, Choghart, Se-Chahun, Lakke Saih and Esfordi. Some

798 deposits, such as Esfordi and Gazestan, have high-grade apatite mineralizations and represent

799 important phosphorus resources.

800 The deposits show a spectrum of mineralization styles such as massive orebodies, metasomatic

801 replacements, stockworks and veins. The dominant minerals are magnetite, apatite and actinolite.

802 Although the main Fe mineral is magnetite, all gradations towards hematite (through martitization)

92 803 occur .

804 Most of the iron ore bodies occur as dome-shaped discordant to concordant structures, which

805 consist mostly of lenses or irregular masses of massive magnetite surrounded by ore breccia and

806 disseminated magnetite in the host rocks. In some places irregular bodies of massive magnetite are

807 enclosed by a stockwork of magnetite, actinolite and apatite veins. An important feature of these

808 deposits is that they frequently display gradational contacts with their host rocks. Sharp contacts are

92 809 generally restricted to structurally controlled zones . The apatite-iron-oxide mineralizations in the

810 Bafq-Saghand zone are hosted by dolomitic and rhyolitic rocks of Cambrian Volcano-Sedimentary

811 Units (CVSU).

812 The geological setting of the KKTZ is linked to a major episode of late Neoproterozoic to Early

43 813 Cambrian orogenic activity in an active continental-margin environment .

814 The Origin of low-Ti apatite-iron oxide deposits of the Bafq-Saghand area has long been a

815 matter of debate. In this case, several models have been proposed for these deposits which include i)

Troll et al. NCOMMS-18-08734B 37 93–95 96 11,97 816 carbonatitic magmatism , ii) liquid immiscibility , iii) magmatic , iv) alkaline magmatism

42 98 92,99–101 817 , v) magmas of the Kiruna-type , vi) hydrothermal Kiruna-type , vii) banded iron

102,103 91 818 formations and viii) magmatic-hydrothermal .

819

820 2. Sampling locations for Layered Igneous Intrusion reference materials

821

822 2.1 Bushveld, South Africa: The magnetite sample in this study comes from the Rustenburg

823 Layered Suite (RLS) of the Bushveld complex, which is an 8 km thick succession of layered mafic

104,105 824 and ultramafic rocks with an age of about 2.1 Ga . The RLS is divided up into the Lower,

825 Critical, Main and Upper zone, with the Critical Zone being the economically most important one

106 826 since it holds the world’s largest chromite and platinum-group element deposits . However,

827 magnetite is mined within the Upper Zone, which contains about 20 m in total thickness of pure

828 magnetite in the form of several magnetite layers within 2 km thick magnetite-bearing gabbroic

107 829 rocks . The magnetite layers vary in thickness between 0.1 and 10 m and contain some silicates,

105 830 mostly plagioclase feldspar . The most prominent layer is called the Main Magnetite Layer from

831 which the magnetite used in this study originates. The magnetite layers in the Upper zone are of

832 magmatic origin and are assumed to have been formed by cycles of magma mixing of different

108 833 FeO-rich magmas and subsequent cumulate emplacement .

834

835 2.2 Panzhihua, Sichuan Province, China: The layered igneous intrusion of Panzhihua is located

836 in the Panxi Mining District, Sichuan Province, in South West China and is part of the Emeishan

837 Large Igneous Province. It is a relatively unmetamorphosed and undeformed, 2 km thick, sill-like

838 gabbroic intrusion which dips about 50°-60° towards the NW and extends about 19 km from NE to

109 839 SW . The intrusion is concordantly emplaced within late Neoproterozoic dolomite limestones,

109 840 permian syenites and Triassic shales and coal measures and is itself 263 Ma old . The intrusion is

841 divided into four zones based on differences in internal structure and iron oxide mineralizations.

Troll et al. NCOMMS-18-08734B 38 842 These four zones are the marginal, lower, middle and upper zone. Iron ore occurs in the lower and

843 middle zones. The mineralizations consist of both, massive lens-shaped or tabular ore bodies up to

844 60 m in thickness as well as disseminated ore. The massive ore bodies consist of >80% Ti-

845 magnetite with variable amounts of clinopyroxene, plagioclase, and olivine. The average ore grade

109 846 comprises 43 wt % FeO, 11.68 wt % TiO2, and 0.30 wt % V2O5 . On the basis of texture (e.g.

847 vesicles), geochemistry and the absence of evidence for a skarn origin, the mineralization is

109 848 interpreted to have formed from an oxide enriched melt . The Panzhihua deposit is currently

109 849 mined and holds a reserve of 1333 Mt of ore . The two Panzhihua samples of this study represent

850 massive Ti-magnetite ore.

851

852 2.3 Ruoutevare, Norrbotten, Sweden: The geology of the Ruoutevare area in Norrbotten northern

853 Sweden comprises ultrabasic rock types such as peridotite and pyroxenite, which are associated

110,111 854 with anorthosite and gabbro of Precambrian age . Associated with the gabbro intrusion is a

855 deposit of iron ore in the form of Ti-bearing magnetite layers with a grade of 54.2 wt. % FeO and

112,113 856 11 wt. % TiO2 .

857

858 2.4 Taberg mine, Småland, Sweden: The Fe-Ti mineralization at Taberg is located in southern

859 Sweden, about 12 km to the south of Lake Vättern, within the Protogine Zone, a 1.2 Ga old. 20 km

114–116 860 wide and several 100 km long belt of ductile and brittle deformation . The ore deposit consists

861 of 1.2 Ga old troctolites (e.g. olivine gabbro) with high contents of Ti-rich magnetite

862 (“titanomagnetite”) and which have been affected by the late Sveconorwegian amphibolite facies

114,116 863 metamorphism . The ore body has a dimension of c. 1 х 0.4 km and is hosted by an

114,117 864 amphibolitised gabbro-dolerite, which has intruded the surrounding Småland granites . The

865 ore holds between 26 and 35 % “titanomagnetite” with a content of 28.7–32.3 % FeO (Sandecki

866 2000). Within the ore are plagioclase-rich layers which give the appearance of a layered igneous

114,117 867 intrusion . The ore is supposed to have been formed as a magmatic cumulate which resulted

Troll et al. NCOMMS-18-08734B 39 114,116 868 from gravitational settling within a gabbroic magma . The sample in this study is a massive Ti-

869 magneite ore from the mine at Taberg.

870

871 2.5 Ulvön, Ångermanland, Central Sweden: The mineral ulvöspinel derive its name after the

118 872 Ulvö island, where the layered igneous Ulvö Gabbro/dolerite Complex is found at the east coast

873 of central Sweden. The intrusion consists of several gently dipping lopoliths, 3080 km in diameter

119,120 874 and 250300 m in thickness . These gabbroic lopoliths contain alternate bands of mafic and

875 more felsic layers with thicknesses between 0.5 cm and 1 m, likely a result of magmatic cumulus

121 876 processes . These rocks are ~1.25 Ga old and have not been affected by regional metamorphic

120,121 877 overprint or deformation . Ti-magnetite occurs in distinct layers, like for example in the

878 rhythmically layered zone of Norra Ulvön, which contains up to 10 cm thick bands with >50 % Fe-

121,122 121 879 Ti oxides . Such layers have also been mined for their metals . Other common minerals in

119,121 880 the Ulvö Gabbro Complex are plagioclase (labradorite), olivine, and clinopyroxene . The

881 sample used in this study is a massive Ti-magnetite ore from Norra Ulvön.

882

883 3. Sampling locations for volcanic reference materials

884 3.1 Canary Islands: Tenerife located in the centre of the Canary archipelago is the largest (2058

2 123,124 885 km ) and highest (3718 m) of the island group which is situated over the Canary hot spot .

886 Volcanic activity on the island dates back to about 6.5 Ma and is today seen in several small

124 887 volcanoes and the Teide-Pico Viejo edifice . Samples for this study originate from dykes in the

125,126 888 NE rift zone on the island. The samples comprise ankaramites and basanites .

889

890 3.2 Cyprus: The island of Cyprus is located in a zone of underthrusting in the eastern part of the

891 Mediterranean Sea, where the African plate is being pushed into the Eurasian plate. It can be

892 divided into five more or less parallel belts, which trend approximately eastwards and are convex

127 893 towards the south . Four of these belts are dominated by sedimentary rocks, most commonly

Troll et al. NCOMMS-18-08734B 40 127,128 894 limestones and loose sediments ranging in age from the Triassic to recent . The fifth belt, the

895 Troodos igneous massif, is dominated by mafic and ultramafic igneous rocks and represents an

127,129 896 ophiolite sequence obducted during the alpine orogeny . The Troodos massif is about 11 km

897 thick and divided up into basic and ultrabasic rocks in the center, the sheeted intrusive complex and

127 898 the peripheral pillow lavas . The rocks have been affected by metamorphism represented by

127 899 diabase and serpentinite . The sample used in this study comes from a dolerite dyke near Agros in

130 900 the central part of the Troodos complex .

901

902 3.3 Iceland: The volcanic island of Iceland is located directly over the point in the North Atlantic,

131 903 where asthenospheric flow interacts with a deep seated mantle plume , whose current plume

904 channel lies beneath the Vatnajökull glacier and represents the plate boundary of the Mid-Atlantic

132 905 ridge . Extensive volcanism is common on the island with more than 18 active volcanoes, which

131 906 are often associated with rift zones and their volcanic fissure swarms . Volcanic eruptions, often

133 907 of explosive nature due to lava-snow interaction, occur every three to four years . The main

134 908 eruption products are tholeiitic basalts as well as basaltic andesites . The magnetite sample used

909 in this study comes from a basaltic lava bomb erupted from the Skjaldbreiður volcano in SW

910 Iceland.

911

135,136 137 912 3.4 Indonesia: The investigated magnetite samples come from the Anak Krakatau , Agung ,

138 139 140 913 Gede , Kelut and Merapi volcanoes on the Indonesian Islands of Java and Bali. These are

914 part of the Western Sunda-Banda arc, which developed during the Cenozoic through subduction of

141,142 915 the Indian-Australian plate under the Eurasian plate . Calc-alkaline volcanism with dacites and

142,143 916 andesites are the typical eruption products in most recent times . Beside volcanic rocks, the

142 917 area also hosts several gold, tin and copper deposits associated with subduction zone volcanism .

918

Troll et al. NCOMMS-18-08734B 41 919 3.5 New Zealand: Mount Ruapehu is a 2797 m high stratovolcano at the southern end of the Taupo

144 920 Volcanic Zone (TVZ) on the North Island of New Zealand . It is the largest and currently active

921 volcano on the Northern Island with the most recent eruption in 2007 and several other eruptive

145,146 922 events during the last hundred years . Volcanic activity in the TVZ is associated with the

923 subduction of the Pacific Plate beneath the Australian Plate along the Hikurangi-Kermadec Trench

147,148 924 system . Mt. Ruapehu is underlain by Mesozoic meta-greywacke, which in turn is underlain by

146,149 925 oceanic, metamorphosed igneous crust . The typical eruption products are subduction zone

150,151 926 andesites and dacites with porphyritic textures . The magnetite content of the volcanic rocks at

146 927 Mt. Ruapehu varies between less than 1 % and up to 6 % and is just over 1% on average .

928 Samples used in this study are two dacite rocks that come from the southern Flank of Mt. Ruapehu.

929

930 4. Sampling sites for low-temperature hydrothermal ore deposit reference materials

931 4.1 Björnberget mines, Grängesberg Mining District, Bergslagen, Central Sweden: The

932 Björnberget mines are located c. 3 km east-southeast of the Grängesberg Export field in the

933 northwestern part of the Bergslagen ore province, south central Sweden. The magnetite-dominated

934 iron ores occur as in part carbonate-banded, skarn-associated types, which may locally progress into

152,153 935 true skarn iron ores . The major iron ore zones are northeast-striking, and variably, but steeply

152 936 (to 80º) dipping towards the southeast . The ore-bearing carbonates and skarns of the Björnberget

937 mines are in turn hosted by c. 1.91-1.88 Ga old felsic metavolcanic rocks (rhyolitic to rhyodacitic in

938 composition), which are variably altered, but of which more well-preserved types exhibit what can

86,153 939 be interpreted as primary laminations . Later regional metamorphism to amphibolite facies

940 grade, as well as three stages of ductile deformation has affected the older rocks in the area,

85,86 941 including the Björnberget ore and its host rocks .

942

943 4.2 Dannemora mine, Bergslagen, southcentral Sweden: The Dannemora skarn iron ore deposit

944 is situated in the eastern part of the Bergslagen ore province. It consists of 25 iron ore bodies, which

Troll et al. NCOMMS-18-08734B 42 945 are typically enriched in manganese, and some smaller sulphide deposits, which are all hosted by c.

946 1.9 Ga old meta-volcanic and meta-sedimentary rocks. These include meta-dacites and meta-

947 rhyolites as well as calcitic and dolomitic meta-limestones (marbles). The latter can locally exhibit

154 948 preserved stromatolitic textures . The metavolcanic rocks have been interpreted as pyroclastic

949 flow and air-fall deposits which together with the limestones were deposited in open marine,

154 950 lagoonal and terrestrial (subaerial) environments . The dolomitic marbles at Dannemora are very

154,155 951 dark-coloured due to a content of 5-30 % of fine magnetite . The area has been affected by

952 greenschist facies metamorphism during the Svecokarelian orogeny and deformed at least twice,

154,156 953 leading to isoclinal folding . The iron ore at Dannemora consists to a great extent of massive

954 strata-bound magnetite ore dipping 65-70° to the west, within an east-south-east syncline, and has

154,155 955 an iron content of between 30 and 50 % . The formation of the ore is believed to relate to

956 circulating metal-bearing hydrothermal fluids. These fluids altered silica-rich units in the area and

957 formed the major, mineralised skarn units through extensive reactions with the pre-exisiting

154 958 limestones . In some cases fluid rock interaction and evaporation may have altered the fluid

959 composition and led to enrichment of heavier elements and isotopes. The Dannemora samples used

960 in this study are all calcite-bearing magnetite ores and come from various locations in the

961 Dannemora mine (see Supplementary Table 1).

962

963 4.3 Striberg mine, Bergslagen, Central Sweden: Banded Iron Formations (BIF), such as the

964 deposit at Striberg, are found in the Bergslagen ore province, and are hosted by the 1.91-1.88 Ga

86,157 965 old metavolcanic rocks . In the Striberg area, extensive banded iron formations occur

966 associated with skarn iron ores, in a complexly folded and deformed succession. The succession

967 shows a main structural trend in a northwest-southeasterly direction, and with moderately steep (c.

152 968 45-60º) dips to the northeast . The main Striberg deposit consists mainly of alternating quartz and

969 hematite-rich layers, normally of 1-10 mm thickness, and the silica content varies between 18 %

24,86,157 970 and 28 % . Typical for BIF deposits, there is a dominance of hematite as the main Fe-bearing

Troll et al. NCOMMS-18-08734B 43 971 mineral, however, magnetite is also present as an alteration product of the latter, and in some BIF

158 972 ore types at Striberg the hematite has been completely converted to magnetite . The iron content

86,157 973 of the deposits lies between 30 % and 55 % .

974

975

Troll et al. NCOMMS-18-08734B 44 976 Supplementary Tables and Figures 977 978 Supplementary Fig.1

979 980 981 Supplementary Fig. 1

Troll et al. NCOMMS-18-08734B 45 982 Greyscale BSE images of samples presented in Fig. 2, taken with a Field Emission-EPMA JXA-

983 8530F JEOL hyperprobe. Except for some samples from El Laco, all chosen magnetite ores appear

984 homogeneous with no discernable zonation or rims of alteration. Sample numbers: a) and b) K-MT-

985 1079-303 c and d) KES090020 e) 13-C-219 f) 13-C-219 g and h) LS-11-4.

986 987 Supplementary Fig.2

988 989 Supplementary Fig. 2

990 Equilibrium source calculations for Fe and O isotopes. a) Calculated isotopic compositions of

991 magma in equilibrium with magnetite samples from volcanic and plutonic reference materials at

Troll et al. NCOMMS-18-08734B 46 992 magmatic T (> 800 C°). Calculated isotopic compositions of magma or fluid in equilibrium with

993 magnetite samples from the Kiruna (b), Grängesberg (c), El Laco and Bafq Mining Districts (d) and

14,48 994 Chilean Iron Belt, Pea Ridge and Pilot Knob apatite-iron oxide ore compared to the reference

5,39,54,57,58,69,70,162,164–168 995 fields for common magmatic and hydrothermal sources . Some of the

996 calculated ore forming magmas and fluids that are in equilibrium with apatite-iron oxide ore

56 997 magnetites are enriched in Fe relative to common magmatic sources and plot above the currently

998 accepted reference fields for intermediate magmas and high-temperature magmatic fluids. This may

999 be the result of fractionation of the heavy isotope into the melt during early silicate crystallization in

1000 some mafic melts and consequently subsequent magnetite crystals may be enriched in the heavy

46 1001 iron isotope (see text for details). Vein, disseminated and overprinted ore samples (Ve-Di) show

1002 equilibrium with common magmatic sources only at low-temperatures (<400 °C) representing a

1003 secondary component under more hydrothermal conditions (f). For simplification only values with

56 1004 δ Fe ≤ +0.6 are plotted.

1005

Troll et al. NCOMMS-18-08734B 47 Supplementary Table 1. Oxygen and iron isotope analysis for magnetite from apatite-iron oxide ores and reference materials δ18O δ56Fe Sample Sample Description Sample Provenance 2σ 2σ in ‰ in ‰ Apatite-iron oxide ore Kiruna Mining District (KMD), Northern Sweden Kiruna-17NY28 Banded massive (Ap-)magnetite ore Kiirunavaara mine, Kiruna 4.1 ±0.2 0.19 ±0.03 K-Mt-1 (907/75) Massive magnetite ore Kiirunavaara mine, Kiruna -1.0 ±0.2 0.20 ±0.03 Ki-Mi-2a (1079/251) Massive magnetite ore Kiirunavaara mine, Kiruna 0.1 ±0.2 0.21 ±0.02 Ki-Mi-2b (1079/251) Massive magnetite ore Kiirunavaara mine, Kiruna -0.7 ±0.2 0.22 ±0.04 K-Mt-1079/303 Massive magnetite ore Kiirunavaara mine, Kiruna -0.3 ±0.2 0.27 ±0.04 K-Mt-1079/437 Massive magnetite ore Kiirunavaara mine, Kiruna 0.6 ±0.2 0.16 ±0.02 M1931* Massive magnetite ore Kiirunavaara mine, Kiruna 0.1 ±0.2 - - M1937* Skeletal magnetite ore Kiirunavaara mine, Kiruna 1.4 ±0.2 - - LVA-3 Massive magnetite ore Luossavaara mine, Kiruna 1.2 ±0.2 0.23 ±0.02 LVA-FW-1 Massive magnetite ore Luossavaara mine, Kiruna 1.4 ±0.2 0.12 ±0.04 LVA-FW-2 Massive magnetite ore Luossavaara mine, Kiruna 1.2 ±0.2 0.27 ±0.03 K-Mt-3 Massive magnetite ore Mertainen mine, Kiruna 1.9 ±0.2 0.41 ±0.03 K-Mt-4 Massive magnetite ore Mertainen mine, Kiruna 1.6 ±0.2 0.29 ±0.03 M7557* Massive magnetite ore Rektorn mine, Kiruna 2.5 ±0.2 - -

Grängesberg Mining District (GMD), Central Sweden DC717-KES090068 Massive (Ap-)magnetite ore Grängesberg mine, Grängesberg 1.2 ±0.2 0.40 ±0.03 DC717-KES090070 Massive (Ap-)magnetite ore Grängesberg mine, Grängesberg 1.8 ±0.2 0.24 ±0.03 DC717-KES090072 Massive (Ap-)magnetite ore Grängesberg mine, Grängesberg 0.9 ±0.2 0.33 ±0.03 Magnetite vein in intermediate volcanic DC717-KES090084 Grängesberg mine, Grängesberg -1.1 ±0.2 0.11 ±0.03 rock DC690-KES090011 Massive (Ap-)magnetite ore Grängesberg mine, Grängesberg 2.8 ±0.2 0.31 ±0.03 DC690-KES090012 Massive (Ap-)magnetite ore Grängesberg mine, Grängesberg 1.2 ±0.2 0.31 ±0.04 DC690-KES090020 Massive Ap-veined magnetite ore Grängesberg mine, Grängesberg 1.1 ±0.2 0.30 ±0.04 DC690-KES090024 Massive (Ap-)magnetite ore Grängesberg mine, Grängesberg 1.0 ±0.2 0.26 ±0.03 Ap-veined/banded massive magnetite DC690-KES090027 Grängesberg mine, Grängesberg 1.2 ±0.2 0.29 ±0.03 ore Silicate-spotted massive (Ap-)magnetite DC690-KES090030 Grängesberg mine, Grängesberg 1.8 ±0.2 0.39 ±0.04 ore Coarse-grained Ap-spotted massive DC690-KES090034 Grängesberg mine, Grängesberg 0.5 ±0.2 0.27 ±0.03 magnetite ore Disseminated magnetite in intermediate DC690-KES090044 Grängesberg mine, Grängesberg -1.0 ±0.2 0.24 ±0.03 volcanic rock DC575-KES103011 Magnetite-dominated massive ore Grängesberg mine, Grängesberg 1.8 ±0.2 0.31 ±0.03 DC575-KES103016 Coarse, massive (Ap-)magnetite ore Grängesberg mine, Grängesberg 1.5 ±0.2 0.27 ±0.04 DC575-KES103003 Magnetite-dominated massive ore Grängesberg mine, Grängesberg 0.2 ±0.2 1.0 ±0.03 KES091013b Massive (Ap-)magnetite ore Blötberget mine, Blötberget 0.1 ±0.2 0.33 ±0.03

El Laco Ap-Fe-oxide deposit, Chile EJ-LS-11-1 Massive magnetite ore Laco Sur, El Laco 1.9 ±0.2 0.28 ±0.03 EJ-LS-11-2 Massive magnetite ore Laco Sur, El Laco -4.3 ±0.2 0.24 ±0.03 EJ-LS-11-3 Massive magnetite ore Laco Sur, El Laco -1.9 ±0.2 0.36 ±0.03 EJ-LS-11-4 Massive magnetite ore Laco Sur, El Laco 4.2 ±0.2 0.34 ±0.03 LS-2 Massive magnetite ore Laco Sur, El Laco1 4.3 ±0.2 0.27 ±0.03 LS-52 Massive magnetite ore Laco Sur, El Laco1 4.4 ±0.2 0.28 ±0.03

Bafq Mining District, Iran 13.C.88 Massive magnetite ore Sechahun, Bafq 2.4 ±0.2 0.20 ±0.06

Troll et al. NCOMMS-18-08734B 48 13.C.106 Massive magnetite ore Lakke Siah, Bafq 2.3 ±0.2 0.26 ±0.05 13.C.216 Massive magnetite ore Chadormalu, Bafq 0.6 ±0.2 0.24 ±0.03 13.C.217 Massive magnetite ore Chadormalu, Bafq 0.6 ±0.2 0.32 ±0.02 13.C.219 Massive magnetite ore Chadormalu, Bafq 2.8 ±0.2 0.27 ±0.07 14.22 Massive magnetite ore Esfordi, Bafq 3.4 ±0.2 0.32 ±0.01

Plutonic reference material Ruoutevare Ti-magnetite, layered igneous intrusion Kvikjokk, Norrbotten, Sweden2 3.2 ±0.2 0.31 ±0.03 Ulvön Ti-magnetite, layered igneous intrusion Ulvön island, Ångermanland, Sweden2 4.0 ±0.2 0.13 ±0.03 Taberg Ti-magnetite, layered igneous intrusion Iron mine, Taberg, Småland, Sweden3 4.1 ±0.2 0.23 ±0.04 EM419 Massive Fe-Ti magnetite ore Northern pit, Panzhihua, China4 4.8 ±0.2 0.61 ±0.05 EM424 Massive Fe-Ti magnetite ore Nalahe, Panzhihua, China4 2.8 ±0.2 0.12 ±0.04 Upper Zone, Bushveld Complex, Bushveld Massive magnetite ore 1.8 ±0.2 - - South Africa4 3 Gabbrobomb Magnetite from a gabbro xenolith NW-Flank, Skjaldbreiður, Iceland 4.6 ±0.2 0.46 ±0.03

Volcanic reference material TEF-NER-18 Magnetite from an ankaramite dyke NE Rift Zone, Tenerife, Spain - - 0.07 ±0.05 TEF-NER-57B Magnetite from an ankaramite dyke NE Rift Zone, Tenerife, Spain 3.7 ±0.2 0.16 ±0.02 TEF-NER-70 Magnetite from a pyroxene phyric dyke NE Rift Zone, Tenerife, Spain 3.7 ±0.2 0.10 ±0.02 MG-07 Igneous magnetite from dacite S-Flank, Mt. Ruapehu, New Zealand5 - - 0.32 ±0.03 MG-09 Igneous magnetite from dacite S-Flank, Mt. Ruapehu, New Zealand5 - - 0.29 ±0.03 Kelut A1 Igneous magnetite from basaltic andesite Mt. Kelut, Java, Indonesia - - 0.10 ±0.04 GD-D-2 Igneous magnetite from basaltic andesite Gede Dome, Java, Indonesia - - 0.12 ±0.03 AK-B1 Igneous magnetite from basaltic andesite SE-Flank, Anak Krakatau, Indonesia - - 0.06 ±0.03 AK-B3 Igneous magnetite from basaltic andesite SE-Flank, Anak Krakatau, Indonesia - - 0.16 ±0.03 A-BA-1 Igneous magnetite from basaltic andesite Mt. Agung, Bali, Indonesia - - 0.18 ±0.05 M-BA06-KA-3 Igneous magnetite from basaltic andesite Mt. Merapi, Java, Indonesia 3.9 ±0.2 0.17 ±0.03 83/CRS/6 Igneous magnetite from a dolerite dyke Agros, Troodos Massif, Cyprus6 - - 0.34 ±0.03

Low-temperature or hydrothermal magnetites KES091007B Calcite bearing-magnetite ore Björnberget, Sweden -0.8 ±0.2 -0.02 ±0.03 DM-1 Iron-skarn magnetite ore Botenhäll, Dannemora, Sweden7 -0.4 ±0.2 -0.36 ±0.03 DM-2 Iron-skarn magnetite ore Norrnäs 3, Dannemora, Sweden7 -0.7 ±0.2 0.01 ±0.03 DM-3 Iron-skarn magnetite ore Konstäng, Dannemora, Sweden7 2.1 ±0.2 -0.43 ±0.03 DM-4 Iron-skarn magnetite ore Strömsmalmen, Dannemora, Sweden7 -0.6 ±0.2 -0.35 ±0.03 EJ092008 Magnetite from a banded iron formation deposit Striberg, Bergslagen, Sweden -1.2 ±0.2 -0.57 ±0.03

* Data from Lundh (2014) (ref.159) 1. Samples donated by Dr. Jan-Olov Nyström, Naturhistoriska Riksmuseet, Stockholm, Sweden 2. Samples from the sample collection of the Geological Survey of Sweden, Uppsala, Sweden 3. Sample from the sample collection, Department of Earth Science, Uppsala University, Sweden 4. Samples collected and donated by Prof. Nicholas Arndt, Université Joseph Fourier, Grenoble, France 5. Samples donated by Prof. John Gamble, Department of Geology, Victoria University of Wellington, New Zealand 6. Samples donated by Prof. Christopher J. Stillman, Department of Geology, Trinity College Dublin, Ireland 7. Samples donated by Gunnar Rauseus at Dannemora Mineral AB, Österbybruk, Sweden 1006 1007

Troll et al. NCOMMS-18-08734B 49 Supplementary Table 2. Results for the magma (900°C)/magmatic water (800°C) equilibrium re-calculation δ18O Value δ18O Value δ18O Value δ56Fe δ56Fe Sample Rock type δ18O δ56Fe andesite fit dacite fit water fit water magma

Kiruna Massive ore 4.1 0.19 8.1 ! 8.4 ! 9.3 ! -0.05 0.16 K-Mt-1 Massive ore -1.0 0.20 3.0 X 3.4 X 4.3 X -0.04 0.17 Ki-Mi-2a Massive ore 0.1 0.21 4.0 X 4.4 X 5.2 ! -0.04 0.18 Ki-Mi-2b Massive ore -0.7 0.22 3.3 X 3.6 X 4.5 X -0.02 0.19 K-Mt-1079/303 Massive ore -0.3 0.27 3.6 X 4.0 X 4.9 X 0.03 0.24 K-Mt-1079/437 Massive ore 0.6 0.16 4.6 X 4.9 X 5.8 ! -0.08 0.13 M1931 Massive ore 0.1 - 4.1 X 4.4 X 5.3 ! - - M1937 Massive ore 1.4 - 5.4 X 5.7 ! 6.6 ! - - LVA-3 Massive ore 1.2 0.23 5.2 X 5.5 X 6.4 ! -0.02 0.19 LVA-FW-1 Massive ore 1.4 0.12 5.4 X 5.7 ! 6.6 ! -0.12 0.09 LVA-FW-2 Massive ore 1.2 0.27 5.1 X 5.5 X 6.4 ! 0.03 0.24 K-Mt-3 Massive ore 1.9 0.41 5.9 ! 6.2 ! 7.1 ! 0.17 0.38 K-Mt-4 Massive ore 1.6 0.29 5.6 X 5.9 ! 6.8 ! 0.05 0.26 M7557 Massive ore 2.5 - 6.5 ! 6.8 ! 7.7 ! - -

DC717-KES090068 Massive ore 1.1 0.40 5.1 X 5.4 X 6.3 ! 0.16 0.37 DC717-KES090070 Massive ore 1.8 0.24 5.8 ! 6.1 ! 7.0 ! -0.01 0.21 DC717-KES090072 Massive ore 0.9 0.33 4.8 X 5.2 X 6.1 ! 0.09 0.30 DC690-KES090011 Massive ore 2.8 0.31 6.7 ! 7.1 ! 8.0 ! 0.07 0.28 DC690-KES090012 Massive ore 1.2 0.31 5.2 X 5.5 X 6.4 ! 0.07 0.28 DC690-KES090020 Massive ore 1.1 0.30 5.0 X 5.4 X 6.3 ! 0.06 0.27 DC690-KES090024 Massive ore 1.0 0.26 5.0 X 5.3 X 6.2 ! 0.02 0.23 DC690-KES090027 Massive ore 1.2 0.29 5.2 X 5.5 X 6.4 ! 0.05 0.26 DC690-KES090030 Massive ore 1.8 0.39 5.8 ! 6.1 ! 7.0 ! 0.15 0.36 DC690-KES090034 Massive ore 0.5 0.27 4.5 X 4.8 X 5.7 ! 0.03 0.24

DC575-KES103003 Massive ore 1.8 0.31 5.7 ! 6.1 ! 7.0 ! 0.07 0.28 DC575-KES103011 Massive ore 1.5 0.27 5.5 X 5.8 X 6.7 ! 0.03 0.24 DC575-KES103016 Massive ore 0.1 0.33 4.1 X 4.4 X 5.3 ! 0.09 0.30 DC575-KES103003 Massive ore 0.2 1.0 4.2 X 4.5 X 5.4 ! 0.76 0.97 Disseminated DC690-KES090044 -1.0 0.24 3.0 X 3.3 X 4.2 X 0.00 0.21 magnetite Magnetite DC717-KES090084 -1.1 0.11 2.8 X 3.1 X 4.1 X -0.13 0.08 vein

EJ-LS-11-1 Massive ore 1.9 0.28 5.9 ! 6.2 ! 7.1 ! 0.04 0.25 EJ-LS-11-2 Massive ore -4.3 0.24 -0.3 X 0.0 X 0.9 X 0.00 0.21 EJ-LS-11-3 Massive ore -1.9 0.36 2.1 X 2.4 X 3.3 X 0.12 0.33 EJ-LS-11-4 Massive ore 4.2 0.34 8.1 ! 8.5 ! 9.4 ! 0.10 0.31 LS-2 Massive ore 4.3 0.27 8.2 ! 8.6 ! 9.5 ! 0.03 0.24 LS-52 Massive ore 4.4 0.28 8.4 ! 8.7 ! 9.6 ! 0.04 0.25

13.C.88 Massive ore 2.4 0.20 6.4 ! 6.7 ! 7.6 ! -0.04 0.17 13.C.106 Massive ore 2.3 0.26 6.2 ! 6.6 ! 7.5 ! 0.02 0.23 13.C.216 Massive ore 0.6 0.24 4.5 X 4.9 X 5.8 ! 0.00 0.21 13.C.217 Massive ore 0.6 0.32 4.6 X 4.9 X 5.8 ! 0.08 0.29 13.C.219 Massive ore 2.8 0.27 6.8 ! 7.1 ! 8.0 ! 0.03 0.24 14.22 Massive ore 3.4 0.32 7.4 ! 7.7 ! 8.6 ! 0.08 0.29

Troll et al. NCOMMS-18-08734B 50 Oxygen:

1. 1000lnα (mt-basalt) = -3.4 ‰; 1000lnα (mt-andesite) = -4.0 ‰; 1000lnα (mt-dacite) = -4.3 ‰; regular range of basalts, arc andesites/dacites +5.7 to +8 ‰ (ref.57,58,160) 161,162 2. 1000lnα (mt-water 800°C) = -5.2 ‰; regular range for magmatic waters 5-10 ‰ (ref. ) Iron: 39 1. 1000lnα (mt-magma) = 0.03 ‰; regular range of arc andesites/dacites +0.00 to +0.12 ‰ (ref. ) 39,70 2. 1000lnα (mt-water 800°C) = 0.24 ‰; regular range for magmatic waters 0.00 to -0.35‰ (ref. ) ! = in equilibrium with magma/magmatic water ; X= not in equilibrium with common magmatic values

1008 1009 Supplementary Table 3. Results for the magmatic water (625°C) equilibrium re-calculation δ18O Value δ56Fe Sample Rock type δ18O δ56Fe water fit water Ki-Mi-2a Massive ore 0.1 0.21 6.3 ! -0.14 K-Mt-1079/437 Massive ore 0.6 0.16 6.8 ! -0.19 M1931 Massive ore 0.1 - 6.3 !

DC690-KES090034 Massive ore 0.5 0.27 6.7 ! -0.08 DC575-KES103003 Massive ore 0.2 1.0 6.4 ! 0.65 DC575-KES103016 Massive ore 0.1 0.33 6.3 ! -0.02

13.C.216 Massive ore 0.6 0.24 6.8 ! -0.11 13.C.217 Massive ore 0.6 0.32 6.8 ! -0.03 161 Oxygen: 1000lnα (mt-water 625°C) = -6.2 ‰ (ref. ) 39 Iron: 1000lnα (mt-water 625°C) = 0.35 ‰ (ref. ) ! = in equilibrium with magma/magmatic water ; X= not in equilibrium with common magmatic values (ref.70,162)

Troll et al. NCOMMS-18-08734B 51 1010 Supplementary Table 4. Results for the volcanic and plutonic reference material equilibrium re-calculation δ18O δ56Fe Sample Rock type δ18O δ56Fe basalt magma TEF-NER-57B Magnetite from an ankaramite dyke 3.7 0.16 7.1 0.13 TEF-NER-70 Magnetite from a pyroxene phyric dyke 3.7 0.10 7.1 0.07 M-BA06-KA-3 Igneous magnetite from basaltic andesite 3.9 0.17 6.2 0.14 TEF-NER-18 Magnetite from an ankaramite dyke - 0.07 7.0 0.04 Kelut A1 Igneous magnetite from basaltic andesite - 0.10 7.0 0.07 GD-D-2 Igneous magnetite from basaltic andesite - 0.12 7.5 0.09 AK-B1 Igneous magnetite from basaltic andesite - 0.06 6.4 0.03 AK-B3 Igneous magnetite from basaltic andesite - 0.16 6.4 0.13 A-BA-1 Igneous magnetite from basaltic andesite - 0.18 6.7 0.15

Gabbrobomb Magnetite from a gabbro xenolith 4.6 0.46 8.0 0.43 Ruoutevare Ti-magnetite, layered igneous intrusion deposit 3.2 0.31 6.6 0.28 Ulvön Ti-magnetite, layered igneous intrusion deposit 3.9 0.13 7.4 0.10 Taberg Ti-magnetite, layered igneous intrusion deposit 4.1 0.23 7.5 0.20 EM419 Massive Fe-Ti magnetite ore 4.8 0.61 8.3 0.58 EM424 Massive Fe-Ti magnetite ore 2.8 0.12 6.2 0.09

Oxygen: 1000lnα (mt-basalt) = -3.4 ‰; 1000lnα (mt-andesite) = -4.0 ‰; 1000lnα (mt-dacite) = -4.3 ‰; regular range of basalts, arc andesites/dacites +5.7 to +8 ‰ (ref.57,58,160) 39 Iron: 1000lnα (mt-magma) = 0.03 ‰; regular range of arc andesites/dacites +0.00 to +0.12 ‰ (ref. ) For volcanic samples where no δ18O-value was obtained estimates for magma δ18O were taken from Jolis (2013) (ref.163). 1011 1012 Supplementary Table 5. Results for equilibrium re-calculation of literature data δ56Fe δ18O δ56Fe Sample δ18O δ56Fe δ18O water water andesite magma Los Colorados, Chilean Iron Belt (Bilenker et al. 2016) 05–3.30 2.41 0.22 6.9 -0.13 6.4 0.19 05–20.7 3.04 0.09 7.5 -0.26 7.0 0.06 05–32 2.75 0.22 7.3 -0.13 6.8 0.19 05–52.2 3.17 0.14 7.7 -0.21 7.2 0.11 05–72.9 2.36 0.13 6.9 -0.22 6.4 0.1 05–82.6 2.76 0.08 7.3 -0.27 6.8 0.05 05–90 2.99 0.21 7.5 -0.14 7.0 0.18 05–106 2.78 0.12 7.3 -0.23 6.8 0.09 05–126.15 2.48 0.1 7.0 -0.25 6.5 0.07 04–38.8 2.04 0.18 6.5 -0.17 6.0 0.15 04–66.7 1.92 0.18 6.4 -0.17 5.9 0.15 04–129.3 2.62 0.22 7.1 -0.13 6.6 0.19 04–104.4 2.43 0.24 6.9 -0.11 6.4 0.21

Pea Ridge and Pilot Knob (Childress et al. 2016) PR18 2.12 0.35 8.3 0 4.9 0.32 PR-64A 4.87 0.2 11.1 -0.15 7.7 0.17 PR-77A 5.11 0.21 11.3 -0.14 7.9 0.18 PR-82A 5.9 0.1 12.1 -0.25 8.7 0.07 PR-82B 7.03 0.07 13.2 -0.28 9.8 0.04 PR-37 4.5 0.07 10.7 -0.28 7.3 0.04

Troll et al. NCOMMS-18-08734B 52 PR-144 5.04 0.26 11.2 -0.09 7.8 0.23 PK-1145-965.8 6.68 0.19 12.9 -0.16 9.5 0.16 PK-1145-979.5 6.21 0.24 12.4 -0.11 9.0 0.21 Oxygen: 14 48 1. 1000lnα (mt-andesite) = -4.0 ‰ (ref. ) or -2.8 ‰ (ref. ) 14 48 2. 1000lnα (mt-water 625°C) = -4.5 ‰ (ref. ) or -6.2 ‰ (ref. ) Iron: 39 1. 1000lnα (mt-magma) = 0.03 ‰ (ref. ) 39 2. 1000lnα (mt-water 625°C) = 0.35 ‰ (ref. )

1013 1014 Supplementary Table 6. Results for hydrothermal fluid (375°C) equilibrium re-calculation δ18O δ56Fe Sample δ18O δ56Fe fluid fluid KES090044 -1.0 0.24 6.80 -0.26 KES090084 -1.1 0.11 6.70 -0.39 EJ-LS-11-2 -4.3 0.24 3.50 -0.26 EJ-LS-11-3 -1.9 0.36 5.90 -0.14 K-mt-1 -0.95 0.20 6.9 -0.30 Ki-mi-2b -0.69 0.22 7.1 -0.28 K-mt-1079/303 -0.33 0.27 7.5 -0.23 161 Oxygen: 1000lnα (mt-water 375°C) = -7.8 ‰ (ref. ) 39 Iron: 1000lnα (mt-water 375°C) = 0.5 ‰ (ref. )

1015

Troll et al. NCOMMS-18-08734B 53 1016 Reference List Supplementary Information 1017

1018 71. Lundbohm, H. Sketch of the Geology of the Kiruna district. Geol. Föreningen i Stock.

1019 Förhandlingar 32, 751–788 (1910).

1020 72. Romer, R. L., Martinsson, O. & Perdahl, J. A. Geochronology of the Kiruna iron ores and

1021 hydrothermal alterations. Econ. Geol. 89, 1249–1261 (1994).

1022 73. Martinsson, O., Billström, K., Broman, C., Weihed, P. & Wanhainen, C. Metallogeny of the

1023 Northern Norrbotten Ore Province, Northern Fennoscandian Shield with emphasis on IOCG

1024 and apatite-iron ore deposits. Ore Geology Reviews 78, 447-492 (2016).

1025 doi:10.1016/j.oregeorev.2016.02.011

1026 74. Lundberg, B. & Smellie, J. A. T. Painirova and mertainen iron ores: Two deposits of the

1027 kiruna iron ore type in Northern Sweden. Econ. Geol. 74, 1131–1152 (1979).

1028 75. Lindblom, S., Broman, C. & Martinsson, O. Magmatic-hydrothermal fluids in the

1029 Pahtohavare Cu-Au deposit in greenstone at Kiruna, Sweden. Miner. Depos. 31, 307-318

1030 (1996). doi:10.1007/BF02280794

1031 76. Bergman, S., Kübler, L. & Martinsson, O. (2001) Description of regional geological and

1032 geophysical maps of northern Norrbotten County (east of the Caledonian orogeny).

1033 Geological Survey of Sweden Ba 56, 110 pp.

1034 77. Berglund, J. & Andersson, U. B. Kinematic analysis of geological structures in Block 34,

1035 Kiirunavaara. (2014).

1036 78. Geijer, P. The iron ores of the Kiruna type. Geological Survey of Sweden C 367, 39 p

1037 (1931).

1038 79. Geijer, P. The Rektor ore body at Kiruna. Geological Survey of Sweden C 514, 18 p (1950).

1039 80. Harlov, D. E. et al. Apatite – monazite relations in the Kiirunavaara magnetite – apatite ore,

1040 northern Sweden. Chem. Geol. 191, 47–72 (2002).

1041 81. Andersson, U.B. Coeval iron oxide and silicate magmas; structural evidence for

1042 immiscibility and mingling at Kiirunavaara and Luossavaara, Sweden. 12th SGA Biennial

Troll et al. NCOMMS-18-08734B 54 1043 Metting, Proceedings, Uppsala, Sweden 12-15/8. Geological Survey of Sweden, 1635

1044 (2013).

1045 82. Jiao, J. Petrography and geochemistry of a section through the Blötberget apatite-iron oxide

1046 deposit, Bergslagen, south central Sweden. Master thesis, Uppsala University, Uppsala,

1047 Sweden, 72 p (2011).

1048 83. Jonsson, E. et al. Oxygen isotopes and geochemistry of Palaeoproterozoic Kiruna-type

1049 deposits in the Bergslagen province, central Sweden. Abstract volume, SGA biennial meeting

1050 2011, Antofagasta, Chile, 494–496 (2011).

1051 84. Johansson, H. Die eisenerzführende Formation in der Gegend von Grängesberg. Geol.

1052 Föreningens i Stock. Förhandlingar 31, 239–410 (1910).

1053 85. Högdahl, K., Troll, V. R., Persson-Nilsson, K. & Jonsson, E. Structural evolution of the

1054 apatite-iron oxide deposit at Grängesberg, Bergslagen, Sweden. In: E. Jonsson et al. (eds.),

1055 Mineral deposit research for a high tech world, 11th SGA Biennial Meeting, Proceedings,

1056 1650-1653 (2013).

1057 86. Stephens, M. B. et al. Synthesis of the bedrock geology on the Bergslagen region,

1058 Fennoscandian Shield, south-central Sweden. Geological Survey of Sweden Ba 58, 259 p

1059 (2009).

1060 87. Maksaev, V., Gardeweg, M., Ramírez, C. F. & Zentilli, M. Aplicación del método trazas de

1061 fisión (fission track) a la datación de cuerpos de magnetita de El Laco e Incahuasi en el

1062 altiplano de la Región de Antofagasta. In: Congreso Geológico Chileno, No. 5, Actas 1: B1-

1063 B23. Santiago (1988).

1064 88. Naranjo, J., Henrquez, F. & Nyström, J. Subvolcanic contact metasomatism at El Laco

1065 Volcanic Complex, Central Andes. Andean Geol. 37, 110–120 (2010).

1066 89. Velasco, F. & Tornos, F. Insights on the effects of the hydrothermal alteration in the El Laco

1067 magnetite deposit (Chile). Macla 16, 210–211 (2012).

1068 90. Torab, F.M. Geochemistry and metallogeny of magnetite apatite deposits of the Bafq Mining

Troll et al. NCOMMS-18-08734B 55 1069 District, central Iran. Univ.-Bibliothek (2008).

1070 91. Majidi, S. A., Lotfi, M., Emami, M. H. & Nezafati, N. The genesis of iron oxide-apatite

1071 (IOA) deposits: evidence from the geochemistry of apatite in Bafq-Saghand district, Central

1072 Iran. Geosciences 27, 233–244 (2017).

1073 92. Torab, F. M. & Lehmann, B. Magnetite-apatite deposits of the Bafq district, Central Iran:

1074 apatite geochemistry and monazite geochronology. Mineral. Mag. 71, 347–363 (2007).

1075 93. Förster, H. & Borumandi, H. Jungpräkambrische Magnetit-Lava und Magnetit-Tuffe aus

1076 dem Zentraliran. Die Naturwissenschaften 58, 524–524 (1971).

1077 94. Darvishzadeh, A. Investigation on Esfordi phosphate deposit. Journal of Science, University

1078 of Tehran, 2–24 (1983).

1079 95. Samani, B. A. Metallogeny of the Precambrian in Iran. Precambrian Res. 39, 85–106 (1988).

1080 96. Förster, H. & Jafarzadeh, A. The Bafq mining district in central Iran - a highly mineralized

1081 Infracambrian volcanic field. Econ. Geol. 89, 1697–1721 (1994).

1082 97. Moore, F. & Modabberi, S. Origin of Choghart iron oxide deposit, Bafq mining district,

1083 Central Iran: new isotopic and geochemical evidence. J. Sci. Islam. Repub. Iran 14, 259–270

1084 (2003).

1085 98. Williams, G. J. & Houshmandzadeh, A. (1966) A petrological and genetic study of the

1086 Choghart iron ore body and the surrounding rocks. Geological Survey of Iran, 18 p.

1087 99. Daliran, F. Kiruna-type iron oxide-apatite ores and “ apatitites ” of the bafq district , iran ,

1088 with an emphasis on the ree geochemistry of their apatites. Porter, TM 2, 303–320 (2002).

1089 100. Jami, M., Dunlop, A. C. & Cohen, D. R. Fluid inclusion and stable isotope study of the

1090 Esfordi apatite-magnetite deposit, central Iran. Econ. Geol. 102, 1111–1128 (2007).

1091 101. Bonyadi, Z., Davidson, G. J., Mehrabi, B., Meffre, S. & Ghazban, F. Significance of apatite

1092 REE depletion and monazite inclusions in the brecciated Se-Chahun iron oxide-apatite

1093 deposit, Bafq district, Iran: Insights from paragenesis and geochemistry. Chem. Geol. 281,

1094 253–269 (2011).

Troll et al. NCOMMS-18-08734B 56 1095 102. Aftabi, A. et al. Fluid inclusion and stable isotope study of the Esfordi apaptite-magnetite

1096 deposit,central Iran – A discussion. Econ. Geol. 104, 137–139 (2009).

1097 103. Mohseni, S. & Aftabi, A. Comment on “Significance of apatite REE depletion and monazite

1098 inclusions in the brecciated Sehchahun iron oxide–apatite deposit, Bafq district, Iran: Insights

1099 from paragenesis and geochemistry” by Bonyadi, Z., Davidson,G.J., Mehrabi, B., Meffre, S.,

1100 Ghazban, F [Chem. Geol. 281, 253–269]. Chem. Geol. 334, 378–381 (2012).

1101 104. Walraven, F., Armstrong, R. A. & Kruger, F. J. A chronostratigraphic framework for the

1102 north-central Kaapvaal craton, the Bushveld Complex and the Vredefort structure.

1103 Tectonophysics 171, 23–48 (1990).

1104 105. Harne, D. M. W. & Von Gruenewaldt, G.. Ore-forming processes in the upper part of the

1105 Bushveld complex, South Africa. J. African Earth Sci. 20, 77-89 (1995). doi:10.1016/0899-

1106 5362(95)00034-Q

1107 106. Saager, R. Metallische Rohstoffe von Antimon bis Zirkonium. Bank Vontobel, Zürich, 176 p.

1108 (1984).

1109 107. Cawthorn, R. & Molyneux, T. Vanadiferous magnetite deposits of the Bushveld Complex.

1110 In: Mineral Deposits of Southern Africa 2, edited by Anhaeusser C.R. and Maske S.,

1111 Geological Society of South Africa, Johannesburg, 2391 p. (1986).

1112 108. Irvine, T. N., & Sharpe, M. R. (1986) Magma mixing and the origin of stratiform oxide ore

1113 zones in the Bushveld and Stillwater Complexes, In: Metallogeny of Basic and Ultrabasic

1114 Rocks, edited by M. J. Gallagher, et al., pp. 183–198, Inst. of Min. Metall., London, U. K.

1115 109. Zhou, M. F. et al. Geochemistry, petrogenesis and metallogenesis of the panzhihua gabbroic

1116 layered intrusion and associated Fe-Ti-V Oxide deposits, sichuan province, SW China. J.

1117 Petrol. 46, 2253-2280 (2005). doi:10.1093/petrology/egi054

1118 110. Peterson W.: 'On the Iron Ore Field of Routivare in Norrbotten Lane.' Geol. Foren. Fork.,

1119 Vol. XV, 1893, p. 45, with map. H. Sjogren: 'Additional Notes on Routivare Iron Ore.' Ibid.,

1120 140 p.

Troll et al. NCOMMS-18-08734B 57 1121 111. Stigh, J. Ultramafiter i svenska delen av fjällkedjan – Möjligt slutförvaringsplats för

1122 radioaktivt avfall? Rapport till Svensk Kärnbränsleförsörjning AB, Geological Survey of

1123 Sweden, 57 p (1982).

1124 112. Landergren, S. On the Geochemistry of Swedish iron ores and associated rocks – A study on

1125 iron-ore formation. Geological Survey of Sweden C 469, 182 p (1948).

1126 113. Fischer, R. P. Vanadium Resources in Titaniferous Magnetite Deposits. (U.S. Government

1127 Printing Office, 1975).

1128 114. Sandecki, J. Mineralogical and genetical aspects of the Smålands Taberg Fe-Ti-V ore,

1129 Protogine Zone of southern Sweden. GFF 122, 351–358 (2000).

1130 115. Söderlund, P., Söderlund, U., Möller, C., Gorbatschev, R. & Rodhe, A. Petrology and ion

1131 microprobe U-Pb chronology applied to a metabasic intrusion in southern Sweden: A study

1132 on zircon formation during metamorphism and deformation. Tectonics 23, TC5005 (2004).

1133 doi:10.1029/2003TC001498

1134 116. Larsson, D. & Söderlund, U. Lu-Hf apatite geochronology of mafic cumulates: An example

1135 from a Fe-Ti mineralization at Sm??lands Taberg, southern Sweden. Chem. Geol. 224, 201–

1136 211 (2005).

1137 117. Hjelmqvist, S. The titaniferous iron–ore deposit of Taberg in the South of Sweden.

1138 Geological Survey of Sweden C 512, 55 p (1950).

1139 118. Mogensen, F. A ferro-ortho-titanate ore from Södra Ulvön. Geol. Föreningen i Stock.

1140 Förhandlingar 68, 578–587 (1946).

1141 119. Magnusson, K. Å. & Larson, S. Å. A palaeomagnetic investigation of the Ulvö dolerite,

1142 Ångermanland, central Sweden. Lithos (1977). doi:10.1016/0024-4937(77)90048-2

1143 120. Lundqvist, T. Beskrivning till berggrundskartan över Västernorrlands län (English

1144 summary). Geological Survey of Sweden Ba 31, 429 p (1990).

1145 121. Larson, S. Å, Hogmalm, K. J. & Meurer, W. P. Character and significance of spectacular

1146 layering features developed in the thin, alkali-basaltic sills of the Ulv?? Gabbro Complex,

Troll et al. NCOMMS-18-08734B 58 1147 Sweden. Mineral. Petrol. 92, 427–452 (2008).

1148 122. Larson, S. Å. & Magnusson, K.-Å. The magnetic and chemical character of Fe-Ti oxides in

1149 the Ulvö dolerite, central Sweden. Geological Survey of Sweden C 723, 29 p (1976).

1150 123. Carracedo, J. C. et al. Hotspot volcanism close to a passive continental margin: the Canary

1151 Islands. Geol. Mag. 135, 591–604 (1998).

1152 124. Ancochea, E. et al. Volcanic evolution of the island of Tenerife (Canary Islands) in the light

1153 of new K-Ar data. J. Volcanol. Geotherm. Res. 44, 231–249 (1990).

1154 125. Deegan, F. M. et al. Crustal versus source processes recorded in dykes from the Northeast

1155 volcanic rift zone of Tenerife, Canary Islands. Chem. Geol. 334, 324–344 (2012).

1156 126. Delcamp, A. et al. Dykes and structures of the NE rift of Tenerife, Canary Islands: A record

1157 of stabilisation and destabilisation of ocean island rift zones. Bull. Volcanol. 74, 963–980

1158 (2012).

1159 127. Gass, I. G. & Masson-Smith, D. The Geology and Gravity Anomalies of the Troodos Massif,

1160 Cyprus. Philosphical Trans. R. Soc. London 255, 417–467 (1963).

1161 128. Henson, F. R. S., Browne, R. V. & McGinty, J. A synopsis of the stratigraphy and geological

1162 history of Cyprus. Q. J. Geol. Soc. 105, 1–41 (1949).

1163 129. Allerton, S. & Vine, F. J. Spreading evolution of the Troodos ophiolite, Cyprus. Geology 19,

1164 637–640 (1991).

1165 130. Stillman, C. J. Some dyke-Gabbro relationships in the Troodos Ophiolite, Cyprus. In:

1166 Gibson, I. L. et al. (eds.) Cyprus Crustal Study Project: initial report, hole CY-4. Geological

1167 Survey of Canada, Ottawa, Ontario, Canada: 107–113 (1989).

1168 131. Trønnes, R. G. Geology and geodynamics of Iceland. Unpub. Ms. (2002).

1169 132. Wolfe, C. J., Th. Bjarnason, I., VanDecar, J. C. & Solomon, S. C. Seismic structure of the

1170 Iceland mantle plume. Nature 385, 245–247 (1997).

1171 133. Gudmundsson, M. T. et al. Eruptions of eyjafjallajökull Volcano, Iceland. in Eos 91, 191

1172 (2010).

Troll et al. NCOMMS-18-08734B 59 1173 134. Sæmundsson, K. Outline of the geology of Iceland. Jokull 29, 11–28 (1979).

1174 135. Dahren, B. et al. Magma plumbing beneath Anak Krakatau volcano, Indonesia: Evidence for

1175 multiple magma storage regions. Contrib. to Mineral. Petrol. 163, 631–651 (2012).

1176 136. Gardner, M. F. et al. Crustal differentiation processes at Krakatau Volcano, Indonesia. J.

1177 Petrol. 54, 150–182 (2013).

1178 137. Self, S. & Rampino, M. R. The 1963-1964 eruption of Agung volcano (Bali, Indonesia).

1179 Bull. Volcanol. 74, 1521–1536 (2012).

1180 138. Handley, H. K., Macpherson, C. G. & Davidson, J. P. Geochemical and Sr-O isotopic

1181 constraints on magmatic differentiation at Gede Volcanic Complex, West Java, Indonesia.

1182 Contrib. to Mineral. Petrol. 159, 885-908 (2010). doi:10.1007/s00410-009-0460-z

1183 139. Jeffery, A. J. et al. The pre-eruptive magma plumbing system of the 2007-2008 dome-

1184 forming eruption of Kelut volcano, East Java, Indonesia. Contrib. to Mineral. Petrol. 166,

1185 275-308 (2013). doi:10.1007/s00410-013-0875-4

1186 140. Chadwick, J. P., Troll, V. R., Waight, T. E., van der Zwan, F. M. & Schwarzkopf, L. M.

1187 Petrology and geochemistry of igneous inclusions in recent Merapi deposits: A window into

1188 the sub-volcanic plumbing system. Contrib. to Mineral. Petrol. 165, 259-282 (2013).

1189 doi:10.1007/s00410-012-0808-7

1190 141. Carlile, J. C. & Mitchell, A. H. G. Magmatic arcs and associated gold and copper

1191 mineralization in Indonesia. J. Geochemical Explor. 50, 91–142 (1994).

1192 142. Marcoux, E. & Milési, J. P. Epithermal gold deposits in West Java, Indonesia: geology, age

1193 and crustal source. J. Geochemical Explor. 50, 393–408 (1994).

1194 143. Van Bemmelen, R. W. (1949) The geology of Indonesia. Government Printing Office, The

1195 Hague, 732 p.

1196 144. Gamble, J. A., Price, R. C., Smith, I. E. M., McIntosh, W. C. & Dunbar, N. W. 40Ar/39Ar

1197 geochronology of magmatic activity, magma flux and hazards at Ruapehu volcano, Taupo

1198 Volcanic Zone, New Zealand. J. Volcanol. Geotherm. Res. 120, 271–287 (2003).

Troll et al. NCOMMS-18-08734B 60 1199 145. Jolly, A. D., Sherburn, S., Jousset, P. & Kilgour, G. Eruption source processes derived from

1200 seismic and acoustic observations of the 25 September 2007 Ruapehu eruption-North Island,

1201 New Zealand. J. Volcanol. Geotherm. Res. 191, 33–45 (2010).

1202 146. Price, R. C. et al. The anatomy of an andesite volcano: A time-stratigraphic study of andesite

1203 petrogenesis and crustal evolution at Ruapehu Volcano, New Zealand. J. Petrol. 53, 2139-

1204 2189 (2012). doi:10.1093/petrology/egs050

1205 147. Gamble, J. A. et al. A fifty year perspective of magmatic evolution on Ruapehu Volcano,

1206 New Zealand: Verification of open system behaviour in an arc volcano. Earth Planet. Sci.

1207 Lett. 170, 301–314 (1999).

1208 148. Stern, T. et al. Crust-mantle structure of the central North Island, New Zealand, based on

1209 seismological observations. J. Volcanol. Geotherm. Res. 190, 58–74 (2010).

1210 149. Price, R. C. et al. Crustal and mantle influences and U-Th-Ra disequilibrium in andesitic

1211 lavas of Ngauruhoe volcano, New Zealand. Chem. Geol. 277, 355–373 (2010).

1212 150. Graham, I. J. & Hackett, W. R. Petrology of calc-alkaline lavas from ruapehu volcano and

1213 related vents, taupo volcanic zone, New Zealand. J. Petrol. 28, 531–567 (1987).

1214 151. Price, R. C. et al. An integrated model for the temporal evolution of andesites and rhyolites

1215 and crustal development in New Zealand’s North Island. J. Volcanol. Geotherm. Res. 140,

1216 (2005).

1217 152. Geijer, P. & Magnusson, N. H. De mellansvenska järnmalmernas geologi. Geological Survey

1218 of Sweden Ca 35, 654 p (1944).

1219 153. Gavelin, S. Studier över berggrunden i Björnbergsfältet. Geol. Föreningen i Stock.

1220 Förhandlingar 55, 455–498 (1933).

1221 154. Lager, I. The geology of the Palaeoproterozoic limestone-hosted Dannemora iron deposit,

1222 Sweden. Sveriges Geologiska Undersökning Rapporter och meddelanden 107, 49 p (2001).

1223 155. Lager, I. The Dannemora iron ore deposit. In: Lundström, I. & Papunen, H. (eds.): Mineral

1224 deposits of southwestern Finland and the Bergslagen province, Sweden. 7th IAGOD

Troll et al. NCOMMS-18-08734B 61 1225 symposium and Nordkalott project meeting, excursion guide No. 3. Geological Survey of

1226 Sweden Ca 61, 26–30 (1986).

1227 156. Dahlin, P., Allen, R. & Sjöström, H. Palaeoproterozoic metavolcanic and metasedimentary

1228 succession hosting the Dannemora iron ore deposits, Bergslagen region, Sweden. GFF 134,

1229 71-85 (2012).

1230 157. Allen, R., Ripa, M. & Jansson, N. Palaeoproterozoic volcanic- and limestonehosted Zn-Pb-

1231 Ag-(Cu-Au) massive sulphide deposits and Fe oxide deposits in Bergslagen, Sweden. IGC

1232 Excursion Nr. 12 (August 2008): 16-21 (2008).

1233 158. Magnusson, N.H. The origin of the iron ores in central Sweden and the history of their

1234 alterations. Sveriges Geologiska Undersökning C 643, 127 p (1970).

1235 159. Lundh J. A lithogeochemical study of Northern Sweden and the Kiruna and Malmberget

1236 iron-apatite ore deposits. Master thesis, Uppsala University, 106 p (2014).

1237 160. Zhao, Z. F. & Zheng, Y. F. Calculation of oxygen isotope fractionation in magmatic rocks.

1238 Chem. Geol. 193, 59–80 (2003).

1239 161. Zheng, Y. Calculation of oxygen isotope fractionation in metal oxides. Chem. Geol. 55,

1240 2299–2307 (1991).

1241 162. Hoefs, J. Stable isotope geochemistry. Springer, Berlin, 286 p (1997).

1242 163. Jolis, E. M. (2013) Magma-Crust Interaction at Subduction Zone Volcanoes, Doctoral

1243 dissertation Acta Universitatis Upsaliensis, Uppsala, Sweden, 244 p.

1244 164. Dauphas, N. & Rouxel, O. Mass spectrometry and natural variations of iron isotopes. Mass

1245 Spectrometry Reviews 25, 515–520 (2006).

1246 165. Hou, Z. et al. Geology, fluid inclusions, and oxygen isotope geochemistry of the

1247 Baiyinchang pipe-style volcanic-hosted massive sulfide Cu deposit in Gansu Province,

1248 Northwestern China. in Econ. Geol. 103, 269-292 (2008). doi:10.2113/gsecongeo.103.1.269

1249 166. Rieger, A. A., Marschik, R. & Díaz, M. The evolution of the hydrothermal IOCG system in

1250 the Mantoverde district, northern Chile: New evidence from microthermometry and stable

Troll et al. NCOMMS-18-08734B 62 1251 isotope geochemistry. Miner. Depos. 47, 359-369 (2012). doi:10.1007/s00126-011-0390-3

1252 167. Beard, B. L. et al. Application of Fe isotopes to tracing the geochemical and biological

1253 cycling of Fe. Chem. Geol. 195, 87-117 (2003). doi:10.1016/S0009-2541(02)00390-X

1254 168. Beard, B. L. & Johnson, C. M. Fe Isotope Variations in the Modern and Ancient Earth and

1255 Other Planetary Bodies. Rev. Mineral. Geochemistry 55, 319–357 (2004).

Troll et al. NCOMMS-18-08734B 63