The Age of the Tseax Volcanic Eruption, British Columbia, Canada Glyn Williams-Jones, René W

Home , Effusive eruption, Lapilli, Lava, Lava Fork, Nass River, Shield volcano

Pagination not ﬁnal (cite DOI) / Pagination provisoire (citer le DOI)

1 ARTICLE

The age of the Tseax volcanic eruption, British Columbia, Canada Glyn Williams-Jones, René W. Barendregt, James K. Russell, Yannick Le Moigne, Randolph J. Enkin, and Rose Gallo

Abstract: A recent volcanic eruption occurred at Tseax volcano that formed a series of tephra cones in northwestern British Columbia, Canada. The explosive to effusive eruption also formed a 32 km long sequence of Fe-rich Mg-poor basanite–trachybasalt lavas covering ϳ40 km2. Oral histories of the Nisg_a’a Nation report that the eruption may have caused as many as 2000 fatalities. The actual eruption date and question of whether there was one or multiple eruptive episodes in the 14th and 18th centuries are, as of yet, unresolved. New radiocarbon dating of wood charcoal from immediately beneath vent-proximal tephra deposits and complementary age information suggest an eruption in 1675–1778 CE (95.4% probability) was responsible for the formation of the tephra cone. New paleomagnetic and geochemical data from the tephra cone and lava ﬂows suggest there is, in fact, no statistically signiﬁcant difference in time between the explosive and effusive deposits and that they formed during a single eruptive episode.

Key words: Tseax volcano, lava ﬂow, tephra cone, paleomagnetism, radiocarbon dating, geochemistry. Résumé : Une éruption volcanique récente a eu lieu au volcan Tseax, qui a produit une série de cônes de téphra dans le nord-ouest de la Colombie-Britannique (Canada). L’éruption explosive à effusive a également produit une séquence longue de 32 km de laves de basanite–trachybasalte riches en Fe et pauvres en Mg couvrant ϳ40 km2. Des récits oraux de la Nation Nisg_a’a mentionnent que l’éruption pourrait avoir causé jusqu’à 2000 morts. Les questions du moment exact de l’éruption et à savoir si elle comportait un seul ou plusieurs épisodes éruptifs aux 14e et 18e siècles demeurent à ce jour sans réponses. De nouveaux résultats de datation au carbone radioactif de charbon de bois prélevé immédiatement sous des dépôts de téphra proximaux à la cheminée et de l’information temporelle complémentaire indiqueraient qu’une éruption vers 1675–1778 EC (probabilité de 95,4 %) est à l’origine de la formation du cône de téphra. De nouvelles données paléomagnétiques et géochimiques sur le cône de téphra et les coulées de lave donnent à penser qu’il n’y a en fait aucune différence statistiquement signiﬁcative dans le temps entre les dépôts explosifs et effusifs et qu’ils se sont formés durant un seul épisode éruptif. [Traduit par la Rédaction]

Mots-clés : volcan Tseax, coulée de lave, cône de téphra, paléomagnétisme, datation au carbone radioactif, géochimie.

For personal use only. Introduction Le Moigne et al. 2020), petrological and geochemical studies Tseax volcano situated in northwestern British Columbia, Canada (Nicholls et al. 1982, 1997; Higgins 2009; Gallo 2018), and dating (55.11085°N, 128.89944°W), comprises several small tephra cones (Lowdon et al. 1971; Symons 1975; Wuorinen 1978; Higgins 2009). and a 32 km long basanite–trachybasalt lava (e.g., Hanson 1923; The Tseax vent area comprises a number of short eruptive fis- Sutherland Brown 1969). The eruption is believed to have occurred in sures and two small tephra cones; the larger of the two cones is the 1700s (e.g., Lowdon et al. 1971; Higgins 2009) and to have de- partially enclosed by a spatter rampart described previously as a dissected tephra cone (Sutherland Brown 1969; Wuorinen 1978). stroyed at least three Nisg_a’a Nation villages (Laxksiluux, Laxksi- wihlg_est, and Ts’oohlts’ap; Fig. 1) located on the banks of the Nass River, This original description, along with radiocarbon dating of a tree ϳ20 km from the volcano. The eruption may have caused as many as from the spatter rampart, raised the possibility of two distinct 2000 fatalities (Nisg_a’a Nation 2004). These fatalities would make it eruptive episodes occurring in 1325 CE and 1700 CE (Wuorinen Canada’s second-worst recorded natural disaster (Hickson and 1978). Recent field work has established the relative sequence of Edwards 2001) after the near contemporaneous Newfoundland volcanic events (Le Moigne et al. 2018, 2020), but the published hurricane of 1775 that caused at least 4100 deaths (Ruffman 1996). radiocarbon dating is sparse (three samples), leaving ambiguity There are no direct written accounts of the Tseax event; however, concerning the absolute timing and duration of eruptive activity. the rich oral history (adaawak - traditional histories) from the Importantly, the question of whether the volcanic field represents Nisg_a’a people provides important observational data for the a monogenetic or a polygenetic system (e.g., Németh and

Can. J. Earth Sci. Downloaded from www.nrcresearchpress.com by Dr. Glyn Williams-Jones on 06/29/20 eruption (Nisg_a’a Nation 2004; see Appendix A). Previous studies Kereszturi 2015), which may have had more than one eruption, have included volcanological and geomorphological mapping has important implications for future activity and hazard mitiga- (Hanson 1923; Sutherland Brown 1969; Roberts and McCuaig 2001; tion efforts and thus merits further study.

Received 18 December 2019. Accepted 25 March 2020. G. Williams-Jones. Centre for Natural Hazards Research, Department of Earth Sciences, Simon Fraser University, Burnaby, BC V5A 1S6, Canada. R.W. Barendregt. Department of Geography, University of Lethbridge, Lethbridge, AB T1K 3M4, Canada. J.K. Russell and R. Gallo. Department of Earth, Ocean and Atmospheric Sciences, The University of British Columbia, Vancouver, BC V6T 1Z4, Canada. Y. Le Moigne. Centre for Natural Hazards Research, Department of Earth Sciences, Simon Fraser University, Burnaby, BC V5A 1S6, Canada; Laboratoire Magmas et Volcans, Université Clermont Auvergne, 63178 Aubière, France. R.J. Enkin. Paleomagnetism and Petrophysics Laboratory, Geological Survey of Canada - Paciﬁc, Sidney, BC V8L 4B2, Canada. Corresponding author: Glyn Williams-Jones (email: [email protected]). Copyright remains with the author(s) or their institution(s) and © Her Majesty the Queen in right of Canada 2020. Permission for reuse (free in most cases) can be obtained from copyright.com.

Can. J. Earth Sci. 00: 1–16 (0000) dx.doi.org/10.1139/cjes-2019-0240 Published at www.nrcresearchpress.com/cjes on 30 March 2020. Pagination not ﬁnal (cite DOI) / Pagination provisoire (citer le DOI)

2 Can. J. Earth Sci. Vol. 00, 0000

Fig. 1. Map showing Tseax cone (red circle) and four valley-filling lava flows; two early stage phenocryst-poor pahoehoe (light and dark green) and two later stage phenocryst-rich àà lavas (yellow and orange). RBn (white diamonds) denote locations of five new palaeomagnetism sites and black circles are sites from Symons (1975). Orange hexagon indicates dendrochronology study site (Hanson 1923) and green triangles show locations of radiocarbon sample sites. Light orange squares and circles show locations of pre-eruption Nisg_a’a villages (V1, Laxksiluux; V2, Laxksiwihlgest; V3, Ts’oohlts’ap) and fish camps (F1) / smokehouses (F2), respectively. Gennu Axwt (purple triangle) is a ridge from which scouts were said to have observed the eruption of Tseax (Nisg_a’a Nation 2004). Modified from Le Moigne et al. (2020). See Table 1 and Appendix Table A2 for sample locations. Inset map shows Tseax (red star) in the context of Neogene–Quaternary volcanic centres and complexes in the Canadian Cordillera, including the Garibaldi Volcanic Belt (GVB), the Wells Gray – Clearwater volcanic field (WGC), the Anaheim Volcanic Belt (AVB), the Wrangell Volcanic Belt (WVB), and the Northern Cordilleran Volcanic Province (NCVP). Modified after Edwards and Russell (2000). Map produced using ESRI ArcGIS 10.7. [Colour online.] For personal use only.

Here we present new geochemical, paleomagnetic, and radio- Tseax volcano carbon data for samples collected from the Tseax volcanic depos- Tseax volcano, Wil Ksi Baxhl Mihl, formerly known as Aiyansh its (Figs. 1, 2) supported by detailed mapping of the tephra cones volcano, Aiyansh River volcano, or Tseax River cone, is located in Can. J. Earth Sci. Downloaded from www.nrcresearchpress.com by Dr. Glyn Williams-Jones on 06/29/20 and lava flows (Gallo 2018; Le Moigne et al. 2018, 2020). The new the Anhluu’t’ukwsim Laxmihl Angwiga’asanskwhl Nisg_a’a (Nisg_a’a Me- radiocarbon dates derive from samples of charred wood collected morial Lava Bed Provincial Park) near Gitlaxt’aamiks (formerly New from beneath explosive tephra deposits ϳ890 m northwest of the Aiyansh) and Gitwinkshilkw (formerly Canyon City), ϳ60 km north- main tephra cone (Fig. 2). The paleomagnetic data sets derive from west of Terrace in northwestern British Columbia, Canada (Fig. 1). the lava field, tephra cone, and spatter rampart. Our sampling Tseax (pronounced see-ax) is the southernmost volcanic centre of campaign was specifically designed to test whether all the Tseax the Northern Cordillera Volcanic Province (Edwards and Russell volcanic deposits, representing explosive and effusive activity, 2000) and is notable for a 32 km long basanite–trachybasalt lava share a common paleomagnetic direction acquired during cool- flow (approximately 0.5 km3 covering ϳ36 km2; Fig. 1). More im- ing. The implication would be that all events have the same pa- portantly, it is the site of the second-youngest (after Lava Fork in leomagnetic age. The alternative is that the apparently earlier ϳ1800 CE; Elliott et al. 1981) and the deadliest volcanic eruption in stratigraphic unit (i.e., the partially dissected spatter rampart; Canada (Hickson and Edwards 2001). The tephra cones and valley- Fig. 2) has a different paleomagnetic direction and age than the filling lavas overlay intercalated sandstone to siltstone and mud- stratigraphically youngest units (i.e., tephra cone and the valley- stone from the Late Jurassic Bowser Lake group (van der Heyden filling lavas), as suggested from 14C dating by Wuorinen (1978). et al. 2000; Evenchick et al. 2008).

Published by NRC Research Press Pagination not ﬁnal (cite DOI) / Pagination provisoire (citer le DOI)

Williams-Jones et al. 3

Fig. 2. Volcanological map of Tseax cone and spatter rampart after Le Moigne et al. (2018, 2020). Basement geology from Evenchick et al. (2008). Inset images: Tseax crater (top right) and inner wall of the spatter rampart (bottom right). RBn (white diamonds) indicate locations of new palaeomagnetism sites and green triangles show locations of new and published radiocarbon samples. Contour interval is 50 m. Map produced using ESRI ArcGIS 10.7. [Colour online.] For personal use only.

Table 1. Published whole rock compositions from explosive and effusive deposits at Tseax. Sample NR5-C* RG-B7† RG-B2† TS-S40a† TS-S19† TS-S20† TS-S52† TS-S57† RG-S16† RG-S25† Lithology Lava flow Bomb Bomb Satellite cone Spatter Spatter Lava flow Lava flow Tephra Tephra

SiO2 46.53 46.4 46.74 45.20 46.85 46.54 46.85 46.86 46.58 46.04 TiO2 3.72 3.64 3.56 3.46 3.59 3.68 3.67 3.58 3.67 3.66 Al2O3 14.40 14.46 14.62 14.25 14.57 14.54 14.64 14.68 14.6 14.36 Fe2O3 2.27 3.56 2.64 1.67 ——— — 2.58 2.31 FeO 12.76 12.15 12.8 14.98 14.78 14.69 14.74 14.63 12.9 13.3 MnO 0.29 0.23 0.23 0.22 0.22 0.22 0.22 0.22 0.22 0.22 MgO 4.52 4.43 4.39 4.24 4.56 4.65 4.64 4.51 4.4 4.38

Can. J. Earth Sci. Downloaded from www.nrcresearchpress.com by Dr. Glyn Williams-Jones on 06/29/20 CaO 7.53 7.45 7.45 6.67 7.42 7.44 7.51 7.45 7.37 7.34 Na2O 4.12 3.82 3.85 4.05 3.95 4.02 4.03 3.9 3.82 3.81 K2O 1.87 1.76 1.81 1.86 1.83 1.78 1.81 1.8 1.76 1.75 P2O5 1.15 1.13 1.2 1.12 1.18 1.12 1.16 1.18 1.17 1.12 CO2 — 0.11 0.11 —————0.84 0.51 SO3 — 0.13 0.15 — 0.14 0.06 0.14 0.09 0.21 0.22 S ———— —————— − H20 0.07 ——— — — — — — — + H20 0.16 ——— — — — — — — Total 99.39 99.27 99.55 97.72 99.09 98.74 99.41 98.90 100.13 99.02 LOI — −1.11 −1.34 0.74 −1.00 −0.37 −1.38 −0.93 −0.82 −0.98 FeOT 15.28 15.35 15.18 18.32 14.78 14.69 14.74 14.63 15.22 15.38 Mg # 37.78 37.88 36.88 32.99 35.49 36.08 35.95 35.46 36.78 36.10 Note: See Appendix Table A1 for sample locations. LOI, loss on ignition. *Data from Nicholls et al. (1982). †Data from Gallo (2018).

Published by NRC Research Press Pagination not ﬁnal (cite DOI) / Pagination provisoire (citer le DOI)

4 Can. J. Earth Sci. Vol. 00, 0000

Fig. 3. Whole rock and trace element geochemistry for select samples from effusive and explosive deposits at Tseax. (A) Total alkali vs SiO2 (see Table 1). (B) Trace element plot showing rare earth element (REE) concentrations normalized to primitive mantle (Sun and McDonough 1989). Trace element data from Gallo (2018). Green symbols/lines represent undifferentiated tephra, black symbol/lines are undifferentiated lava. [Colour online.]

Detailed volcanological mapping shows Tseax volcano to com- trace element geochemistry of select samples compiled from the prise a ϳ70 m high, 400 m diameter tephra cone situated within literature are essentially identical across lava, tephra and ballis- an oxidized horseshoe-shaped spatter rampart (Fig. 2)(Hanson tics, the tephra and satellite cones, and spatter rampart (Fig. 3). 1923; Sutherland Brown 1969; Wuorinen 1978; Le Moigne et al. This strongly suggests that there has been no signiﬁcant residence 2020). Another smaller (ϳ20 m high, ϳ50–55 m diameter) highly time, or associated fractional crystallisation, within the magmatic oxidized tephra cone, unnamed and thus referred to here as Sat- plumbing system (e.g., Hawkesworth et al. 2000). All samples plot ellite cone, is located 470 m to the north of Tseax and in close as Fe-rich Mg-poor basanite–trachybasalt (Fig. 3) with a mean ␴ proximity to an eruptive ﬁssure consisting of small tephra chemical composition (±1 ): SiO2 of 46.5% ± 0.5%, Na2O+K2Oof mounds (Fig. 2; Le Moigne et al. 2020). These structures are sur- 5.74% ± 0.1%, FeOT of 15.5% ± 1.1%, and Mg# of 36.1 ± 1.4 (Table 1; rounded by a broad tephra deposit covering ϳ22 km2 (Fountain Nicholls et al. 1982; Higgins 2009; Gallo 2018). tephra in Fig. 2, Table 1; Gallo 2018; Le Moigne et al. 2020). Four

For personal use only. valley-filling lavas (two pahoehoe and two àà flows) have been The age controversy mapped and, due to their uniform composition, are believed to The exact date of the eruption of Tseax remains unclear despite have been emplaced during a single eruptive episode (Fig. 1; the number of previous studies. Early reports by missionaries of Hanson 1923; Sutherland Brown 1969; Higgins 2009; Le Moigne local Nisg_a’a adaawak (e.g., Collinson 1915; McCullagh 1918) men- et al. 2020). tion trees cut in 1898 at a place where the Nisg_a’a had established Two porphyritic pahoehoe lavas (10%–15% phenocrysts of pla- their first refuge after evacuating their villages seven generations gioclase, olivine, and oxides) initially flowed 5 km down Crater earlier (i.e.,7×21years); the trunks are reported to have shown Creek before blocking the Tseax River and forming Lava Lake evidence of bark being stripped 128 years earlier, in ca. 1770 CE, (Fig. 1, Hanson 1923; Roberts and McCuaig 2001). The lavas flowed immediately after the eruption (Roberts and McCuaig 2001; a further 15 km northward down the Tseax River valley to the Nass Higgins 2009). However, there is likely significant uncertainty in River floodplain where the first flow spread out to form an approx- these estimates due to poor translation from Nisg_a’a to English imately 12 km long lava plain that parallels the Nass River (Fig. 1); (e.g., Higgins 2009). Hanson (1923, p. 39) reports that “trees one the second flow halted at the entrance to the Nass Valley. The hundred and seventy years old have been found growing on the pahoehoe lavas thicken to at least 12 m along the northern margin lava”, which would constrain the date of the eruption to before with the Nass River (Symons 1975; Purssell 1993) and results in 1753 CE (Fig. 4). From his work in the Nass Valley, Barbeau (1935) numerous basaltic pillows and pillow tubes (Le Moigne et al. concluded that the most recent eruption occurred in the late 2020). Two later, stratigraphically distinct, and smaller volume 18th century. Similarly, Wuorinen (1976) reports that the Nisg_a’a Can. J. Earth Sci. Downloaded from www.nrcresearchpress.com by Dr. Glyn Williams-Jones on 06/29/20 àà lavas flowed from the source area down Crater Creek but adaawak (see Appendix A) dates the eruption back to 1770 CE and stopped near Lava Lake (Fig. 1). All flows are characterized by states that it occurred over a period of a few days or at most a few numerous intact and collapsed lava tubes, both proximal and weeks. distal to the source (Sutherland Brown 1969; Marshall 1975; Le Moigne et al. 2020). Published radiometric studies Plagioclase megacrysts (up to 20 mm) are present in the four A number of radiocarbon-dating studies were made on trees lava flows (pahoehoe and àà) but represent less than 1% of killed by the eruption. Sutherland Brown (1969) and Souther the volume. There is nevertheless a notable difference in the crys- (1970) reported a date of 220 ± 130 14C years BP for a lava-encased tal size distribution with the two pahoehoe flows containing cottonwood sampled near the Nass River; however, Lowdon et al. less <8%–10% plagioclase phenocrysts, whereas the two àà flows (1971) state that this date was uncorrected and should in fact be have ϳ15%–20% plagioclase phenocrysts (Higgins 2009; Le Moigne 250 ± 130 14C years BP (sample GSC-1124, Table 2, Fig. 4). More et al. 2018). Higgins (2009) proposed that this indicated an erup- recently, Roberts and McCuaig (2001) dated a wood fragment of a tion sourced from two separate batches of magma. Interestingly, lava-encased tree at 280 ± 50 14C years BP (sample GSC-6150); as it in spite of the textural differences of the lavas, the whole rock and was missing ϳ8 cm of the trunk (with 5 rings/cm) and several rings

Published by NRC Research Press Pagination not ﬁnal (cite DOI) / Pagination provisoire (citer le DOI)

Williams-Jones et al. 5

Fig. 4. Published and new eruption dates from oral history (Collinson 1915; McCullagh 1918; Barbeau 1935), dendrochronology (Hanson 1923), paleomagnetism (Symons 1975; Higgins 2009), and radiocarbon studies. New and recalibrated radiocarbon dates (within cyan border) are presented as a calibrated probability age distribution (prior likelihood, i.e., each measurement taken separately, in light grey, and combined or posterior likelihood in dark grey). Both 68.2% and 95.4% probability confidence intervals are shown (black bars). Combined reflects the likelihood curve for an event postdating all the radiocarbon dates and predating 1778 CE based on the complementary chronological studies. Plots generated from OxCal version 4.3.2 (Ramsey 2009). Uncertainties are shown by error bars for paleomagnetic dates. No uncertainties are shown for oral or dendrochronology studies. See Table 2 and Appendix Table A1 and Fig. A1. [Colour online.] For personal use only. Table 2. Compilation of radiocarbon data. Posterior calibrated date range Radiocarbon date Source Sample ID Sample type 14C year (BP) (68.2 % prob., CE) (95.4 % prob., CE) Sutherland Brown (1969); GSC-1124 Standing cottonwood log encased in lava 250 ± 130 1488–1683 (68.2%) 1435–1769 (95.4%) Souther (1970); near Nass River Lowdon et al. (1971) Wuorinen (1978) S-1046 Carbonised log in “outer cone” 625 ± 70 1339–1420 (68.2%) 1283–1436 (95.4%) Roberts and McCuaig (2001) GSC-6150 Lava-encased tree trunk in growth position 230 ± 50 1564–1642 (48.9%) 1526–1720 (95.4%) + N(50 ± 10) 1669–1704 (19.3%) This study CUIAMS-215254 Charred wood beneath tephra near cone 390 ± 15 1451–1485 (68.2%) 1446–1508 (86.2%) 1601–1616 (9.2%) This study CUIAMS-215255 Charred branch wood beneath tephra near 190±15 1662–1683 (68.2%) 1658–1685 (72.1%) cone 1735–1785 (23.3%) Combined Tseax eruption Five 14C dates Event postdating 14C and predating 1800 CE 1678–1710 (23.2%) 1674–1800 (95.4%) Can. J. Earth Sci. Downloaded from www.nrcresearchpress.com by Dr. Glyn Williams-Jones on 06/29/20 1749–1799 (45.0%) Note: Dates are presented as 14C year BP and calibrated calendar ranges (with OxCal + IntCal13 curve) with confidence intervals for 68.2% and 95.4% probability. The age model uses the complementary information constraints which suggest that all five samples predate a single volcanic event that occurred before 1800 CE. Boldface type indicates dates with highest confidence intervals. See Appendix A for OxCal script and Appendix Fig. A1 and Table A1.

were pulverized for the bulk date, they give a corrected age of and dated it at 625 ± 70 14C years BP, which calibrates to between 230±5014C years BP. Higgins (2009) recalibrated the radiocarbon 1283 and 1436 CE (95.4% probability; sample S-1046) (Figs. 2 and 4, dates from the most recent of the three samples (GSC-6150) using Table 2); we were unable to locate this tree trunk during our the CALIB software (Reimer et al. 2004; Stuiver and Reimer 1993) extensive mapping in 2016 and 2017. Wuorinen (1978, p. 1037) also and reinterpreted the age of the youngest eruption at between notes a difference in appearance between the “well-eroded sur- 1668 and 1714 CE (68.2% probability). face of the old [spatter rampart] cone and the comparatively pris- Importantly, Wuorinen (1978) sampled part of a carbonized tree tine surface and form of the new [inner] cone”. Along with the trunk found standing in the vertical wall of the spatter rampart “minimal vegetative growth (lichen and small bushes)” on the

Published by NRC Research Press Pagination not ﬁnal (cite DOI) / Pagination provisoire (citer le DOI)

6 Can. J. Earth Sci. Vol. 00, 0000

Fig. 5. Stereographic (Wulff, equal angle) plot of magnetic remanence cone in contrast to the spatter rampart that hosts mature trees, directions. (A) Plot of five new sites (RB1–5) sampled at Tseax and their Wuorinen (1978) suggested this raised doubt on the hypothesis of ␣ 95 (p = 0.05) confidence circles (see Table 3). Site RB4 is an outlier a single eruption. affected by post-cooling collapse of a lava tube and is not included in the mean (blue circle). (B) Mean of RB1, RB2, RB3, RB5, and the 23 sites Published paleomagnetic studies ␣ sampled by Symons (1975) and their 95 confidence circle (blue; Table 3). The orientation of the local geomagnetic field can be preserved The star marks the time-average geocentric axial dipole direction at in the remanent magnetization of volcanic rocks when they cool

the Tseax latitude (55.11085°N). (C) Quantile–quantile plot assessing below the Curie temperature (TC) of their principal ferromagnetic goodness-of-fit of the Fisher distribution to the paleomagnetic site mineral (typically magnetite, TC = 580 °C). The geomagnetic field mean directions. Site RB4 is an outlier, while all the other new sites direction changes rapidly through time and can be compared (black circles) are compatible with and span the range defined by the with local and global reference curves (e.g., Turner 1987; Korte Symons (1975) study (open circles), despite sampling different phases of et al. 2009) to constrain emplacement/deposition ages of lava the eruption and 35 years difference in technology and methods. flows and pyroclastic deposits. This approach has been success- [Colour online.] fully applied to a number of current and geologically recent eruptions (e.g., Vulcano, Italy, Lanza and Zanella 2003; Tongariro, New Zealand, Greve et al. 2016; El Metate, Mexico, Mahgoub et al. 2017; Colli Albani, Italy, Trolese et al. 2017; Reykjanes Peninsula, Ice- land, Pinton et al. 2018). Symons (1975) completed an extensive paleomagnetic study of Tseax based on 197 specimens from 122 cores (ϳ20 cm long) at 23 sites on the early stage, phenocryst-poor, pahoehoe lava flows. It is important to note that no samples were collected from the later stage phenocryst-rich à à lavas, the tephra cone, or the spatter rampart (Figs. 1, 2). In situ orientation (to within ϳ1°) was accomplished using a solar compass and Brunton (for topograph- ical control). Symons (1975) determined a mean magnetic remanence direction of 13.9° declination and 72.9° inclination with an ␣ 95 of 0.8° (Fig. 5B). Symons (1975) noted that the natural remanent magnetizations of the pahoehoe lavas were strong and remarkably stable, falling within 7.6–24.8 A/m; variations were ascribed to variations in lava vesicularity. Observations of the declination of the magnetic field for British Columbia were initially made in 1778 and 1792 CE by Captain Cook and Captain Vancouver, respectively (Herbert 1926). Between 5000 and 1500 BP, observations suggest westward drift of the non- dipole field at a rate of about ϳ0.075°/year in western Canada (Yukutake 1967; Turner 1987). Based on this, Symons (1975) extrap- For personal use only. olated the observatory data linearly back from 1770 CE; the pro- jected declination intersected the measured flow declination at 1650±40CE(Fig. 4). Higgins (2009) applied a number of regional and global models of geomagnetic secular variation (GUFM1, Jackson et al. 2000; a dipole model, Hagstrum and Champion 2002; CALS3k.2 and CALS7k.2, Korte and Constable 2005). How- ever, while the predicted declinations were close to the observed values, inclinations were always overestimated and likely due to limitations of the databases used to define the geomagnetic secular variation models (fig. 3 in Higgins 2009). Sampling and methods Radiocarbon sampling and analysis Two samples of wood charcoal were collected from a carbonized horizon immediately below the tephra deposit in the imme- diate vicinity of Tseax cone (Fig. 2, Appendix Table A1). Sample

Can. J. Earth Sci. Downloaded from www.nrcresearchpress.com by Dr. Glyn Williams-Jones on 06/29/20 CUIAMS-215254 was a piece of charred wood collected at a depth of ϳ12–15 cm (under ϳ6 cm of modern surface soil overlying ϳ6 cm of lapilli). Sample CUIAMS-215255 consisted of charcoal fragments of partially charred branch wood collected at a depth of ϳ8–14 cm (under ϳ8 cm of modern surface soil overlying ϳ6cm of lapilli). These samples were processed by Alice Telka, Paleotech Services (Ontario, Canada); CUIAMS-215254 was cleaned of inter- nal and surface ﬁne roots and rhizomes and CUIAMS-215255 had minor amounts of surface material shaved off. Suitable fragments of wood charcoal were subsequently submitted to the University of California Irvine Keck Carbon Cycle AMS Facility for radiocarbon analysis. Dates are reported as conventional radiocarbon ages corrected for fractionation with measured ␦13C according to Stuiver and Polach (1977). Calendar ages are presented as 68.3%

Published by NRC Research Press Pagination not ﬁnal (cite DOI) / Pagination provisoire (citer le DOI)

Williams-Jones et al. 7

Table 3. Paleomagnetic analyses. ␣ Sample site Nt Lithology Ns Longitude (°W) Latitude (°N) Decl. Incl. k 95 RB1 LLM001–LLM012 Coherent lava 10 129.20435 55.18505 3.5 72.9 379 2.5 RB2 LLM013–LLM022 Large pyroclastic blocks in tephra 13 128.89856 55.11148 20.1 72.3 221 2.8 RB3 LLM023–LLM036 Inner rampart wall and spatter 14 128.89600 55.11108 8.9 77.8 248 2.5 RB3s LLM023–LLM025 Spatter only 3 128.89600 55.11108 6.0 77.6 490 5.6 RB3w LLM026–LLM036 Inner rampart wall only 11 128.89600 55.11108 9.7 77.9 208 3.2 RB4 LLM037–LLM046 Coherent lava in collapsed tube 13 128.90291 55.11151 50.0 77.9 482 1.9 RB5 LLM047–LLM054 Coherent lava 9 129.22176 55.18524 13.3 73.4 403 2.6 Mean (RB1–RB5) 5 17.6 75.4 247 4.9 Mean (excl. RB4) 4 11.7 74.2 641 3.6 Symons (1975) - approximate locations S1 Coherent lava 5 129.24138 55.16590 11.5 75.5 1830 1.5 S2 Coherent lava 5 129.22400 55.17255 11.6 74.4 1978 1.4 S3 Coherent lava 5 129.22280 55.18073 7.3 73.1 2619 1.2 S4 Coherent lava 5 129.22573 55.18430 24.1 74.9 936 2.1 S5A Coherent lava 2 129.22029 55.19103 12.4 73.9 82 10.9 *S5B Coherent lava 3 129.22029 55.19103 139.8 30.0 45 12.0 S6 Coherent lava 5 129.21349 55.18586 12.2 73.8 373 3.3 S7 Coherent lava 5 129.20130 55.18695 10.5 71.7 701 2.4 S8 Coherent lava 5 129.18158 55.19260 17.5 73.8 938 2.0 S9 Coherent lava 5 129.17184 55.19538 22.3 73.8 1860 1.5 S10 Coherent lava 5 129.16180 55.19853 11.3 73.2 914 2.1 S11 Coherent lava 6 129.15050 55.20330 14.0 73.2 355 3.0 S12 Coherent lava 5 129.13457 55.21478 9.4 74.9 423 3.1 S13 Coherent lava 6 129.13229 55.21037 0.9 73.1 640 2.3 S14 Coherent lava 6 129.10684 55.21097 16.0 72.4 428 2.8 S15 Coherent lava 5 129.11953 55.21106 22.5 71.8 811 2.2 S16 Coherent lava 5 129.08596 55.18999 13.8 71.1 794 2.2 S17 Coherent lava 6 129.07608 55.18634 9.5 72.0 2004 1.3 S18 Coherent lava 6 129.05995 55.17893 7.3 72.0 900 1.9 S19 Coherent lava 6 128.99590 55.14444 16.3 69.4 502 2.6 S20 Coherent lava 6 128.98981 55.14079 12.9 71.8 457 2.7 S21 Coherent lava 5 128.98546 55.12878 18.1 71.6 598 2.6 S22 Coherent lava 6 128.97814 55.11758 18.2 73.4 880 1.9 S23 Coherent lava 6 128.97806 55.11533 19.4 72.0 478 2.6 Mean (*excluding S5B) 23 13.9 72.9 1372 0.8 Grand mean (RB1, RB2, RB3, RB5, and Symons (1975) 27 13.5 73.1 1296 0.7 For personal use only. Note: Nt, sample numbers for each site; Ns, total number of samples used in calculation of mean direction for each site; Decl., declination; Incl., inclination; k, ␣ precision parameter; 95, radius of 95% conﬁdence level. See Figs. 1 and 2 for location.

and 95.4% probability, respectively, and calibrated using OxCal using an ASC Scientific D-2000 demagnetizer with a three-axis version 4.3.2 (Ramsey 2009) and the IntCal13 atmospheric curve manual tumbler and carried out at 10 millitesla (mT) steps (up to (Ramsey et al. 2013)(Table 2). 200 mT). Thermal demagnetization was carried out at 100, 200, 300, 400, 500, 525, and 550 °C, using an ASC Model TD48 dual- Paleomagnetic sampling and analysis chamber thermal demagnetizer to confirm that alternating field Oriented cylindrical samples were collected from coherent lava demagnetization was sufficient to resolve the primary rema- and large intact blocks in pyroclastic deposits at the Tseax cone nence. Directions of characteristic remanent magnetization were and nearby sites (Figs. 1, 2, Table 3). Cylindrical cores 2.5 cm in determined for each sample by principal component analysis diameter were extracted using a portable gasoline-powered rock (Kirschvink 1980) using Remasoft version 3.0 (Chadima and Hrouda drill and water swivel attachment, commonly used in standard 2006). Mean characteristic remanent magnetization directions were paleomagnetic work (Fig. 6). Ten to fourteen cores were collected calculated for each site and an overall mean was also calculated at each site, usually spread out over several metres to minimize (Table 3). the chance of lightning or localized disturbances affecting results. Five sites were sampled on the valley-filling lava, tephra cone, and Results and implications Can. J. Earth Sci. Downloaded from www.nrcresearchpress.com by Dr. Glyn Williams-Jones on 06/29/20 outer spatter rampart, providing 54 independently oriented cores (Fig. 1, Table 3). All samples were oriented using a solar compass as Radiocarbon results the Brunton magnetic compass displayed large deflections due to To provide some insight on the age of the eruption and for the significant magnetic field created by the outcrop. consistency, we recalibrated the published radiocarbon age esti- Magnetic measurements were made in the paleomagnetic labora- mates (Fig. 4, Table 2). Note that for Robert and McCuaig’s (2001) tory at the University of Lethbridge, Alberta. Magnetic susceptibility sample (GSC-6150), we corrected for the missing ϳ50 years (due to was measured with a Sapphire Instruments (SI-2B) susceptibility me- missing trunk material) by adding a Gaussian likelihood of 50 ± ter. The magnetization of each sample was measured with an AGICO 10 years after calibration, rather than to the C14 age; this results in JR-6A spinner magnetometer prior to demagnetization and again a similar likelihood function but with less likelihood of a modern after each level of stepwise demagnetization. Samples were stored in age. It is important to note that these wood charcoal fragments magnetic shields following field collection and between laboratory were not all from in situ trunks and are considered to be the measurements. Most samples were subjected to alternating field de- maximum ages. magnetization with one-third of the collection subjected to thermal Based on our analysis of the mapping, geochemistry, and the demagnetization. Alternating field demagnetization was performed paleomagnetism, we interpret that there was a single eruptive

Published by NRC Research Press Pagination not ﬁnal (cite DOI) / Pagination provisoire (citer le DOI)

8 Can. J. Earth Sci. Vol. 00, 0000

Fig. 6. Images and stereographic plots (inset) from site RB3 of the inner eastern wall of the outer spatter rampart (A), with closeup of sampled ␣ agglutinated spatter (B), and inner wall (C). Stereographic plots show individual measurements (in black) and means (in red) and 95 conﬁdence circles. See Fig. 1 for location. [Colour online.]

event that postdated all the charred wood measured in the 14C revealed a signiﬁcantly more easterly declination (40° east of the samples. We combined the ﬁve dates as a “phase” using the OxCal mean declination of the other four sites). Re-examination of the

For personal use only. age analysis (Ramsey 2009) and set the age of the Tseax eruption to field site in 2018 showed that the original sample was from a very postdate that phase. See Appendix A for OxCal script. The oral large block on the partially collapsed and rotated wall of an ex- history and dendrochronology suggest that the eruption occurred posed lava tube and this likely accounts for the deviated direction by 1780 CE and the paleomagnetism declination constrains the for this site. The offset is explained with a 10° dip down to the age to predate James Cook’s first direct observation in 1778. Thus, northwest (i.e., around strike 238°). The mean direction of our we constrain the lower bounding age at 1778 CE. The OxCal Bayes- four sites (excluding site RB4) is 11.7° declination, 74.2° inclination ian age modelling of the five 14C ages defines a relatively constant (Fig. 5A, Table 3), which is only 0.9° ± 2.3° from the mean (13.9°, likelihood between 1675 and 1778 CE (95.4%) (Fig. 4). 72.9°) calculated from 23 sites on the early stage lavas reported by Symons (1975). The new and published sites are combined to de- Paleomagnetic results rive the paleomagnetic direction for analysis of the eruption age: All paleomagnetic samples produced stable directions of mag- 13.5° declination, 73.1° inclination, Fisher precision factor k = netization following stepwise demagnetization. Thermal demag- ␣ netization had little effect until >400 °C and then near complete 1296, 95 = 0.7°, N = 27 sites (Fig. 5B). demagnetization at 550 °C, typical of fine-grained (single domain) The high-precision measurements of paleomagnetic directions magnetite as the magnetic carrier (see LLM014B, Fig. 7C). Alternat- also addresses the question of origin of the inner tephra cone and ing field demagnetization, applied to most specimens, exhibited outer spatter rampart. Importantly, site RB3 is located on the linear decay to the origin on orthogonal projections (Fig. 7). The inner eastern wall of the outer spatter rampart (Fig. 2) and shows Can. J. Earth Sci. Downloaded from www.nrcresearchpress.com by Dr. Glyn Williams-Jones on 06/29/20 median destructive fields range from 20–80 mT and final direc- evidence of stratigraphically younger (and darker) spatter ejected tions are stable up to 200 mT alternating field demagnetization ballistically and agglutinated onto the inclined inner wall (Fig. 6). (see LLM011A, Fig. 7A) indicative of single-domain magnetite. Note Characteristic remanent magnetization directions of only the that single-domain magnetite is the optimal magnetic mineral younger, darker spatter (6.0°, 77.6°) is an insignificant separation and grain size for producing reliable paleomagnetic directions. (0.8° ± 3.8°) from the mean direction of the inner rampart wall The magnetic susceptibility values of basalt range from 3.0 to without spatter (9.7°, 77.9°) (Fig. 6). Furthermore, while the direc- 67.3 × 10−3 SI with a median value of 15.1 × 10−3 SI. Mean natural tion of site RB3 is the most distant from the N = 27 site mean, the remanent magnetization for the basalts is 30.5 A/m (range 4.5– outlier test of Fisher et al. (1993) shows that it is concordant with 74.0 A/m). the rest of the site means (p = 83.4%), assuming a Fisher distribu- All samples are normally magnetized, as expected for basalt tion (Fig. 5C). Note that using the same test, site RB4 has only p = from a late Holocene eruption (Table 3). The site mean directions 1.53 × 10−3% probability of belonging to the Fisher distribution of are remarkably consistent in direction. Of the five sites sampled the other N = 27 sites. The Fisher precision factor k = 1296 is higher for paleomagnetic analysis, four have uncertainty circles that fall than 91% (i.e., rank 189 of 206) of the extremely well constrained within the uncertainty circle about the mean (Fig. 5A). Site RB4 lava flows measured in Hawaii (Hagstrum and Champion 2002).

Published by NRC Research Press Pagination not ﬁnal (cite DOI) / Pagination provisoire (citer le DOI)

Williams-Jones et al. 9

Fig. 7. Examples of typical, well-behaved demagnetization data obtained from two of the sampling sites at Tseax (Table 3). For each sample, the stereographic plots (left figures) show magnetization directions after stepwise demagnetization, where the filled circles lie on the lower hemisphere. The orthogonal plots (right figures) show horizontal projections as filled circles and vertical projections as open circles. Both alternating field and thermal demagnetization plots are given for sample LLM014A/B from site RB2. Natural remanent magnetization is shown with a circle and cross. Units are in amperes per metre. [Colour online.] For personal use only. Can. J. Earth Sci. Downloaded from www.nrcresearchpress.com by Dr. Glyn Williams-Jones on 06/29/20

Published by NRC Research Press Pagination not ﬁnal (cite DOI) / Pagination provisoire (citer le DOI)

10 Can. J. Earth Sci. Vol. 00, 0000

Fig. 8. Combined age information. (A) The likelihood curve for the age of the Tseax eruption constrained to postdate the 14C ages and predate the oral history, dendrochronology, and the historical geomagnetic declination record. The 95.4% age limits are marked by the light grey rectangle. (B) The observed declination, with 95% confidence limits, of the paleomagnetic study (cyan band), and the time evolution estimated by five methods. In black, the written historical measurements (Herbert 1926) starting with three dots from Captain Cook in 1778 and Captain Vancouver in 1793. The other paths are described in the text. The dark grey trapezoid displays the most likely range of declination trajectories from before the written historical measurements. (C) Geomagnetic direction time evolution estimated by the same five methods compared with the ␣ observed paleomagnetic direction (cyan, 95 confidence circle). The black circle segment marking the written historical record neglects any inclination information. The star marks the time-average geocentric axial dipole direction for Tseax latitude (55.11085°N). [Colour online.] For personal use only.

Thus, in spite of uncertainties in radiocarbon dates, the new paleomagnetic direction could be compared with a reference sec- paleomagnetic data, detailed geological mapping by Le Moigne ular variation curve based on written historical observations and Can. J. Earth Sci. Downloaded from www.nrcresearchpress.com by Dr. Glyn Williams-Jones on 06/29/20 et al. (2020), and uniform geochemical compositions unambigu- dated archeomagnetic and paleomagnetic magnetic remanence ously support the hypothesis that Tseax cone, the Satellite cone, directions. As there are no direct measurements in the region the spatter rampart, tephra deposits, and lava flows were all com- before Cook’s voyage, one can use global spherical harmonic geo- ponents of a single contemporaneous eruptive episode. magnetic models constructed from historical (Jackson et al. 2000) and archeomagnetic and paleomagnetic records (CALS3k.3, Korte Combined age constraints The Tseax volcanic eruption took place over a short interval. et al. 2009; CALS3k.4, Korte and Constable 2011). These can be The oral history and dendrochronology data constrain the event calculated for any location in northwestern North America; how- to before the end of the 19th century. The reliable written histor- ever, they are dominated by data from other continents. The clos- ical observations of declination (Herbert 1926) all come from a est compilation is from southwestern North America (Hagstrum period during which the local geomagnetic field had a far greater and Blinman 2010). The directions are transferred to the study easterly declination than the time of the Tseax eruption, 13.5°, region by assuming that the geomagnetic field is 100% dipolar, with the implication that the eruption took place before the first converting each direction to a virtual geomagnetic pole, and then direct measurements by James Cook in 1778 CE. In principle, the deriving the dipolar field direction for the Tseax location. Note

Published by NRC Research Press Pagination not ﬁnal (cite DOI) / Pagination provisoire (citer le DOI)

Williams-Jones et al. 11

that Casas and Incoronato (2007) estimate a maximum relocation Casas, L., and Incoronato, A. 2007. Distribution analysis of errors due to reloca- error of 0.25°/100 km due to non-dipole effects. tion of geomagnetic data using the ‘Conversion via Pole’ (CVP) method: implications on archaeomagnetic data. Geophysical Journal International, The important observation from these independent geomag- 169(2): 448–454. doi:10.1111/j.1365-246X.2007.03346.x. netic secular variation curves (Figs. 8B, 8C) is that they bracket the Chadima, M., and Hrouda, F. 2006. Remasoft 3.0 – A user-friendly paleomagnetic observed paleomagnetic direction from the Tseax volcanics, but the data browser and analyzer. Travaux Géophysiques, 27: 20–21. variability argues against attempting any quantitative likelihood Collinson, W.H. 1915. In the wake of the war canoe. Musson Book Company, method of age estimation. All records display a similar trend back Toronto, Ont. Edwards, B.R., and Russell, J.K. 2000. Distribution, nature, and origin of Neo- to about 1650 CE, with a relatively uniform sweep in declination of gene–Quaternary magmatism in the northern Cordilleran volcanic province, about 7.5°/100 years. There is no apparent agreement before Canada. Geological Society of America Bulletin, 112(8): 1280–1295. doi:10.1130/ 1650 CE. Extrapolating 8° ± 2°/100 years declination drift towards 0016-7606(2000)112<1280:DNAOON>2.0.CO;2. the north (based on the range of slopes of the declination curves in Elliott, R.L., Koch, R.D., and Robinson, S.W. 1981. Age of basalt flows in the Blue River Valley, Bradfield Canal Quadrangle. In the U.S. Geological Survey of Fig. 8B), we achieve concordance with our paleomagnetic declina- Alaska, Accomplishments during 1979. U.S. Geological Survey Circular 823-B: tion between 1600 and 1700 CE depending on the starting point of pp. B115–B116. the extrapolation. Combining this interval with the complemen- Evenchick, C.A., Mustard, P.S., Greig, C.J., McMechan, M.E., Ritcey, D.H., Smith, G.T., tary 14C constraints would suggest the Tseax eruption took place and Ferri, F. 2008. Geology, Nass River, British Columbia. Geological Survey of Canada, Open File 5705. doi:10.4095/225792. in the last quarter of the 17th century. Fisher, N.I., Lewis, T., and Embleton, B.J.J. 1993. Statistical Analysis of Spherical Data. Cambridge University Press, 329pp. Conclusions Gallo, R. 2018. History and dynamics of explosive volcanism at Tseax Cone, British Columbia. B.Sc. dissertation, University of British Columbia. doi:10. In spite of being one of Canada’s youngest and deadliest volca- 14288/1.0366164. nic eruptions, important questions have remained, particularly Greve, A., Turner, G.M., Conway, C.E., Townsend, D.B., Gamble, J.A., and whether the Tseax tephra cone was formed during two distinct Leonard, G.S. 2016. Palaeomagnetic refinement of the eruption ages of Holo- eruptive episodes (14th and 18th centuries; Wuorinen 1978). The cene lava flows, and implications for the eruptive history of the Tongariro new radiocarbon data based on wood charcoal fragments from a Volcanic Centre, New Zealand. Geophysical Journal International, 207(2): 702–718. doi:10.1093/gji/ggw296. carbonized horizon immediately beneath tephra deposits proxi- Hagstrum, J.T., and Champion, D.E. 2002. A Holocene paleosecular variation mal to the Tseax tephra cone is similar to earlier measurements record from 14C-dated volcanic rocks in western North America. Journal of and, while not definitive, suggests that this eruption likely oc- Geophysical Research: Solid Earth, 107(B1): 2025. doi:10.1029/2001JB000524. curred after 190 ± 15 14C years BP. When combined with earlier Hagstrum, J.T., and Blinman, E. 2010. Archeomagnetic dating in western North radiocarbon data, the paleomagnetism, and oral and dendro- America: an updated reference curve based on paleomagnetic and archeomagnetic data sets. Geochemistry, Geophysics, Geosystems, 11:Q06009. doi: chronology information, the event is constrained to have oc- 10.1029/2009GC002979. curred during the interval 1675–1778 CE (95.4% probability), and Hanson, G. 1923. Reconnaissance between Skeena River and Stewart, B.C. Can- likely before 1700 CE. Importantly, the characteristic remanent ada Department of Mines, 2031: 17. magnetization directions of the main tephra cone, spatter ram- Hawkesworth, C.J., Blake, S., Evans, P., Hughes, R., Macdonald, R., Thomas, L.E., et al. 2000. Time scales of crystal fractionation in magma chambers—integrating parts, and lava flows are indistinguishable. Thus, by integrating physical, isotopic and geochemical perspectives. Journal of Petrology, 41(7): new paleomagnetic and geochemical data along with detailed 991–1006. doi:10.1093/petrology/41.7.991. volcanological mapping (Le Moigne et al. 2020), we clearly show Herbert, W. 1926. The march of the compass in Canada. Topographic Survey of that there is no statistically significant difference in geochemistry Canada Bulletin, 58: 3–20. doi:10.1139/tcs-1967-0104. or paleomagnetic direction between the explosive and effusive Hickson, C.J., and Edwards, B.R. 2001. Volcanoes and volcanic hazards in Canada. In A synthesis of geological hazards in Canada. Edited by G.R. Brooks. Geolog- For personal use only. deposits making up Tseax volcano and its surroundings. As such, ical Survey of Canada Bulletin, 548: 1–248. the Fe-rich Mg-poor basanite–trachybasalt lava and tephra cones Higgins, M.D. 2009. The Cascadia megathrust earthquake of 1700 may have formed during a single eruptive episode. rejuvenated an isolated basalt volcano in western Canada: Age and petro- graphic evidence. Journal of Volcanology and Geothermal Research, 179(1–2): Acknowledgements 149–156. doi:10.1016/j.jvolgeores.2008.10.016. Jackson, A., Jonkers, A.R.T., and Walker, M.R. 2000. Four centuries of geomag- This work benefited greatly from detailed reviews of two anon- netic secular variation from historical records. Philosophical Transactions of ymous reviewers and was supported by Natural Sciences and En- the Royal Society A: Mathematical, Physical and Engineering Sciences, gineering Research Council of Canada (NSERC) Discovery grants 358(1768): 957–990. doi:10.1098/rsta.2000.0569. to G. Williams-Jones, J.K. Russell, and R.W. Barendregt. Many Kirschvink, J.L. 1980. The least-squares line and plane and the analysis of palaeomagnetic data. Geophysical Journal of the Royal Astronomical Society, 62(3): thanks to Deana Nyce of the Wilp Wilxo’oskwhl Nisg_a’a Institute 699–718. doi:10.1111/j.1365-246X.1980.tb02601.x. (WWNI) for her continued and enthusiastic support and Harry Korte, M., and Constable, C.G. 2005. Continuous geomagnetic field models for Nyce (Nisg_a’a Lisims Government) for his knowledge of the the past 7 millennia: 2. CALS7K. Geochemistry Geophysics Geosystems, 6: salmon ecology and fishery. Mansell Griffin (Nisg_a’a Lisims Gov- Q02H16. doi:10.1029/2004GC000800. ernment) generously shared his extensive knowledge of Tseax and Korte, M., and Constable, C. 2011. Improving geomagnetic field reconstructions for 0–3 ka. Physics of the Earth and Planetary Interiors, 188(3–4): 247–259. the Nisg_a’a lava beds. We are grateful to Irene Squires (WWNI) for doi:10.1016/j.pepi.2011.06.017. her generous hospitality, to Rachel Warwick and Alex Wilson for Korte, M., Donadini, F., and Constable, C.G. 2009. Geomagnetic field for 0–3 ka: assistance during field work, and to Margaret Vanderberg for her 2. A new series of time᎑varying global models. Geochemistry, Geophysics, Can. J. Earth Sci. Downloaded from www.nrcresearchpress.com by Dr. Glyn Williams-Jones on 06/29/20 assistance in providing historical information and field notes. Geosystems, 10:Q06008. doi:10.1029/2008GC002297. Thanks also to Michael Higgins (Université du Québec à Chi- Lanza, R., and Zanella, E. 2003. Paleomagnetic secular variation at Vulcano (Aeolian Islands) during the last 135 kyr. Earth and Planetary Science Letters, coutimi) for supplying many of the older references. This research 213(3–4): 321–336. doi:10.1016/S0012-821X(03)00326-1. was carried out as part of BC Parks/NLG permit 108312. Author Le Moigne, Y., Williams-Jones, G., Kelfoun, K., Labazuy, P., Vigouroux, N., and contributions: The project was conceived by GWJ, RWB, and JKR; Russell, J.K. 2018. Investigating Canada’s deadliest volcanic eruption: from the paleomagnetic sampling, methodology, and analysis was com- field observations to lava flow modeling. Cities on Volcanoes 10 conference, IAVCEI, Naples, Italy. pleted by RWB and RJE; YL and RG completed the detailed geolog- Le Moigne, Y., Williams-Jones, G., Russell, K., and Quane, S. 2020. Volcanological ical mapping and collected samples for radiocarbon dating; and map of Tseax volcano, British Columbia, Canada. Journal of Maps, 16(2): all authors contributed to generation of the figures and the writ- 363–375. doi:10.1080/17445647.2020.1758809. ing of the paper. Lowdon, J.A., Robertson, I.M., and Blake, W. 1971. Geological Survey of Canada Radio- carbon Dates XI. Radiocarbon, 13(2): 255–324. doi:10.1017/S0033822200008456. Mahgoub, A.N., Böhnel, H., Siebe, C., and Chevrel, M.O. 2017. Paleomagnetic References study of El Metate shield volcano (Michoacán, Mexico) confirms its monoge- Barbeau, M. 1935. Volcanoes on the Nass. Canadian Geographical Journal, 10(5): netic nature and young age (ϳ1250 CE). Journal of Volcanology and Geother- 215–225. mal Research, 336: 209–218. doi:10.1016/j.jvolgeores.2017.02.024.

Published by NRC Research Press Pagination not ﬁnal (cite DOI) / Pagination provisoire (citer le DOI)

12 Can. J. Earth Sci. Vol. 00, 0000

Marshall, P. 1975. Ice-blocked tubes in the Aiyansh flow, British Columbia. Arctic Newfoundland and Saint-Pierre et Miquelon Hurricane of September 1775”. and Alpine Research, 7(4): 399–400. doi:10.2307/1550184. The Northern Mariner, 6(3): 11–23. McCullagh, J. 1918. Ignis, a parable of the Great Lava Plain. Mission Press, Ai- Souther, J.G. 1970. Recent volcanism and its influence on early native cultures of yansh, British Columbia. northwestern British Columbia. In Early Man and Environments in North- Németh, K., and Kereszturi, G. 2015. Monogenetic volcanism: personal views and west North America. pp. 53–64. 14 discussion. International Journal of Earth Sciences, 104(8): 2131–2146. doi:10. Stuiver, M., and Polach, H. 1977. Discussion reporting of C data. Radiocarbon, 19 1007/s00531-015-1243-6. (3): 355–363. doi:10.1017/S0033822200003672. Stuiver, M., and Reimer, P.J. 1993. Extended 14C data base and revised CALIB 3.0 Nicholls, J., Stout, M.Z., and Fiesinger, D.W. 1982. Petrologic variations in Qua- 14C age calibration program. Radiocarbon, 35(1): 215–230. doi:10.1017/ ternary volcanic rocks, British Columbia, and the nature of the underlying S0033822200013904. upper mantle. Contributions to Mineralogy and Petrology, 79(2): 201–218. Sun, S.S., and McDonough, W.F. 1989. Chemical and isotopic systematics of doi:10.1007/BF01132888. oceanic basalts: implications for mantle composition and processes. Geolog- Nicholls, J., Stout, M.Z., Machacek, J., and Michael, P. 1997. Volume-composition ical Society, London, Special Publications, 42(1): 313–345. doi:10.1144/GSL.SP. relations in concentrically zoned crystals; application to thermodynamic 1989.042.01.19. modeling of igneous processes. The Canadian Mineralogist, 35(5): 1311–1323. Sutherland Brown, A. 1969. Aiyansh lava flow, British Columbia. Canadian Jour- doi:10.1007/BF01132888. nal of Earth Sciences, 6(6): 1460–1468. doi:10.1139/e69-149. Nisg_a’a Nation. 2004. Laxmihl, Nisg_a’a Final Agreement 2002/2003 Annual Re- Symons, D.T.A. 1975. Age and flow direction from magnetic measurements on port. Available from http://nnkn.ca/files/u28/Implementation%20Annual%20 the historic Aiyansh flow, British Columbia. Journal of Geophysical Research, Report%202002%20-%202003%20Fire.pdf. 80(17): 2622–2626. doi:10.1029/JB080i017p02622. Pinton, A., Giordano, G., Speranza, F., ÞórðarsonÞ 2018. Paleomagnetism of Ho- Trolese, M., Giordano, G., Cifelli, F., Winkler, A., and Mattei, M. 2017. Forced locene lava flows from the Reykjanes Peninsula and the Tungnaá lava se- transport of thermal energy in magmatic and phreatomagmatic large vol- quence (Iceland): implications for flow correlation and ages. Bulletin of ume ignimbrites: Paleomagnetic evidence from the Colli Albani volcano, Volcanology, 80: 10. doi:10.1007/s00445-017-1187-8. Italy. Earth and Planetary Science Letters, 478: 179–191. doi:10.1016/j.epsl.2017. Purssell, T. 1993. Gitwinksihlkw Bridge Foundation Design Summary Report. 09.004. Ministry of Transportation and Highways, North West Region, Province of Turner, G.M. 1987. A 5000 year geomagnetic palaeosecular variation record from British Columbia, File M52-45-2998. western Canada. Geophysical Journal International, 91(1): 103–121. doi:10.1111/ j.1365-246X.1987.tb05215.x. Ramsey, C.B. 2009. Bayesian analysis of radiocarbon dates. Radiocarbon, 51: 337–360. doi:10.1017/S0033822200033865. Van Der Heyden, P., Woodsworth, G.J., and Snyder, L.D. 2000. Reconnaissance geological mapping in southwest Nass River map area, British Columbia. Ramsey, C.B., Scott, E.M., and Van der Plicht, J. 2013. Calibration for archaeolog- Geological Survey of Canada, Current Research 2000-A6. doi:10.4095/211132. ical and environmental terrestrial samples in the time range 26–50 ka cal BP. Wuorinen, V. 1976. The Aiyansh volcano. British Columbia Geographical Series, Radiocarbon, 55: 2021–2027. doi:10.2458/azu_js_rc.55.16935. 22: 143–152. Reimer, P.J., Baillie, M.G.L., Bard, E., Bayliss, A., Beck, J.W., Bertrand, C.J.H., et al. Wuorinen, V. 1978. Age of Aiyansh Volcano, British Columbia. Canadian Journal 2004. Intcal04 terrestrial radiocarbon age calibration, 0-26 Cal Kyr BP. Radio- of Earth Sciences, 15(6): 1037–1038. doi:10.1139/e78-111. carbon, 46: 1029–1058. doi:10.1017/S0033822200032999. Yukutake, T. 1967. The westward drift of the Earth’s magnetic field in historic Roberts, M.C., and McCuaig, S. 2001. Geomorphic responses to the sudden block- times. Journal of Geomagnetism and Geoelectricity, 19(2): 103–116. doi:10. ing of a fluvial system: Aiyansh lava flow, northwest British Columbia. The 5636/jgg.19.103. Canadian Geographer, 45(2): 319–323. doi:10.1111/j.1541-0064.2001.tb01492.x. Ruffman, A. 1996. The multidisciplinary rediscovery and tracking of “The Great Appendix appears on the following pages. For personal use only. Can. J. Earth Sci. Downloaded from www.nrcresearchpress.com by Dr. Glyn Williams-Jones on 06/29/20

Published by NRC Research Press Pagination not ﬁnal (cite DOI) / Pagination provisoire (citer le DOI)

Williams-Jones et al. 13

Appendix A natural being) suddenly emerged to block the lava’s ad- vance. For days Gwaxts’agat fought back the lava by blowing Radiocarbon analyses on it with its great nose. Finally, the lava cooled and OxCal script for calibration of radiocarbon calendar ranges Gwaxts’agat retreated into the mountain where it remains Sequence() to this day.” {Curve(“Terrestrial”,“IntCal13.14c”); In 1874, the missionary William Henry Collinson recorded a {Boundary(“BeforeTseax”); story from an old man who was “evidently over eighty years of age Phase(“UnderlyingDates”) … that the eruption occurred when his grandfather was a boy” {R_Date(“GSC-1124”,250,130); (Collinson 1915, p. 269): R_Date(“S-1046”, 625,70); “The river did not always flow where it does now” said he. “It R_Date(“GSC-6150”, 280,50)+N(50,10); flowed along by the base of the mountains on the farther R_Date(“CUIAMS-215254”, 390,15); side of the valley some miles away. It was there the people R_Date(“CUIAMS-215255”, 190,15); }; were encamped when the Nak-nok (Naxnok, supernatural { Boundary(“Tseax”); being) of the Mountain became angry and the fire-stone Phase(“UpperLimit”) flowed down. They were all busy in catching, cleaning, and { Date(1800); };};}; }; cutting up the salmon, to dry in the smoke…Then, when we saw the Nak-nok of the Mountain rushing towards us Historical written and oral records clothed in fire, we fled for our lives. All that day we fled, and There has been speculation (e.g., Akrigg and Akrigg 1977) that at sunset, as we looked back, we saw the spirit cloud with its the eruption of Tseax was observed by Europeans in 1775. Specif- huge wings outspread following us. We reached the foot- ically, Lieutenant Juan Francisco de la Bodega y Quadra, com- hills on this side [current location of Gitwinksilhkw, Fig. 2], manding the schooner Sonora as part of the Hezeta Expedition, which we ascended, and there we took refuge, as all were had anchored in Bucareli Bay (Prince of Wales Island, Alaska) on exhausted, and could run no farther. The river of fire-stone, 24 August 1775 and noted that “The nights are exceptionally swept on by the cloud spirit, drove the river before it across bright and mild because of seven volcanoes of snow and fire, the valley, until it also reached the base of the foot-hills. which their vapours illuminate and temper” (Tovell 2009, p. 36). Here it heaped up, the river which quenched and cooled the The pilot of the Sonora, Francisco Antonio Mourelle de al Rua, and fire-stone, boiling and thundering, and leaving it heaped up chaplain, Padre de la Campa, also wrote that “They felt the heat along the bank as it is today. As night fell, the spirit cloud which they considered would be from the quantity of flames disappeared in the darkness, but the whole valley was on fire, which were, emitted by a volcano, which erupted four or five which continued for many days, until all the trees, and even times a day, the whole locality being illuminated at night by the the ground, were consumed. It was then that we separated glare” (de la Sierra et al. 1930, p. 241). However, as the Sonora was and settled in the two encampments of Giatlakdamiksh anchored more than 280 km west of Tseax across mountainous (Gitlaxt’aamiks) and Giatwinikshilk (Gitwinksilhkw).” terrain, it is extremely unlikely that crew observed the eruption of The anthropologist Barbeau (1935, p. 222) also reported ac- Tseax (e.g., Higgins 2009). Despite the lack of direct written accounts from the Nisg_a’a people living in the Nass Valley at the counts, important details can be learned from local communities beginning of the 19th century: who have preserved information through oral traditions. “…the volcanic eruption soon after broke out. First there For personal use only. was smoke, like that coming out of a house, a big pillar of Nisg_a’a adaawak Catastrophic natural disasters often become an integral part smoke. It was as if a house was burning on the mountain top. of local history (e.g., Masse et al. 2007; Cashman and Cronin 2008). The people saw a big fire. The fire came down the side in In the early Indigenous communities of northwestern North their direction, but not as fast as forest fire. It moved down America, natural disasters were often transformed into stories slowly, very slowly. It was strange and frightful. It was dan- gerous! There were fumes spreading ahead, and those who and legends passed from one generation to the next through oral smelled them were smothered. They died and their body traditions (e.g., Cashman and Cronin 2008). The catastrophic erup- stiffened like rock. Frightened, the people of one tribe dug tion of Tseax forms an important part of the Nisga’a culture, as it _ holes in the ground like underground lodges, and hid significantly impacted the local environment and may have within, scared as they were of the mountain spirits. Like- caused up to 2000 fatalities. Importantly, these adaawak (tradi- wise, the other tribe. That did not keep other people from tional histories and stories) ultimately preserve a detailed observa- dying of the fumes, mostly in the lower of the villages. As tional account of the Tseax eruption (from Nisg_a’a oral traditions; soon as the smoke dispersed some people ran away; a great Nisg_a’a Nation 2004,p.1): many others stayed on. They did not suffer any more from “Long ago, two children were playing down by the river. One the smoke. The fire then rolled down like a river, filled the child caught a salmon and slit open its back. The child stuck lake, and for a time the water was a bed of flames. The stone sticks into the salmon’s back, set it on fire, and returned the was red and hot there for many days. As far as it went, all the

Can. J. Earth Sci. Downloaded from www.nrcresearchpress.com by Dr. Glyn Williams-Jones on 06/29/20 fish to the river. The children were amused to see the salmon way, it was flowing red. It started from the river where swim erratically, smoke rising from its back. The other child the people fished salmon, away up there, and ran down to caught a salmon and slit open its back, inserted a piece of the place where the canyon now is…”. shale, and put it back into the river. The salmon floated on These accounts highlight potentially important observations its side, weighed down by the shale. The children laughed at regarding the eruption of Tseax. Salmon play a central cultural the struggling fish. An elder happened upon the scene and and economic role in the Indigenous communities of northwest- warned the children: “Take care what you do. The salmon ern North America. Mention of the salmon in the adaawak may will curse you and the Creator will respond in kind”. The ground began to tremble and shake. Natures harmony had suggest that the eruption occurred during the spawning season of been upset. A scout was sent to investigate. From the top of the pink salmon (H. Nyce, personal communication, 2019), which Gennu’axwt [Fig. 1], he saw smoke and flames and ran to in the Tseax River is between June and September (Nisg_a’a Fisheries warn the people of their fiery destiny. In panic, some villag- and Wildlife Department 2018). The accounts recorded by Collinson ers fled up the mountain. Others canoed to the far side of the (1915) and Barbeau (1935) suggest that the lava flow was emplaced river but were killed by the lava. As the people watched the relatively slowly, at least where it entered the Nass Valley. The lava lava flow over their villages, Gwaxts’agat (a powerful super- also likely displaced the Nass River from the southern to northern

Published by NRC Research Press Pagination not ﬁnal (cite DOI) / Pagination provisoire (citer le DOI)

14 Can. J. Earth Sci. Vol. 00, 0000

Table A1. Locations for 14C and geochemical samples. Longitude Latitude Sample Sample No. Location / Sample type (°W) (°N) type Source UCIAMS-215254 ϳ700 m northwest of Tseax cone 128.91154 55.114327 C This study UCIAMS-215255 ϳ750 m northwest of Tseax cone 128.91193 55.114107 C This study GSC-1124 Lava-encased cottonwood tree near Nass River 129.13798* 55.214033* C Sutherland Brown (1969); Lowdon et al. (1971) S-1046 Charred trunk in spatter rampart wall 128.89682* 55.110795* C Wuorinen (1978) GSC-6150 Lava-encased tree trunk in growth position Unknown Unknown C Roberts and McCuaig (2001) NR5-C Lava 128.90000* 55.116667* G Nicholls et al. (1982) RG-B2 Volcanic bomb 128.89555 55.111226 G Gallo (2018) RG-B7 Volcanic bomb 128.89372 55.110806 G Gallo (2018) TS-S19 Spatter 128.89605 55.110774 G Gallo (2018) TS-S20 Spatter 128.89605 55.110772 G Gallo (2018) TS-S40a Lapilli from Satellite cone 128.89943 55.115161 G Gallo (2018) TS-S52 Lava 129.17951 55.192567 G Gallo (2018) TS-S57 Lava 128.89773 55.112694 G Gallo (2018) RG-S16 Tephra main cone 128.90724 55.108679 G Gallo (2018) RG-S25 Tephra main cone 128.88559 55.118367 G Gallo (2018) Note: C, radiocarbon; G, geochemistry. *Approximate locations.

margin of the Nass Valley. Although speculative, possible causes of volcanic gases, either from direct degassing of the lava or from the the reported fatalities may be inferred, notably that “There were interaction of lava and water bodies leading to haze and smog fumes spreading ahead, and those who smelled them were smoth- (“laze” or “vog”) (e.g., Williams-Jones and Rymer 2015; Carlos et al. ered” (Barbeau 1935, p. 222). This may refer to the hazards of 2018). For personal use only. Can. J. Earth Sci. Downloaded from www.nrcresearchpress.com by Dr. Glyn Williams-Jones on 06/29/20

Published by NRC Research Press Pagination not ﬁnal (cite DOI) / Pagination provisoire (citer le DOI)

Williams-Jones et al. 15

Fig. A1. Radiocarbon age estimates presented as 14C year BP and calibrated date. Calibrations performed using OxCal version 4.3.2 (Ramsey 2009) and the IntCal13 atmospheric curve (1 ␴; blue curves) (Ramsey et al. 2013). Red curve indicates the conventional radiocarbon age (±1 ␴)in the sample and the grey histogram represents the calibrated probability age distribution. Calendar date ranges shown for 68.2% and 95.4% probability conﬁdence intervals (black lines beneath histogram). Plots generated from OxCal version 4.3.2 (Ramsey 2009). [Colour online.] For personal use only. Can. J. Earth Sci. Downloaded from www.nrcresearchpress.com by Dr. Glyn Williams-Jones on 06/29/20

References Carlos, W.G., Gross, J.E., Jamil, S., Dela Cruz, C.S., Damby, D., and Tam, E. 2018. Akrigg, G.P.V., and Akrigg, H.B. 1977. British Columbia Chronicle, 1847–1871: Volcanic Eruptions and Threats to Respiratory Health. American Journal of Gold & Colonists. Discovery Press. Respiratory and Critical Care Medicine, 197(12): 21–22. doi:10.1164/rccm. Barbeau, M. 1935. Volcanoes on the Nass. Canadian Geographical Journal, 10(5): 19712P21. 215–225. Cashman, K.V., and Cronin, S.J. 2008. Welcoming a monster to the world: Myths,

Published by NRC Research Press Pagination not ﬁnal (cite DOI) / Pagination provisoire (citer le DOI)

16 Can. J. Earth Sci. Vol. 00, 0000

oral tradition, and modern societal response to volcanic disasters. Journal of Nisg_a’a Nation 2004. Laxmil. Nisg_a a Final Agreement 2002/2003 Annual Report. Avail- Volcanology and Geothermal Research, 176(3): 407–418. doi:10.1016/j.jvolgeores. able from http://nnkn.ca/ﬁles/u28/Implementation%20Annual%20Report%20 2008.01.040. 2002%20-%202003%20Fire.pdf. Collinson, W.H. 1915. In the wake of the war canoe. Musson Book Company, Ramsey, C.B. 2009. Bayesian analysis of radiocarbon dates. Radiocarbon, 51: Toronto, Ont. 337–360. doi:10.1017/S0033822200033865. de la Sierra, B., Baker, A.J., and Wagner, H.R. 1930. Fray Benito de la Sierra’s Ramsey, C.B., Scott, E.M., and Van der Plicht, J. 2013. Calibration for archaeolog- ical and environmental terrestrial samples in the time range 26–50 ka cal BP. account of the Hezeta Expedition to the Northwest Coast in 1775. California Radiocarbon, 55: 2021–2027. doi:10.2458/azu_js_rc.55.16935. Historical Society Quarterly, 9(3): 201–242. doi:10.2307/25178081. Tovell, F.M. 2009. At the far reaches of empire: the life of Juan Francisco de la Masse, W., Barber, E., Piccardi, L., and Barber, P. 2007. Exploring the nature of Bodega y Quadra. UBC Press. 496pp. [ISBN: 9780774813679.] myth and its role in science. Geological Society Special Publication, 273: Williams-Jones, G., and Rymer, H. 2015. Chapter 57 - Hazards of volcanic gases. In 9–28. doi:10.1144/GSL.SP.2007.273.01.02. The encyclopedia of volcanoes. Edited by H. Sigurdsson, B. Houghton, Nisg_a’a Fisheries and Wildlife Department. 2018. 2018 Nass stock assessment H. Rymer, J. Stix, and S. McNutt. 2nd ed. Academic Press. pp. 985–992. doi: update. Nisg_a’a Lisims Government. 10.1016/B978-0-12-385938-9.00057-2. For personal use only. Can. J. Earth Sci. Downloaded from www.nrcresearchpress.com by Dr. Glyn Williams-Jones on 06/29/20

Published by NRC Research Press