Magmatic recharge during formation and resurgence of the Valles caldera, New Mexico, USA: evidence from quartz compositional zoning and geothermometry

Transcription

1 Magmatic recharge during formation and resurgence of the Valles caldera, New Mexico, USA: evidence from quartz compositional zoning and geothermometry Jack Wilcock Department of Earth and Planetary Sciences McGill University Montréal, Québec, Canada August 2010 A thesis submitted to McGill University in partial fulfillment of the requirements of the degree of Master of Science Jack Wilcock, 2010

2 Abstract The Valles caldera complex in north-central New Mexico, USA, represents the type example of resurgent caldera system, characterised by eruption of two voluminous high-silica rhyolite ignimbrites (the Otowi and Tshirege Members of the Bandelier Tuff) at ± Ma and ± Ma, respectively. Refined dating has shown that resurgence occurred shortly after eruption of the Tshirege, or Upper Bandelier Tuff (UBT). Central resurgence of ~1000 m was accompanied by small-volume eruptions of the Deer Canyon Rhyolite, followed closely by the Redondo Creek Rhyodacite. The Cerro del Medio Rhyolite lava dome complex is a product of ring fracture volcanism following resurgence, erupting at ± Ma. A central aim of this study was to find evidence for magmatic recharge during this geologically short (~ 27 ka) time period. We have combined cathodoluminescence (CL) imaging and titanium-in-quartz geothermometry techniques to individual quartz crystals from 1) different stratigraphic horizons of the UBT ignimbrite, 2) samples of the Deer Canyon Rhyolite and 3) the Cerro del Medio Rhyolite lavas. CL imaging reveals that ~80% of the erupted volume UBT ignimbrite contains unzoned quartz crystals (average concentration = 28 ± 2 ppm Ti), recording relatively isothermal temperatures of C. An abrupt occurrence of compositionally zoned quartz crystals ) within the mid-to-late erupted UBT ignimbrite units 3-5 reveals evidence for interaction with hotter magma. Corresponding titanium-in-quartz measurements of outer, bright CL rims (71 ± 9 ppm Ti) reveal temperature increases of ~100 C relative to the start of the UBT eruption. We have discovered an interesting heterogeneity within the ii

3 Deer Canyon Rhyolite lavas, with strong spatial control on eruption of porphyritic lavas containing complexly zoned quartz crystals onto the western regions of the resurgent dome. Conversely, crystal-poor to aphyric lavas containing small, unzoned quartz crystals are confined to eastern areas of the resurgent dome. The Cerro del Medio rhyolite lavas are sparsely porphyritic to aphyric, and contain unzoned quartz with titanium concentrations more than 40 ppm greater than the cores of UBT quartz. The quartz-free Redondo Creek Rhyodacite is the most primitive silicic material erupted during the Valles caldera cycle. Intrusion of this hotter magma into a residual UBT crystal mush zone may have facilitated 1) eruption and geochemical/thermal heterogeneity within the Deer Canyon Rhyolite, 2) resurgence of the caldera and 3) expulsion of hot, crystal-poor rhyolite batches from the mush zone, which were erupted as the Cerro del Medio complex. iii

4 Résumé Le complexe de caldeiras de Valles au centre-nord du Nouveau-Mexique, Etats- Unis, constitue l exemple type d un système de caldeira résurgente, caractérisé par l éruption de deux volumineuses ignimbrites rhyolitiques riche en silice (les Membres Otowi et Tshirege de la Bandelier Tuff) à ± Ma et ± Ma, respectivement. Des techniques de datation de haute precision ont montré que la résurgence s est produite rapidement après l éruption de la Tshirege ou Upper Bandelier Tuff (UBT). La résurgence du centre de la caldeira d environ 1000 m s est accompagnée des éruptions mineures des rhyolites de Deer Canyon, puis de la rhyodacite de Redondo Creek. Le complexe de dômes de lave rhyolitiques de Cerro del Medio est le produit de volcanisme de fracture annulaire post-résurgence à ± Ma. Un point clé de cette étude était de trouver des évidences de recharge magmatique durant ce court episode ~ 27 ka. Nous avons analysé des cristaux de quartz par imagerie de cathodoluminescence (CL) et géothermométrie du titane dans 1) plusieurs niveaux stratigraphiques de l ignimbrite d UBT, 2) des échantillons des rhyolites de Deer Canyon et 3) des laves des rhyolites de Cerro del Medio. L imagerie CL révèle que ~80% du volume d UBT contient des cristaux de quartz non zonés (concentration moyenne = 28 ± 2 ppm Ti), représentant des températures relativement homogènes de C. Une transition abrupte vers des quartz optiquement zonés dans les unités 3-5 des phases intermediaires et tardives de l éruption d UBT suggère une interaction avec un magma plus chaud. Les mesures correspondantes de titane dans les bordures externes lumineuses des cristaux (70 ± 9 ppm Ti) indiquent iv

5 une augmentation de température d environ 100 C depuis le début de l éruption d UBT. Nous avons découvert une hétérogénéité intéressante dans les laves rhyolitiques de Deer Canyon: les laves porphyriques contenant des cristaux avec zonations complexes sont restreintes aux régions à l Ouest du dôme résurgent, tandis que les laves aphyriques ou pauvres en crisaux, contentant seulement des petits cristaux non zonés sont restreintes aux regions à l Est du dôme résurgent. Les laves rhyolitiques de Cerro del Medio sont faiblement porphyriques à aphyriques, et contiennent des quartz non zonés avec des concentrations de titane supérieures de plus de 40 ppm à celles des noyaux des quartz d UBT. La rhyodacite de Redondo Creek est le materiel silicique le plus primitif émis durant le cycle de la caldeira de Valles et ne contient pas de quartz. L intrusion de ce magma plus chaud dans le résidu cristallin de la chambre magmatique de l éruption d UBT a probablement facilité 1) l éruption des thermiquement et géochimiquement hétérogènes rhyolites de Deer Canyon, 2) la resurgence de la caldeira et 3) l expulsion de portions chaudes et pauvres en cristaux du réservoir rhyolitique, qui ont formé le complexe de Cerro del Medio. v

6 Preface The following thesis consists of original research which is ultimately intended to form a manuscript to be submitted to a peer-reviewed journal. Fieldwork, sample cutting for thin sections, sample crushing and grain picking, cathodoluminescence imaging, LA-ICP-MS analysis and sample preparation for XRF analysis were performed by the author. X-ray fluorescence analyses were performed by Glenna Keating and electron microprobe Ti-in-glass measurements by Shi Lang at the Department of Earth and Planetary Sciences, McGill University. Electron microprobe analyses of feldspar were performed by Marc-Antoine Longpré at the same department. The author was responsible for writing and formatting the following thesis. Likewise, all new scientific data is the responsibility of the author. Prof. John Stix acted as principal research supervisor and Dr. Bill Minarik and Prof. Robert Martin as co-supervisors who formed additional members of the student advisory committee. Dr. Fraser Goff of the Department of Earth and Planetary Sciences, University of New Mexico, acted as principal field guide to the author and also provided critical evaluation of the thesis. vi

7 Acknowledgements I arrived at McGill University new to both the science of volcanology and Canada in general. Studying here and living in Montreal has been made a thoroughly enjoyable experience by numerous individuals from both an academic and social perspective. At this point however, I would like to acknowledge my family for their love and continuous encouragement throughout my university career, reminding me on many an occasion that university is the best years of your life. To this, I have no objection. I extend my gratitude to John Stix, my principal supervisor, who has been a constant source of inspiration throughout this project. I greatly appreciate the unwavering time and energy dedicated to aiding my work whilst still maintaining a hectic departmental chair position and family commitments. Within the field, this project would certainly have been weakened in the absence of Fraser Goff's vast knowledge of the Valles caldera. I thank both Fraser and his wife Cathy for their hospitality and refreshing New Mexican spirit. At McGill, the volcanology research group has been incredibly supportive during my project. I would like to praise Crystal Mann, Christoph Helo, Marc-Antoine Longpré, Guillaume Girard and Michelle Campbell amongst others, for helpful scientific discussion, multiple proof readings, abstract translations, poster advice and general good humour. I also thank Enrique Cardenas, an exchange student, for help with sample preparation. Marc-Antoine was especially generous with his time for help with microprobe work. In all, the students and staff of the Earth Planetary Sciences department provided a enjoyable social and academic environment which I am proud to have been a vii

8 part of. Both Bill Minarik and Bob Martin provided interesting and insightful comments on my work, from their respective research areas. I thank Bill for providing orientation for the ICP-MS work and answering several late-night calls regarding instrument problems. I thank Bob for his enthusiasm for the CL imaging and suggestions for further work. Don Marshall from Relion Industries is acknowledged for his assistance with initial installation and troubleshooting advice with the CL unit. I appreciate the meticulous sample preparation by George Panagiotidis, electron microprobe assistance and advice from Shi Lang and XRF analyses from Glenna Keating. Finally, outside of academia, I would like to thank the good people of Montreal for their vibrant culture and spirit, and making this a fantastic city in which to live and study for the past two years. Merci beaucoup. viii

9 Table of contents Abstract... Résumé... Preface... Acknowledgements... Table of contents... List of figures... List of tables... List of appendices... ii iv vi vii ix xi xii xiii Section 1: Introduction... 1 Section 2: Geologic setting... 4 Section 3: Methodology Sample collection Sample preparation Cathodoluminescence (CL) imaging Laser ablation inductively coupled plasma mass spectrometry Electron microprobe X-ray fluorescence Titanium-in-quartz geothermometry (TitaniQ) Section 4: Results Quartz CL zoning patterns Upper Cerro Toledo Formation and early UBT (units 1g and 2) Mid-to-late erupted UBT (units 3-5) Deer Canyon Rhyolite Cerro del Medio Rhyolite Ti concentrations in quartz Upper Cerro Toledo Formation and early UBT (units 1g and 2) Mid-to-late erupted UBT (units 3-5) Deer Canyon Rhyolite Cerro del Medio Rhyolite Ti concentrations in glass Hornblende dacite pumices Section 5: Discussion Constraining a TiO2 during the UBT eruption and subsequent resurgence Upper Cerro Toledo Formation Early-erupted UBT (Tsankawi Pumice Bed, Units 1g, 1v, 2) Mid-to-late erupted UBT (units 3-5) Sources of error in the rutile saturation model Redondo Creek Rhyodacite TitaniQ crystallisation temperatures for the UBT ix

10 5.3 A thermal signature recorded in quartz? Relative timing of magmatic recharge Nature of the injected magma Pre-eruptive thermal regime of the UBT Resurgence: origin of the post-ubt magmas Deer Canyon Rhyolite Redondo Creek Rhyodacite Cerro del Medio Rhyolite Section 6: Conclusions References Appendix Appendix Appendix Appendix Appendix 5 A preliminary study of cathodoluminescene zoning and geochemistry of 136 feldspar in the Upper Bandelier Tuff and resurgence units... Appendix 5.1 Feldspar cathodoluminescence zoning patterns Appendix Feldspar zoning classifications Appendix 5.2 CL zoning patterns results Appendix 5.3 Alkali feldspar geochemistry Appendix Electron microprobe methodology Appendix Electron microprobe results for alkali feldspar Appendix 5.4 Summary Appendix 5 References Appendix 5.5 UBT and Deer Canyon Rhyolite feldspar electron microprobe data 154 Appendix 5.6 Calibration data for feldspar EMP analysis Appendix x

11 List of figures Figure 1. Location and tectonic setting of the Valles caldera, New Mexico, USA... 7 Figure 2. Formalized stratigraphy of the Tewa Group... 9 Figure 3. Extracaldera sample locations Figure 4. Intracaldera geological map and sample locations Figure 5. Quartz cathodoluminescence (CL) zoning pattern classifications Figure 6. Quartz cathodoluminescence (CL) zoning patterns correlated to stratigraphic position... Figure 7. Titanium-in-quartz LA-ICP-MS data Figure 8. Mosaic of characteristic CL zoning patterns in quartz crystals correlated to stratigraphic position... Figure 9. Ternary diagram showing feldspar compositions in the UBT Figure 10. Trace and major element plots showing relationships among the units used in this study... Figure 11. Mosaic of CL images of feldspar crystals from the hornblende dacite pumice, units of the UBT and resurgence-related lavas... Figure 12. Photomicrographs of minerals and textures seen in the Redondo Creek Rhyodacite... Figure 13. Trace element mixing models for selected units Figure 14. Thermal evolution model of the Bandelier magmatic system during collapse and resurgence of the Valles caldera Appendix 5 Figure A5.1. Feldspar cathodoluminescence (CL) zoning pattern classifications Figure A5.2. Feldspar cathodoluminescence (CL) zoning pattern classifications correlated to stratigraphic position... Figure A5.3. Synneusis texture of feldspar and quartz in the Deer Canyon Rhyolite 145 Figure A5.4. UBT and Deer Canyon Rhyolite feldspar geochemistry xi

12 List of tables Table 1. Summary of Ti-in-quartz concentrations obtained with LA-ICP-MS Table 2. Constraining a TiO2 for the Bandelier system Table 3. Crystallisation temperatures for stratigraphic units used in this study Table 4. Titanium diffusion in quartz and timescales of outer zone growth for Unit 3 of the UBT xii

13 List of appendices Appendix 1. Compilation of bulk XRF analyses and ages for rocks from various formations and units of the JMVF... Appendix 2. Titanium concentrations in quartz obtained by LA-ICP-MS and corresponding zoning types... Appendix 3. Un-normalised XRF bulk analyses of rocks from various formations and units of the JMVF... Appendix 4. Standardisation data for LA-ICP-MS analysis of Ti-in-quartz, XRF analysis of bulk rock and glass, and EMP Ti-in-glass analysis... Appendix 5. A preliminary study of cathodoluminescene zoning and geochemistry of feldspar in the Upper Bandelier Tuff and resurgence units... Appendix 5.5 Electron microprobe analysis of feldspar from UBT units 2 and 3 and the Deer Canyon Rhyolite... Appendix 5.6 Calibration data for EMP analyses of feldspar Appendix 6. Sample location table xiii

14 1. Introduction Magmatic recharge is an important issue in volcanology, not least because this process thermally sustains long-lived magmatic systems and also serves as a potential eruption trigger. Recharge events have been recorded at many large silicic centres (e.g., Taupo, New Zealand; Long Valley, California, USA; Valles, New Mexico, USA; Yellowstone, Wyoming, USA), which have undergone multiple eruptive episodes over time intervals on the order of several million years (e.g. Shane et al., 2008; Smith et al., 2010; Wark et al., 2007; Campbell et al., 2009; Girard & Stix, 2009; Vasquez et al., 2009). The evolution of large silicic centres is linked to the duration and rate of magmatic recharge, which in turn controls the repose time and volume of successive eruptive events (Smith, 1979; Spera & Crisp, 1981; Lipman, 2007; Annen, 2009). Magmatic systems of this size often produce caldera-forming-eruptions (CFEs). These events erupt tens to thousands of cubic kilometres of material, which deposits in the form of large ignimbrite sheets (Smith, 1979; Cole et al., 2005). In the case of large, shallow chambers, evacuation of the magma during eruption destabilizes the overlying roof, leading to structural collapse which forms large depressions of a diameter approximately equal to the magma chamber (e.g. Roche et al., 2000; Kennedy et al., 2004). After caldera formation some calderas undergo resurgence, whereby the caldera floor is substantially uplifted. Several studies suggest resurgence is largely driven by magmatic recharge in response to the partial draining of the magma chamber during a CFE (Smith & Bailey, 1966; Lipman, 1984; Cole et al., 2005; Acocella, 2010). 1

15 The Valles caldera located in the Jemez Mountains Volcanic Field, north-central New Mexico, USA, represents the type example of a resurgent caldera as identified in the seminal work of Robert Smith and Roy Bailey (1966; 1968) (Fig. 1 A, B). Silicic volcanism has spanned roughly 2 Ma, punctuated by eruption of two voluminous highsilica rhyolite ignimbrites (the Otowi and Tshirege Members of the Bandelier Tuff) totalling 800 km 3 of pyroclastic material (Fig. 2.; Smith & Bailey, 1966; Smith, 1979; Heiken et al., 1990; Goff, 2010). Furthermore, the petrologic textures (e.g. widespread resorption of quartz) of the youngest lavas in the caldera suggests new silicic melt is being generated by remelting of an underlying relict crystal mush zone by new magma. This evidence suggests that the Valles caldera could potentially erupt in the future (Wolff & Gardner, 1995). Therefore, the Valles caldera remains a site of continued interest in understanding the role of magmatic recharge in the caldera lifecycle. Recent 40 Ar/ 39 Ar geochronology work has shown that central resurgence of ~1000 m occurred rapidly within 54 ka of the eruption of the Tshirege or Upper Bandelier Tuff (UBT) (Phillips et al., 2007). Thus, recharge was almost certainly occurring immediately following eruption of the UBT, and was a major control upon resurgence. Extensive mapping of the UBT divides the ignimbrite into six principal flow units which overlie a plinian fall deposit (Fig. 2.; Bailey et al., 1969; Broxton et al., 1995a; Goff et al., 2007). Advances in application of micro-analytical techniques combined with geothermometry allow us to assess the thermal evolution of the UBT eruption by looking at minerals preserved at different stratigraphic horizons and thus different periods of caldera formation. Specifically, this study aims to 1) establish evidence and timing for 2

16 magmatic recharge and 2) determine the role of magmatic recharge on eruption of the UBT ignimbrite, the resurgence-related units, and initial lavas erupted into the caldera moat region after resurgence had completed. Cathodoluminescence (CL) imaging represents a powerful tool for assessing magmatic evolution recorded in mineral zonation (e.g. Müller et al., 2003; Ginibre et al., 2007; Streck, 2008). Compositional zoning of quartz is largely due to variable concentrations of trace elements, in particular titanium (e.g. Götze et al., 2001; Müller et al., 2003; Larsen et al., 2004). Quartz is ubiquitous throughout the samples used in this study, except for the Redondo Creek Rhyodacite (Smith & Bailey, 1966; Goff et al., 2007; Warren et al., 2007). Together with CL imaging, we apply the titanium-in-quartz geothermometer (TitaniQ) of Wark & Watson (2006) to individual quartz crystals sampled from 1) various stratigraphic positions within the UBT, 2) lavas erupted during resurgence (Deer Canyon Rhyolite) and 3) lavas erupted into the caldera ring fracture zone after resurgence (Cerro del Medio Rhyolite). Using this technique, the aim of this study is to generate a thermal evolution model of the Bandelier magmatic system from ~ 1.25 to ~ 1.1 Ma. This establishes magmatic temperatures during 1) pre-eruption periods (Cerro Toledo Formation), 2) the cataclysmic UBT eruption, 3) resurgence and 4) ring fracture volcanism. Within this study we also focus attention on the Redondo Creek Rhyodacite, which was erupted immediately after the Deer Canyon Rhyolite and represents the most primitive silicic material erupted throughout the Valles caldera cycle. The lack of quartz within this unit precludes any analysis by TitaniQ, but the interesting mineralogy and geochemistry of this unit yields interesting information. Importantly, this material may represent a product of magmatic recharge into a residual crystal mush following caldera collapse and can potentially aid in 3

17 answering key questions about resurgent calderas in general. 2. Geologic setting The Jemez Mountains Volcanic Field (JMVF) in north-central New Mexico is a Miocene-Quaternary suite of basaltic to rhyolitic rocks that are an expression of volcanism occurring at the intersection of the north-trending Rio Grande Rift and the northeast-trending Jemez Lineament (Fig. 1 B.; Aldrich, 1986; Self et al., 1986; Heiken et al., 1990; Wolff et al., 2002; Goff & Gardner, 2004; Rowe et al., 2007; Phillips et al., 2007). The Rio Grande Rift consists of a series of north trending en-echelon sedimentary basins with minor late-cenozoic basaltic volcanism extending the length of New Mexico into southern Colorado (Fig. 1 B; Self et al., 1986; Wolff et al., 2002). Palaeostress determinations show that regional east-west extension of the RGR commenced around 32 Ma and continues episodically to the present (Aldrich et al., 1986). A tectonic lull at 18 to 13 Ma preceded renewed extension and formation of much of the JMVF within a several million year period (Rowe et al., 2007). Although a subject of debate, volcanism in the JMVF began within this period, with eruption of two-pyroxene andesites and basalts of the Keres Group around 16.5 Ma. This was associated with an episode of extension in the adjacent Española Basin (Gardner et al., 1986). From seminal work conducted by Bailey et al. (1969) and later revised by Gardner et al. (2010), the formalized stratigraphy divides the rocks of the JMVF into the Keres, Polvadera and Tewa Groups (oldest to youngest, respectively). Of most relevance to this study, the Tewa Group (1.78 ± 0.07 to 0.04 ± 0.10 Ma) 4

18 consists of the La Cueva, Otowi and Tshirege Members of the Bandelier Tuff, Cerro Toledo Formation and Valles Rhyolite (Fig. 2.). Previously unrecognised in the initial nomenclature of Bailey et al. (1969), at least two silicic tuffs (La Cueva Member) have been identified underlying the Otowi Member of the ensuing Bandelier Tuff. Refined 40 Ar/ 39 Ar dates indicate that these two <10 km 3 high-silica rhyolitic ignimbrites (75-78 wt. % SiO 2 ) were emplaced at 1.78 ± 0.07 Ma, 170 ka prior to the Otowi eruption (Spell et al., 1990). Petrologically, the La Cueva Member ignimbrites exhibit geochemical similarity to the Otowi Member Bandelier Tuff, and thus appear to represent an early product of the evolving Bandelier magmatic system (Self et al., 1986; Gardner et al., 1986; Heiken et al., 1990; Rowe et al., 2007). The Otowi Member of the Bandelier Tuff (Bailey et al., 1969), was erupted at ± Ma (Spell et al., 1990), contemporaneous with collapse of the Toledo caldera. The Otowi consists of a basal ash fall bed, the Guaje Pumice Bed of up to 8 m thickness, distributed mainly to the east of Toledo caldera (Self et al., 1986). The Guaje Pumice Bed underlies a massive composite high-silica (76-77 % wt. % SiO 2 ) rhyolite ignimbrite of ~ 350 km 3 volume DRE (Griggs, 1964; Bailey et al., 1969; Self et al., 1986; Gardner et al., 1986). Extracaldera thicknesses can reach 180 m, while within intracaldera settings the Otowi reaches at least 800 m. The Otowi member exhibits surge beds and massive ignimbrites spread radially from Toledo caldera, with deposits thinning with increasing distance from the eruptive centre. The Otowi member was erupted from ring fractures which were more or less coincident with the ensuing Valles caldera (Self et al., 1986; Balsley 1988; Heiken et al., 1990). 5

19 Figure 1. A) Location of the Jemez Mountains Volcanic Field (JMVF) in the western USA. B) Tectonic setting of the JMVF at the intersection of the north-trending Rio Grande Rift and northeast-trending Jemez Lineament. Adapted from Wolff et al. (2002) and Rowe et al. (2007). 6

20 7

21 Figure 2. Formal stratigraphy of the Tewa Group, Jemez Mountains Volcanic Field, New Mexico, USA. The units applicable to this study are in colour. Composite column to the right is the currently accepted subunit classification of the Upper Bandelier Tuff (Broxton & Reneau, 1995; Warren et al., 2007). Subunits 1g, 1v and 2 of the UBT represent ~80 % of the total eruptive volume (Goff, pers. comm.). The Tewa Group was originally defined by Bailey et al. (1969) and revised most recently by Gardner et al. (2010). Ages from Gardner et al. (1986), Phillips et al. (2007) and Spell et al. (1990; 1996). 8

22 9

23 The Cerro Toledo Formation (CTF) consists of grouped rhyolitic domes and associated tephras distributed along the southeastern and northern walls of Valles caldera and within the Toledo embayment (Fig. 3), erupted episodically within a ~ 350 ka period between the two upper (Otowi and Tshirege Member) Bandelier Tuff eruptions (Fig. 2; Gardner et al., 1986; Heiken et al., 1990; Spell et al., 1996). The spatial distribution of domes is believed to represent ring fracture volcanism from Toledo caldera, although the formation of the Toledo embayment may have been coincidental with caldera collapse and could represent a structural weakness (Self et al., 1986; Heiken et al., 1990; Gardner & Goff, 1996). The domes comprise grey, flow-banded to massive rhyolite, intruding domes of older Cerro Rubio dacite and underlying welded ignimbrite of the Otowi Member Bandelier Tuff. The CTF tephras comprise fallout units interspersed with reworked horizons. The Upper Bandelier Tuff (UBT) or Tshirege member (Bailey et al., 1969) followed a ~ 350 ka hiatus in large ignimbrite-forming eruptions. Eruption of this ~ 400 km 3 (DRE) ignimbrite at ± Ma (Phillips et al., 2007) initiated collapse of Valles caldera, overprinting the earlier Toledo caldera (Fig. 3.; Goff, 2010). The rhyolitic basal pumice fall, the Tsankawi Pumice Bed (Bailey et al., 1969), consists of four coarse pumice-fall units and two finer-grained fall units. Each unit has various dispersal patterns (Self et al., 1986) with a maximum aggregate thickness of 3.5 m. The Tsankawi Pumice Bed underlies a non-welded to densely welded rhyolitic ignimbrite that is distributed radially from the Valles caldera (Fig. 3). Extracaldera thicknesses range from 15 to >270 m and from 400 to >1100 m within the caldera (Heiken et al., 1990). In comparison to the Otowi, the UBT is chemically and petrographically more heterogeneous. The UBT displays three dominant components: 1) early-erupted high-silica rhyolite (76-78 wt. % 10

24 SiO 2 ) to late-erupted low-silica rhyolite (~73 wt. % SiO 2 ); 2) distinctive hornblende dacite (69 wt. % SiO 2 ) pumices first appearing within the Tsankawi Pumice Bed and sporadically interspersed within the UBT (Self & Lipman, 1989; Broxton & Rogers, 2007); and 3) syenitic crystal 'clots' which are prominent within late-erupted units (Stimac, 1996; Caress 1994, 1996). From petrologic, geochemical and field relations, the UBT is attributed to be a compound cooling unit constructed of six principal flow units (Fig. 2). The Tsankawi Pumice Bed represents the most fractionated and evolved material, enriched in incompatible elements; this melt is interpreted to have resided in roof regions of a compositionally zoned chamber (Smith & Bailey, 1966; Balsley, 1988; Wolff et al., 2002). Each successive subunit of the UBT ignimbrite is envisaged to have tapped incrementally deeper levels of the chamber, such that the up-section compositional variation noted in the field is an inverted representation of the stratified Bandelier magmatic system prior to eruption (Bailey et al., 1969). The Valles Rhyolite (Griggs, 1964; Bailey et al., 1969; Gardner et al., 2010) comprises domes, flows and tuffs erupted within the Valles caldera following eruption of the UBT. This group covers all units emplaced from Ma to 0.04 Ma, which is a result of central caldera resurgence progressing to ring fracture volcanism, although recent studies have highlighted the heterogeneity among successive domes within the caldera (e.g. Spell & Kyle, 1989; Spell et al., 1996). The Deer Canyon Member ( Ma) includes small-volume high-silica rhyolite lavas, lithic tuffs and breccias emplaced onto the northeast and southwest flanks of the actively-growing resurgent dome immediately after the UBT eruption (Goff & Gardner, 2004; Phillips et al., 2007). From stratigraphic relationships and refined geochronology, this unit slightly predates the 11

25 Figure 3. Radial distribution of the Upper Bandelier Tuff (UBT) from the Valles caldera showing extracaldera sample locations for this study. Numbers with stars correspond to sample numbers used in this study (Appendix 6). Isopach lines indicate thickness (m) of the Tsankawi plinian fallout unit B of the UBT as defined by Self et al. (1986). Greatest deposition of this unit occurred to the northwest, reaching 3.5 m thickness. 12

26 13

27 Redondo Creek Member ( Ma), a series of rhyodacite lavas erupted during the mid-to-late stage of resurgence onto the central and west portions of the dome, and also within the western caldera moat (Goff & Gardner, 2004). Both resurgence-related units are interbedded with caldera fill sediments of lacustrine and fluvial origin, which are products of an intracaldera lake environment formed during resurgence (Fig. 4) (Goff et al., 2007). Seven progressively-younging high-silica rhyolite dome complexes represent a product of ring fracture volcanism following termination of resurgence within Valles caldera. The oldest dome, Cerro del Medio, erupted at ± Ma (Fig. 4; Phillips et al., 2007), comprising five individual rhyolite lava flows with at least two associated pyroclastic flow deposits, and a cumulative volume of ~ 5 km 3 (Gardner et al., 2007). 3. Methodology 3.1 Sample collection We conducted comprehensive sampling of the uppermost tephra units of the Cerro Toledo Formation (Unit 7-14 of Stix, 1989), all subunits of the Upper Bandelier Tuff (UBT) including the Tsankawi Pumice Bed, the Deer Canyon Rhyolite and Redondo Creek Rhyodacite resurgence-related lavas, and all subunits of the first post-caldera moat rhyolite, Cerro del Medio (after Gardner et al., 2007). This sample suite essentially covers formation of the Valles caldera, from pre-ubt volcanism, the entire UBT, resurgencerelated lavas to ring fracture volcanism. We also sampled hornblende dacite pumices present within the Tsankawi Pumice Bed and unit 1g of the UBT at several locations 14

28 (Appendix 6). Spatial variations in mineralogy, welding and crystallization of the UBT outflow sheets have produced numerous subunit classifications which differ from the original stratigraphy of Smith & Bailey (1966). The nomenclature applied in this study uses the subunits as originally defined by Broxton et al. (1995a) and further constrained by Warren et al. (2007) for the upper ignimbrite sections (Fig. 2). The latter study splits unit 4 of Broxton et al. (1995a) into units 4 and 5, since this uppermost flow unit exhibits substantial geochemical, physical and spatial differences (Goff et al., 2007). The plinian Tsankawi Pumice Bed of the UBT was sampled at a roadcut approximately 10 km northwest of the caldera at a location situated within the thickest accumulation of the Tsankawi deposit as indicated by the isopach mapping of Self et al. (1986). UBT subunits were collected from both intracaldera and extracaldera locations (see Appendix 6 for sample location coordinates). Flow units underlying UBT subunit 4 are not recognised within the intracaldera environment. Hence subunits 1-3 of the UBT were sampled from locations on the Parajito Plateau, southeast of the caldera. Samples of the Deer Canyon and Redondo Creek lavas and their associated tuffs were sampled from multiple locations on the resurgent dome, to ensure that we had samples from several vents to assess any spatial relationships (Fig. 4). 3.2 Sample preparation Individual quartz crystals were handpicked following application of a light crush to a bulk sample. We used the sieved fraction (between -1 and 0.5 phi units) yielding the highest return of complete, undamaged crystals. Quartz crystals were mounted in epoxy within 1-inch grain mounts. This technique was used for CL imaging and LA-ICP-MS 15

29 Figure 4. Geologic map and stratigraphic relations of intracaldera units, Valles caldera complex. Deer Canyon Member pyroclastic units are not divided. Numbered stars correspond to sample numbers used in this study (Appendix 6). Adapted from Goff et al. (in prep), Gardner et al. (2010), Phillips et al. (2007) and Smith et al. (1970). 16

30 17

31 analysis of quartz grains from CTF unit 7-14, the Tsankawi Pumice Bed, UBT Unit 1g and Cerro del Medio. Polished thin sections of the remaining units were used for the same methodological approach. Two hornblende dacite pumices were first cleaned with deionised water, dried for 72 hours and the weathered outer surface removed with a rotary grinder to eliminate any erroneous data prior to XRF analysis for major and trace elements. 3.3 Cathodoluminescence imaging Cathodoluminescence (CL) analysis of quartz crystals in thin section and grain mounts was performed using a Reliotron III cold-cathode unit mounted on a Leitz petrographic microscope stage. Beam voltages ranged from kv at 0.5 ma current and 6 Pa operating pressure. The voltage was increased in order to resolve numerous thin oscillatory zones, which may not be apparent at lower voltages. Once a stable beam was obtained, the beam was focussed, reducing the width to ~ 2 mm which was sufficient to excite most quartz samples. For a circular beam on a quartz of 1 mm diameter, the calculated beam power density was ~ 420 W cm -2. A QImaging Retiga EXi camera orientated vertically down the microscope objective provided digital colour images at a spatial resolution of 6.45 μm. Image acquisition times varied depending upon the CL intensity but were generally in the range of 5-30 s. Once a desirable exposure time was calibrated for each sample, successive images used the same exposure time in order to maintain experimental reproducibility. Post-capture image processing was performed with Adobe Photoshop in order to increase the brightness and contrast of zoned quartz grains and to highlight any low- 18

32 contrast diffuse zones. No colour adjustment was made to any image. 3.4 Laser ablation inductively coupled plasma mass spectrometry Titanium concentrations in quartz crystals were obtained using a New-Wave UP- 213 nm laser ablation system attached to a Perkin Elmer/SCIEX ELAN 6100 DRC plus ICP-MS at McGill University. Use of such a system is favoured over the electron microprobe due to significantly lower detection limits of the ICP-MS. Experiments were conducted as either 1) spot analyses to resolve thin compositional zones or 2) line scans to traverse zone boundaries, both at 10 Hz and ~10 J cm -2 laser fluence. Beam sizes for each method ranged from μm, depending upon the thickness and friability of individual quartz crystals and also to avoid melt inclusions. Data reduction, including signal selection and background count determination, was performed with GEMOC 'Glitter' software. Two isotopes of titanium were chosen for this study. The greater abundance of 48 Ti (73.8 %) over 49 Ti (5.5 %) results in higher counts rates for the former isotope, hence this isotope was chosen as most representative. In addition, 29 Si was selected as an internal reference isotope, with 27 Al, 43 Ca and 96 Zr monitored to identify melt inclusions, crystallographic defects or surface contamination. In order to resolve an isobaric interference of Ca on 48 Ti, benitoite (BaTiSi 3 O 9 ), a calcium-free titanium silicate, was used as the primary standard. To compare multiple sessions, a pooled average of 48 Ti concentrations and a pooled standard deviation were calculated. The pooled standard deviation (S p ) is defined by: 19

33 (1) S p = n 1 1 s 2 1 n 2 1 s n k 1 s k 1/ 2 n 1 n 2...n k k where n 1 is the number of analyses of the first session, s 1 the standard deviation of that session and k the overall number of sessions being combined. The pooled standard deviation yields an overall error of 7.3 % (RSD) relative to the ideal stoichiometric Ti in benitoite, and 7.2 % (RSD) relative to the pooled average concentration (Appendix 4). Low-Ti, unzoned quartz from the Tsankawi Pumice Bed was also analysed periodically to assess reproducibility at low Ti concentrations. Two crystals, which were periodically analysed yielded precisions of 3.97 % and 5.19 %, respectively (Appendix 4). 3.5 Electron microprobe We used the electron microprobe to determine Ti concentrations in glass (Table 2). Glass analyses were performed using a JXA JEOL 8900L electron microprobe at McGill University. In order to minimize sodium loss during analysis, a defocused beam of 20 μm diameter, 15 kv and 20 na was used. Count times for major elements (Si, Al, Fe, Mn, Mg, Na, K) were 20 s. The count time for Ti was extended to 100 s to improve statistical precision. Standardisation was performed on homogeneous Yellowstone rhyolite glass (for Na, Al, K, Si) and two basaltic glasses, BMAK (Ti) and BIND (Fe, Mg, Ca), which are international standards from the Smithsonian Institution (Jarosewich et al., 1980). The relative standard deviation for Ti in the BMAK glass standard was 0.8%, and its analysis exhibited excellent agreement with reference values (within 0.01 wt.%, Appendix 4). Prior to analysis, we took care to identify areas of homogenous glass which were free from microlites and surface contamination. For the Tsankawi Pumice Bed, we analysed 20

34 glass adhering to individual quartz crystals, in addition to melt inclusions from each respective crystal. 3.6 X-ray fluorescence A Phillips PW kw spectrometer at McGill was used to acquire major (Si, Ti, Al, Fe, Mn, Mg, Ca, Na, K, P) and trace element (Nb, Rb, Sr, Th, Y, Zr) analyses of several glass separates from the Cerro Toledo Formation and two bulk hornblende dacite pumices from UBT Unit 1g. Two natural rhyolitic glasses, UTR-2 and NBS-278 (Stix et al., 1995; Gladney & Bower, 1985) were used as reference materials, both yielding excellent experimental reproducibility (Appendix 4). 3.7 Titanium-in-quartz geothermometry (TitaniQ) Temperature-dependent substitution of trace elements for Si 4+ within the quartz lattice has been shown to be especially applicable to titanium. Ti is a common trace element in quartz, exhibiting a range of concentrations from <1 to >100 ppm (e.g., Dennen, 1964, 1966; Götze et al., 2001; Müller et al., 2003; Larsen et al., 2004; Wark & Watson, 2006; Shane et al., 2008). For magmatic quartz, temperatures exceeding 500 C are required to fit the Ti 4+ into the fourfold coordinated quartz lattice, since Ti usually proceeds to sixfold coordinated octahedra with O 2-. Elevated temperature increase the elasticity of the quartz lattice, facilitating incorporation of larger ions (Müller et al., 2003). A titanium-in-quartz geothermometer has been calibrated from synthesizing quartz in the presence of rutile at temperatures ranging from C at 1.0 GPa (Wark & 21

35 Watson, 2006). Using this approach, a temperature of crystallisation is given by: (2) T o 3765 C = log X qtz Ti where X is the concentration of Ti in quartz in ppm. Refinements to the thermodynamic principals of the geothermometer suggest that most silicic systems are not saturated in rutile, thereby placing the activity of titania (a TiO2 ) below unity (e.g. Wark & Watson, 2006; Wiebe et al., 2007; Wark et al., 2007). Thus, in the absence of an independent determination of a TiO2, TitaniQ yields a minimum temperature of crystallisation if a TiO2 =1 is applied in the equation: (3) T o 3765 C = log X qtz Ti /a TiO Estimates of a TiO2 can be derived by several means. One approach is via application of the revised rutile saturation model of Hayden & Watson (2007). Rutile saturation in hydrous silicic melts is a function of both temperature and melt composition (FM): (4) log Ti, ppm = T FM Where T is temperature in K and FM a melt compositional parameter derived from an earlier saturation model (Ryerson & Watson, 1987): (5) FM = 1 2 Ca Mg Fe Na K Si Al where the chemical symbols represent cation fractions. The activity of titania is thus calculated using the ratio of the observed melt Ti concentration divided by the solubility 22

36 predicted by the saturation model: (6) a TiO2 = Ti glass measured glass Ti saturation Another approach to estimating the activity of titania is to compare crystallisation temperatures as predicted by Fe-Ti oxide and two-pyroxene geothermometers and adjust the a TiO2 of the TitaniQ equation to match those temperatures predicted from each method. In the Bandelier system, we have temperature estimates obtained by both methods (see section 5.1 below). Hence we can constrain a TiO2 by rearrangement of the TitaniQ geothermometer such that: [log X 3765/T 5.69] (7) a =10 Ti TiO2 where T is a temperature in K determined by an independent method. In summary, this approach is robust, as we can (a) measure Ti concentrations in quartz accurately and precisely, and (b) independently establish a TiO2 values by application of other geothermometers. 4. Results 4.1 Quartz CL zoning patterns A total of 224 individual quartz crystals and fragments were characterised for their respective CL zonation patterns following classifications used by Peppard et al. (2001; see their Fig 11A, B) for quartz from the Bishop Tuff, erupted from the Long Valley 23

37 caldera. Figure 5 shows the zoning classifications applied in this study. Type 1 zoning represents quartz crystals with dark CL cores overgrown by a bright CL rim, or in some cases, multiple bright CL rims. Type 2 zoning represents a quartz crystals with a bright CL core and dark CL rim. Type 3 was used to classify unzoned quartz crystals. The type 1 classification was further subdivided (1a-1d) to account for multiple zones of varying CL intensity and the nature of the core/rim boundary, whether diffuse or sharp (Fig. 5, Appendix 2). All samples displayed variable blue to violet CL colours and also exhibited time-dependent loss of the blue emission wavelength within ~30 seconds of irradiation, gradually reverting to an underlying red CL colour. This phenomena was particularly noticeable within quartz of the upper Cerro Toledo Formation and Tsankawi Pumice Bed units. The Valles caldera sample suite used for this study is characterised by four distinct variations in the nature of CL zoning with respect to stratigraphic position. Figure 6 depicts the quantitative abundance of each quartz zoning type within each unit Upper Cerro Toledo Formation and early UBT (Tsankawi, UBT units 1g and 2) Figure 6 shows a predominance of unzoned quartz crystals (Type 3, Fig 5) within the upper Cerro Toledo Formation (Unit 7-14 of Stix, 1989), the Tsankawi Pumice Bed and units 1g and 2 of the UBT. A minor population of quartz exhibits fine-scale (< 10 μm) oscillatory zones, characterised by alternating bands of bright and dark CL intensity. Additionally, unit 2 contains several normally zoned quartz crystals (Type 2), whereby the core region exhibits increased CL intensity relative to the darker rim. The quartz of the 24

38 Figure 5. Quartz cathodoluminescence (CL) zoning pattern classifications following the terminology of Peppard et al. (2001). White lines define zone boundaries where applicable. The coloured boxes correspond to the colours used on Fig. 6. (A) Type 1, reverse zoning. (B) Type 2, normal zoning. (C) Type 3, unzoned. Additional notes for each subclassification are given in Appendix 2. All scale bars are 0.5 mm. 25

39 26

40 Figure 6. Quartz cathodoluminescence (CL) zoning patterns, correlated to stratigraphic position. Subclassifications follow the terminology of Peppard et al. (2001). Subunit 1v of the UBT was not sampled for this project. The lack of quartz in the Redondo Creek Rhyodacite prevents any CL analysis for this unit. Notes for each subclassification are given in Appendix 2, and images of each zoning type are shown in Fig. 5. Int. = intracaldera setting, Ext. = extracaldera setting. The total number of quartz crystals imaged is represented by n, and thus the abundance of each zoning type is a proportion of this number. 27

41 28

42 Tsankawi Pumice Bed is characterised by ubiquitously distributed melt inclusions which induce local areas of increased CL emission, but are not representative of primary quartz growth zones. We note the predominance of small, μm embayments and rounded morphology of most quartz crystals within these early-erupted units. Together these units represent about 80 % of the total erupted volume of the UBT (F. Goff, pers. comm.) Mid-to-late erupted UBT (ignimbrite units 3-5) By contrast, quartz crystals from the mid to late erupted ignimbrite units (UBT 3, 3T, 4 and 5) show dominantly type 1 zoning (82-95 % abundance). There is a small abundance (5-15 %) of type 3, unzoned quartz (Fig. 6). Zone boundaries within zoned quartz crystals are commonly sharp and concordant, particularly within units 3 and 3T, with zone boundary thicknesses for unit 3 averaging 28.5 μm. The quartz crystals from these units are also sometimes characterised by μm-thick zones of intermediate CL intensity situated between the core and rim and separated by sharp, concordant boundaries. The variation in type 1 zoning (i.e., relative proportions of type 1a, 1b, 1c and 1d) is relatively uniform for the mid-to-late UBT units, which is likely a product of differing degrees of diffusion within the crystal lattice Deer Canyon Rhyolite The resurgence-related Deer Canyon Rhyolite exhibits variation in CL zoning in terms of its spatial distribution on the resurgent dome, although our two samples are derived from the same mapped eruptive unit (Fig. 6). Porphyritic rhyolite sampled from 29

43 the western resurgent dome (sample #13, Fig. 4) has quartz crystals which are both reversely zoned (90% abundance) and unzoned (10% abundance). Zone boundaries range from sharp and concordant to diffuse. Intermediate zone widths range from μm, some showing complex cross-cutting relationships. Embayments are ubiquitous for almost all zoned quartz grains, although distinguishing the temporal relationship between dissolution and rim growth is difficult to reconcile with the resolution of our images. Preliminary observations suggest that dissolution may have occurred prior to growth, as thin (<20 μm) zones of bright CL intensity often surround embayments (see Fig. 8 C, inset). Melt inclusions are confined to core regions, and the sharp CL boundaries imply rapid growth of the outer rims. We also sampled Deer Canyon Rhyolite from a central resurgent dome location (sample #14, Fig. 4). The range of crystal size here is greater ( mm) than the porphyritic sample, exhibiting no dominant crystal size. Excluding two normally-zoned quartz grains, all crystals are unzoned, rounded and/or embayed (Fig. 6). Synneusis texture (after Vance, 1969) of the smaller quartz grains and feldspar in this sample stands in pronounced contrast to the large 2-3 mm diameter quartz crystals from the porphyritic sample Cerro del Medio Rhyolite CL analysis of quartz from the Cerro del Medio ring fracture rhyolite marks a return to dominant type 3 zoning: unzoned, small ( mm), rounded quartz phenocrysts (Fig. 6). Unit South represents the most crystal-rich sample of the Cerro del Medio complex (5 vol. %) (Fig. 4). Under bombardment by the electron beam, we did not witness the high degree of short-lived blue CL emission prevalent with quartz from the 30

44 upper Cerro Toledo Formation and Tsankawi Pumice Bed. In summary, we note (a) the abrupt occurrence of zoned quartz crystals commencing within unit 3 of the UBT ignimbrite and (b) variation of CL zoning in quartz from the resurgence and ring fracture volcanic rocks (Fig. 6). The corresponding titaniumin-quartz concentrations discussed below are closely linked to observed CL zoning. 4.2 Ti concentrations in quartz We analysed 130 individual quartz crystals which had been previously characterised for their respective CL zonation patterns. A number of interesting trends emerge and in this section we correlate the titanium values to the CL zonation patterns. Comprehensive results for Ti-in-quartz are reported in Appendix 2 and plotted collectively on figure 7. Table 1 summarizes the Ti-in-quartz measurements. Since we are dealing with grouped data, it is useful to consider two important statistics. Firstly, the pooled average concentration (X T ) was calculated using the relation: (8) X T = X A N A X B N B X n N n... N A N B N n... Where X A is the average concentration of the first group and N A is the number of individual samples in that group. Secondly, a pooled standard deviation (S T ) is calculated by: (9) = S N A T X 2 A s 2 A N B X 2 B s 2 B N n X 2 n s 2 n N A N B N n X 2 T Where N A is the number of samples, X A is the average concentration and s A is the standard 31

45 Table 1. A summary of titanium-in-quartz concentrations obtained by LA-ICP-MS analysis. The range of values together with the average concentrations is given for each stratigraphic unit. The pooled average concentration and pooled standard deviation was calculated using the methods given in equations (8) and (9). 32

46 33

47 deviation of the group. Equations (8) and (9) are from Downie & Heath (1970) Upper Cerro Toledo Formation and early UBT (units 1g and 2) The unzoned quartz crystals of the upper Cerro Toledo Formation, the Tsankawi Pumice Bed and units 1g and 2 of the UBT fall within a small range of ppm Ti. The pooled average concentration and standard deviation is ± 2.25 ppm Ti (Table 1). The two outlier core Ti values on of 42 ppm and 56 ppm Ti for UBT unit 2 correspond to bright-cl core regions which are overgrown by rims of lower CL-intensity and lower Ti content, referred to as normal zoning (Fig. 7) Mid-to-late erupted UBT (units 3-5) The abrupt appearance of dominant Type 1 CL zoning (dark core with bright CL rim overgrowths) within UBT unit 3 (Fig. 6) is reinforced by core-to-rim increases in titanium within all zoned quartz. Core values are similar to those in quartz from the preceding units, falling within a relatively narrow range of ppm Ti. The pooled average concentration and standard deviation is ± 4.71 ppm Ti (Table 1). A normally zoned quartz crystal (bright cores and dark CL rims) from unit 3 plots with a high core Ti value reflected in the outlier on Fig 7. Similarly, the outlier value of 70 ppm Ti for UBT unit 5U represents a bright CL unzoned quartz. Throughout units 3-5 of the UBT, outer rims show a ppm increase in Ti with respect to core values. The range of rim concentrations is largest within UBT unit 3 ( ppm Ti). Conversely, the smallest range of rim Ti values is seen for unit 3T (76-93 ppm Ti). The maximum rim concentration analysed was 104 ppm Ti from UBT unit 5U, 34

48 Figure 7. Box-whisker plot showing the results of LA-ICP-MS analysis of quartz cores and rims, with respect to stratigraphic position. DC (P) represents the porphyritic Deer Canyon Rhyolite sampled from the western resurgent dome. DC represents the Deer Canyon Rhyolite sampled from the central resurgent dome. The plot is constructed to show the spread of data as defined by the maximum, minimum, lower and upper quartiles and mean concentration. The number of individual analyses (n) on cores and rims is shown for each unit. Outlier values, represented by open diamond symbols, are calculated as 1.5 times the inter-quartile range plus the value of the upper quartile of the dataset (see legend). An a TiO2 of 0.5 was chosen as a representative value for the activity of the system at the onset of eruption of the UBT. The temperature scale is for general reference only. More accurate temperatures for each subunit are given in Table 3. 35

49 36

50 and the minimum rim value of 43 ppm Ti from a complexly zoned quartz crystal with diffuse zone boundaries from UBT unit 3 (Table 1, Appendix 2). The highest average rim concentration is calculated for quartz of UBT unit 3T (82 ppm), and the lowest for UBT unit 4U (63 ppm). Zones of intermediate CL intensity also exhibit Ti concentrations which are transitional between core and rim values. Rim Ti concentrations of quartz from UBT units 3-5 are ± 8.58 ppm when a pooled average concentration and standard deviation are considered (Table 1) Deer Canyon Rhyolite Cores of Deer Canyon Rhyolite quartz crystals for both the western porphyritic sample (DC (P) on Fig. 7) and the central resurgent dome sample (DC on Fig. 7) show titanium concentrations similar to those in quartz cores from the upper CTF and UBT. These fall within a range of ppm Ti. The outlier core value for the DC sample represents a bright CL core of a normally-zoned, type 2 quartz crystal. The porphyritic Deer Canyon Rhyolite sample, containing zoned quartz, exhibits quartz rim concentration values ranging from ppm Ti, similar to the range of concentrations of quartz rims from units 3-5 of the UBT. The CL zoning variation between the two Deer Canyon samples is thus further reflected in compositional differences Cerro del Medio Rhyolite Finally, we note that the unzoned Cerro del Medio quartz samples have Ti contents which are generally similar and overlap with the rim concentrations of UBT units 3-5 and 37

51 the porphyritic Deer Canyon Rhyolite sample. Ti concentrations are grouped within a range of ppm. In summary, all unzoned quartz and core regions of zoned quartz, excluding Cerro del Medio, fall within a range of ppm Ti. The abrupt occurrence of zoned quartz crystals commencing at UBT unit 3 (Fig. 6) is reflected by ppm increases in rim Ti concentrations compared to core values. Quartz from the porphyritic Deer Canyon Rhyolite exhibits rim Ti concentrations similar to rims of zoned quartz from the mid-tolate erupted UBT. The central Deer Canyon sample exhibits quartz with Ti concentrations similar to the cores of UBT quartz. 4.3 Ti concentrations in glass Titanium concentrations within glass of the plinian Tsankawi Pumice Bed, UBT units 1g, 4U, 5U and the Redondo Creek Rhyodacite show an almost threefold increase with respect to decreasing eruption age (Table 2). A total of 14 melt inclusions with an average of 293 ± 2.47 ppm Ti and 20 samples of adhering glass on Tsankawi quartz with an average of 308 ± 2.35 ppm Ti are virtually indistinguishable with respect to Ti concentrations. Analysis of homogeneous glass in UBT units 1g, 4U and 5U exhibit average concentrations of 560 ± 4.48, 814 ± 6.51 and 938 ± 7.51 ppm Ti, respectively. The Redondo Creek Rhyodacite sample yields a glass concentration of 915 ± 7.33 ppm Ti. In summary, there is a clear increase in Ti concentration upsection in the UBT stratigraphy. Bulk analysis of several flow units of the Cerro del Medio Rhyolite complex gives Ti values of 662 ± 110 ppm (Gardner et al., 2007). The low crystal content (<5% 38

52 Figure 8. Mosaic of characteristic CL zoning patterns in quartz crystals with respect to stratigraphic position. All scale bars are 500 μm unless otherwise stated. The symbol * indicates a quartz crystal imaged in a grain mount. CL analysis reveals predominantly zoned (dark core and bright rim) quartz within the mid-late UBT ignimbrite units and porphyritic Deer Canyon Rhyolite. Average titanium concentrations (ppm) calculated from LA-ICP-MS analyses are given for core and rim regions. (A) Unzoned quartz crystals with a bright CL signal from Cerro del Medio flow Unit South, sample #16. (B) Large, complexly zoned quartz crystal from the porphyritic Deer Canyon Rhyolite from the western resurgent dome. (C) A quartz crystal from the same unit exhibiting several embayments. (C, inset) A monochrome enlargement of a bright CL rim encompassing a pre-existing embayment. (D) Synneusis texture (crystal attachment) of three quartz crystals from the crystal-poor Deer Canyon Rhyolite from a central resurgent dome location. The crystal at the bottom of the image exhibits type 2, normal zoning. (E) Unzoned quartz crystal from the same Deer Canyon Rhyolite sample. (F) Reversely zoned quartz crystal from UBT units 3, showing a bright CL rim and resorbed dark CL core. (G) Zoned quartz grain from UBT unit 3 exhibiting (1) dark core and (3) bright rim, separated by a zone of intermediate CL intensity (2). (H) Raw counts s-1 during line scan (core-rim) of a zoned quartz crystal (I), passing through a sharp zone boundary. Note the abrupt signal increase, indicative of a ~50 ppm increase in Ti from core to rim. (J) Unzoned quartz crystal from the upper Cerro Toledo Formation. (K) Unzoned quartz crystal from the early-erupted UBT unit 1g. (L) Normally zoned quartz crystal from UBT unit 2. (M) A normally zoned (Type 2) quartz crystal from UBT unit 2 showing corresponding titanium concentrations of core and rim. 39

53 40

54 phenocrysts) of these lavas indicates that this value is generally representative of glass concentrations. Thus, Ti in glass decreases for the Cerro del Medio eruption cycle, but based on isotopic and geochemical data, its genetic relationship to the UBT is debatable. 4.4 Hornblende dacite pumices Samples of hornblende-bearing dacite pumice (HbDP) were obtained from the Tsankawi Pumice Bed and UBT unit 1g (samples 24HD1, 24HD2 in Appendix 1). The pumices are light grey, finely porphyritic and vesicular, with visible hornblende crystals. The pumices range in size from 1-10 cm, with most ~ 2 cm. Following Stimac (1996), we note the occurrence of both phenocrysts and a xenocrystic component set within a finely vesicular glassy matrix containing ~70% microlites (see Fig. 12, D). Reacted xenocrysts of alkali feldspar crystals show sieve-textured cores and a large grainsize (~3 mm) whereas the phenocrysts (plagioclase, two pyroxenes, hornblende, biotite, and Fe-Ti oxides) are generally euhedral, mm in size, and form crystal aggregates. Both major and trace element compositions of the two pumices from UBT Unit 1g agree well with published values (see Appendix 1; Self et al., 1986; Stix et al., 1988; Balsley, 1988; Stimac, 1996). All are dacitic (67-69 wt % SiO 2 ), but show appreciable variation in trace element chemistry, notably Sr, Zr and Ba (Appendix 1). 5. Discussion Using cathodoluminescence imaging of individual quartz crystals correlated to stratigraphic position, we have identified an abrupt occurrence of zoned crystals commencing within the mid-to-late UBT ignimbrite units (Fig. 6). Ti concentrations of all 41

55 unzoned quartz and the cores of zoned quartz are relative uniform (23-45 ppm), excluding the Cerro del Medio samples (60-72 ppm) (Fig. 7). Titanium concentrations of the rims of zoned quartz crystals show a ppm increase with respect to core concentration. The maximum rim concentration measured was 104 ppm Ti from a quartz crystal from the mid to late stages of the UBT eruption. Outlier core values and the range of rim concentrations demonstrate that some crystals may have either been derived from elsewhere in the chamber or were subject to multiple changes in temperature. This suggests a degree of mixing and convection, especially within the lower part of the UBT (Tshirege) magma chamber. The change from an average Ti concentration of 29 ± 5 ppm (unzoned quartz of the upper CTF, Tsankawi Pumice Bed, UBT units 1g and 2) to 72 ± 17 ppm (outer rims of UBT 3-5) indicates abrupt changes in the prevailing thermal conditions within the Bandelier system associated with the caldera-forming eruption at 1.25 Ma (Fig. 7). The abrupt appearance of reversely zoned quartz crystals beginning in UBT unit 3 (Fig. 6), and the corresponding increase from 20 to 50 ppm Ti within outer rims, suggests a temperature increase of ~100 C. This pattern is repeated throughout the remaining units of the UBT and within porphyritic samples of the Deer Canyon Rhyolite. Conversely, the crystal-poor Deer Canyon Rhyolite with unzoned quartz crystals (average concentration 30 ± 2 ppm Ti) from the central resurgent dome may be indicative of relatively hot temperatures which prevented crystallisation of high-ti rims. This discrepancy between the two Deer Canyon Rhyolite samples may reflect either thermal disequilibrium within the chamber during resurgence or the presence of residual UBT magma. The absence of quartz within the Redondo Creek Rhyodacite, as well as the microscopic textures, 42

56 mineralogy and geochemistry of this unit, preclude any simple genetic relationship to the preceding Bandelier ignimbrite (see section 5.7.2). The dominance of unzoned but relatively high-ti quartz (average concentration 64 ± 4 ppm Ti) from Cerro del Medio suggests comparatively hotter temperatures, although we are unable to derive a genetic relationship to the UBT based on Ti-in-quartz data alone (Fig. 7). 5.1 Constraining a TiO2 during the UBT eruption and subsequent resurgence Through application of the rutile saturation model (Hayden & Watson, 2007) and by rearranging the TitaniQ geothermometer (Wark & Watson, 2006), we can assess potential temperature differences induced by a changing activity of titania before, during and after the eruption of the UBT (see section 3.7). The empirical saturation model is strongly affected by temperature and less so by composition, such that an accurate temperature estimate is required as an input. The temperature estimates of Warshaw & Smith (1988) for the UBT were derived by several techniques, depending on the mineral phases present. In addition, for some UBT subunits, both a lower and an upper temperature estimate were made. In this case, the mean temperature was used as the input parameter to the saturation model. The melt composition parameter (FM) was calculated from geochemical data obtained using numerous analytical techniques (Table 2). We used whole rock data selectively, since the existence of mineral phases containing significant Ti concentrations (e.g., biotite within the Redondo Creek Rhyodacite) yielded unreliable Ti saturation concentrations. By contrast, accurate glass analysis by both electron microprobe and LA-ICP-MS yielded robust FM values. 43

57 Rearrangement of the TitaniQ geothermometer to derive a TiO2 was performed using the same temperature estimates. Average titanium-in-quartz concentrations of either unzoned quartz crystals or conversely the outermost rims of zoned quartz crystals were used as the input parameters. We assume that these regions reflect the final magmatic conditions prevalent before eruption Upper Cerro Toledo Formation The FM values for subunits of the Cerro Toledo Formation ranged from (Table 2). This derived an a TiO2 of for the early phases of CTF volcanism (unit 15-8) and for the late stages (unit 6-8) using the Hayden & Watson (2007) saturation model. The calculated activity for unit 15-8 is in good agreement with an a TiO2 of 0.43 ± 0.09 calculated by Campbell et al. (2009). Using average titanium concentrations of quartz crystals from unit 6-8, the rearranged TitaniQ geothermometer yields an a TiO2 of Early-erupted UBT (Tsankawi Pumice Bed, Units 1g, 1v, 2) For the early-erupted units of the UBT (Tsankawi, 1g, 1v, 2), the range of FM values is (Table 2). This yields activity values of for the Tsankawi Pumice Bed, for UBT unit 1g and 0.62 for UBT unit 2 using the Hayden & Watson (2007) method. Conversely, the rearranged TitaniQ geothermometer calculates a lower and more restricted a TiO2 range for the same units; 0.44 for the Tsankawi Pumice Bed, 0.27 for UBT unit 1g and 0.31 for unit 2. 44

58 Table 2. Constraining a TiO2 for the Bandelier system using the rutile saturation model (Hayden & Watson, 2007) and rearrangement of the TitaniQ titanium-in-quartz geothermometer (Wark & Watson, 2006). Whole rock, matrix glass and melt inclusion geochemical analyses were determined by various techniques (XRF, EMP and LA-ICP- MS), for units applicable to this study. 45

59 46

60 5.1.3 Mid-to-late erupted UBT (Units 3-5) The later units of the UBT (units 3-5) exhibit an increased FM range of The resulting a TiO2 values from the Hayden & Watson (2007) saturation model are for UBT unit 3, 0.57 for unit 3T, 0.79 for unit 4 and for unit 5. TitaniQ yields a TiO2 values of 0.47 for UBT unit 3, 0.55 for unit 3T, for unit 4, and for unit 5. Excluding UBT unit 4, there appears to be closer agreement for the later UBT units between a TiO2 derived by both the saturation model and the TitaniQ geothermometer Sources of error in the rutile saturation model The FM value for UBT unit 4 is derived from a bulk rock geochemical analysis (Krier et al., 1998) due to the lack of glass analyses. Thus, the relatively high a TiO2 of 0.79 may be influenced significantly by the presence of ferromagnesian phases. Warren et al. (2007) show a significant increase in the abundance of mafic phases (especially Fe-Ti oxides, magnetite and ilmenite) at higher stratigraphic positions, notably within UBT units 4 and 5. Table 2 indicates that the Ti concentration for unit 4 is ~ 1420 ppm Ti, which is ppm higher than values analyzed or determined for UBT units 3T and 5U respectively. Compared to our Ti in glass analysis, bulk rock values may not be reliable proxies for Ti concentrations in the late erupted units. Similarly, the bulk rock values of Balsley (1988) and the melt inclusion analyses of Stix & Layne (1996) for the Tsankawi Pumice Bed affect the average a TiO2 value for this unit and may warrant further examination. The error estimates on the crystallisation temperatures given in Table 3 are 47

61 calculated from the change in a TiO2 from application of the lower and upper independent temperature estimates of Warshaw & Smith (1988). For example, using the lower temperature estimate for UBT unit 3 of 1033 K, the a TiO2 value changes by 0.17 and the calculated temperature difference is thus 36 C Redondo Creek Rhyodacite Due to the lack of an independent temperature estimate for the Deer Canyon Rhyolite, we cannot calculate a TiO2 for this unit. For the Redondo Creek Rhyodacite, Phillips et al. (2007) calculated a Fe-Ti oxide crystallisation temperature of 1083 K (810 C). We analysed uncontaminated glass from this sample by several techniques, and the calculated a TiO2 according to Hayden & Watson (2007) is , showing very good agreement with an a TiO2 value of 0.54 derived using the temperature estimate and bulk rock analyses of Phillips (2004). These values are also similar to those for the UBT using the Hayden and Watson (2007) approach. In summary, when the average values are considered, the Hayden & Watson (2007) saturation model results in slightly decreasing ( ) a TiO2 throughout most of the UBT eruption sequence (from the Tsankawi Pumice Bed through unit 3T) whereas the TitaniQ geothermometer method calculates a lower and relatively constant a TiO2 of 0.41 ± 0.14 for the same period of the eruption. If we include titanium-in-quartz concentrations from UBT units 4 and 5, the average a TiO2 value increases slightly to In fact, there is little variation in a TiO2 calculated using TitaniQ between the early and late erupted units. 48

62 5.2 TitaniQ crystallisation temperatures for the UBT We have shown that the activity of titania can be effectively calculated by several methods and that over the course of the UBT eruption this value is either 1) decreasing slightly according to the rutile saturation model or 2) remaining relatively constant as shown by rearrangement of the TitaniQ geothermometer. Considering the rutile saturation model, titanium-in-quartz concentrations yield temperatures of 640 C at the end of Cerro Toledo Formation volcanism (unit 6-8) (Table 3). The Tsankawi Pumice Bed and UBT units 1g and 2 record temperatures of C. Concurrent with the abrupt occurrence of reversely zoned quartz crystals (Type 1), UBT unit 3 records a temperature increase of 673 C to 762 C. The subsequent units of the UBT (3T, 4L, 4U and 5) indicate elevated temperatures within a range of C. The highest temperatures (800 C) are recorded in the late-erupted unit 5U from an extracaldera location (Table 3). The second highest temperature (793 C) is calculated from unit 3T. Both these temperatures are evidence that the later units were tapping hotter regions of the Valles magma chamber. The high a TiO2 of 0.79 for UBT unit 4 may be an anomalous value and not representative of the melt Ti content (see section 5.1.4). Hence the temperature estimate of 730 C for unit 4L may be inaccurate (Table 3). Our TitaniQ crystallisation temperatures show good agreement with the UBT temperature estimates of Warshaw & Smith (1988) and the late CTF temperature estimates of Stix & Gorton (1990a). Temperatures recorded in early-erupted units are ~ 50 C higher than those derived from the saturation model, while the temperature increase of 750 C to 786 C from unit 2 to unit 3 is less pronounced, and the late-erupted units 49

63 record relatively uniform, elevated temperatures of 800 ± 7 C. 5.3 A thermal signature recorded in quartz? Based on our analysis, we can state with confidence that the observed compositional variations, commencing within quartz of UBT unit 3, is the result of changing thermal conditions within the Valles magma chamber, and not a result of variations in titanium activity. Multiple independent lines of evidence provide additional support for a temperature increase at higher stratigraphic levels within the UBT outflow sheets. First, a systematic increase in pre-eruptive temperatures and also oxygen fugacity is recorded by oxide-silicate and two-pyroxene geothermometry (Warshaw & Smith, 1988). The Tsankawi Pumice Bed contains titanomagnetite in the presence of fayalite, deriving a minimum crystallisation temperature of 697 C and oxygen fugacity of log UBT unit 3 contains both augite and hypersthene but a lack of fayalite, thus the two-pyroxene geothermometer calculates a mean temperature of 785 C. The upper ignimbrite unit 4 records a mean two-pyroxene temperature of 805 C, a Fe-Ti oxide temperature of 837 C and log f O Several mineralogical observations also support a temperature increase. Caress (1995, 1996) and Phillips (2004) document an increased compositional variation in rims of alkali feldspars erupted during the waning stages of the UBT (Or <28 - Or 44 ), as compared to phenocrysts sampled from the early flow units (Or ) (Fig. 9). The occurrence of syenitic crystal clots within UBT units 2 and 3, adds to the range in alkali 50

64 Table 3. Crystallisation temperatures for stratigraphic units of the Upper Bandelier Tuff, Deer Canyon Rhyolite, Redondo Creek Rhyodacite and Cerro del Medio Rhyolite. 51

65 52

66 feldspar Or content. However, considering the rhyolitic component alone, an increase in Δ Or is still apparent (Caress, 1995). The syenitic clots are thought to derive from a crystal mush zone tapped during the late stages of the UBT eruption (Caress 1995). Sanidine is the dominant feldspar within the UBT, but the emergence of prominent anorthoclase as well as plagioclase armoured by anorthoclase within UBT unit 4 is also documented (Smith & Bailey, 1966; Caress, 1995, 1996; Warren et al., 2007). Concurrent with the increase in compositional variation, there are a greater number of grains exhibiting disequilibrium (resorption) textures in the late-erupted units. Quartz exhibits a similar trend to the modal phenocryst content, increasing from 5% in the Tsankawi Pumice Bed to around 30% in unit 3, then showing a decrease in the late-erupted units 4 and 5. In comparison, the alkali feldspar content shows a steady increase as a proportion of total phenocrysts, from 60% in the Tsankawi Pumice Bed to 85% in unit 5 (Smith & Bailey, 1966; Warren et al., 2007). A decrease in the modal phenocryst content, partial resorption of quartz (and feldspar to some extent), and an increase in the alkali feldspar compositional variation are suggestive that the lower part of the chamber was interacting with a hotter magma. Our analyses of Ti in glass for several subunits of the UBT show that Ti concentrations increase upsection through the UBT and continue and continue for postcaldera units such as the Redondo Creek Rhyodacite (Table 2). These data also suggest hotter conditions at later stages of the Bandelier eruptions, as Ti is more soluble in the silicic melt at higher temperatures. Smith & Bailey (1966) showed that the degree of welding within successive ignimbrite units was related to emplacement temperatures. The Tsankawi Pumice Bed and 53

67 Figure 9. Ternary diagram (Ab = albite, An = anorthite, Or = orthoclase) showing feldspar compositions in the Upper Bandelier Tuff. The early UBT samples include feldspars from the Tsankawi Pumice Bed and early-erupted ignimbrite units 1g, 1v and 2. The late UBT samples are feldspars from units 4 and 5. The 750 C (at 100 MPa) and 900 C (at 50 MPa) isotherms are plotted from Deer et al. (1992). Feldspar compositions are from Phillips (2004). 54

68 55

69 early ignimbrite units grade from unwelded and crystal-poor to partially welded and vapour-phase altered. UBT unit 3T generally exhibits the highest degrees of welding outside the caldera. The late ignimbrite units also show similar horizons of dense welding, especially vitrophyric horizons prominent within intracaldera unit 5U. Collectively, this upsection increase in the degree of welding cannot be explained by an ignimbrite emplaced at uniformly low temperatures (Smith & Bailey, 1966). A minor population of fine-grained, hornblende dacite pumices (67-69 wt. % SiO 2 ) are present within the Tsankawi Pumice Bed and interspersed throughout the lower UBT. Importantly, there is no evidence for this pumice type within the immediately underlying tephra units of the Cerro Toledo Formation. Therefore the dacite pumices are suggestive of new magma entering the Bandelier system within a short period prior to eruption. A number of studies have shown that inputs of more mafic magma induce a temperature increase within a chamber containing silicic magma (e.g., Snyder, 2000; Wiebe et al., 2007; Wark et al., 2007; Campbell et al., 2009). Several of the Cerro del Medio eruptive units are mapped as obsidian flows, with flow unit 4 comprising roughly half of the total 5 km 3 volume of the complex (Gardner et al., 2007). All of the flow units have <5 % phenocrysts, mostly sanidine. Both hornblende and embayed quartz crystals are rare, as is plagioclase. Ti-in-quartz analysis indicates the Cerro del Medio lavas to be relatively hot (~ 800 C) compared to the early UBT (Fig. 7). If the Cerro del Medio lavas are related to the UBT, they represent a significant volume of hotter, crystal-poor rhyolite from the Valles magma chamber. Various lines of evidence support the hypothesis that the UBT demonstrates evidence for a pronounced thermal gradient within the magma chamber prior to eruption 56

70 (e.g. Hildreth, 1981; Wark et al., 2007). A key question is whether the thermal gradient represents tapping of incrementally deeper and hotter levels of the chamber or does it reflect input of hot new magma, or both? We use the quartz zoning patterns to answer this question. Firstly, rims on zoned quartz crystals are invariably high-temperature. Secondly, core regions are ubiquitously rounded. These two observations are highly suggestive of interaction with hot, new magma entering into the chamber. The timing of injection, and origin of this magma is discussed in the following sections. 5.4 Relative timing of magmatic recharge A key question is whether the quartz zoning is a reflection of a relatively short or protracted magmatic recharge event or events. We have multiple lines of evidence which suggest that this event occurred relatively soon before eruption of the UBT. The lack of hornblende dacite pumices within the preceding Cerro Toledo Formation suggests that the parental material was incorporated into the chamber within a geologically short period following termination of CTF volcanism, just prior to eruption of the UBT (e.g. Self et al., 1986; Stimac, 1996; Warren et al., 2007). Petrographic analysis of the pumices reveals both cogenetic phenocrysts and a component of reacted xenocrysts of alkali feldspar and resorbed quartz derived from the UBT, which suggests that this parental magma mingled with the rhyolitic UBT (Stimac, 1996). In addition, the abundance of plagioclase and hornblende within the Tsankawi Pumice Bed and UBT unit 1g suggests that these crystals are derived from mixing of two distinct magmas prior to and during the early phases of eruption (Warren et al., 2007). Within some UBT rhyolitic 57

71 pumices, it is common to find a trace component of hornblende, plagioclase and pyroxene crystals with chemistry similar to that of the hornblende dacite pumices. Some of these minerals are themselves rimmed by a fine-grain matrix, suggesting they were derived from disaggregation of the hornblende dacite pumice (Stimac, 1996). The outer bright CL zones of quartz crystals from UBT units 3-5 record temperatures ~100 C hotter than at the start of the caldera-forming eruption. Concurrent with the occurrence of zoned quartz crystals, the boundaries between core and rim are often sharply defined (Fig. 8) with Type 1 reverse zoning dominant (Fig. 5). The resulting increase in the titanium concentration occurs over a distance <50 μm (Fig. 8, H). The ubiquitous rounding of the core regions of all quartz crystals suggests an event that affected a significant proportion of the Bandelier magma chamber, leading to widespread resorption of quartz crystals. This evidence suggests that the temperature of the Bandelier chamber was raised significantly to promote resorption, with subsequent growth of hightemperature outer rims on quartz from the later-erupted ignimbrite units. We can use the known diffusivity of titanium within quartz at a given temperature to assess the relative growth rates of these outer zones. The diffusivity of Ti (D Ti ) is given by the Arrhenius relation: (10) D Ti = exp 273± 12 kj mol 1 / RT m 2 s 1 where R is the universal gas constant and T the absolute temperature. From this equation, we can determine the time elapsed since the onset of diffusion (D) over a distance (d), given by the relation: (11) t = d 2 /4D 58

72 In this case the thickness of the zone boundary is representative of d. For example, the calculated average boundary thickness within quartz from UBT unit 3 is 28.5 μm. Based on the temperature estimates of Warshaw & Smith (1988), a mean temperature of 1058 K (785 C) is calculated for unit 3 (Table 3). Using a range of Ti diffusion rates and the calculated zone boundary widths, the time for the observed diffusion within the quartz crystals used for this study ranges from 1080 to 7340 years (Table 4). This is similar to the ~ 10 4 year duration of a recharge event occurring before eruption of the Otowi member of the Bandelier Tuff, calculated using Sr diffusion in sanidine (Hervig & Dunbar, 1992). Thus, the temperature increase required for subsequent crystallisation of high-ti rims on quartz may have occurred a few thousand years prior to eruption. 5.5 Nature of the injected magma Based on field exposures, the dacite pumices comprise <1 vol. % of the UBT (Smith & Bailey, 1966; Smith et al., 1970; Balsley, 1988; Caress, 1995; Stimac, 1996), hence <4 km 3 of this more mafic magma was erupted, presumably having resided at the roof region of the reservoir for some period before eruption as it is found primarily within the Tsankawi Pumice Bed. Smith (1979) envisaged that the Bandelier chamber (which contained both the Otowi and Tshirege, or Upper Bandelier Tuff) was underlain by a substantial volume of melt, rhyodacitic in composition, which was in turn underlain by a deeper basaltic zone. With regard to the hornblende dacite pumices, several possibilities exist as to their ultimate origin. Firstly, they may simply be a sample of a dacite source that underwent little or no interaction with the UBT, although mineral exchange likely 59

73 Table 4. Titanium diffusion in quartz and timescales of outer zone growth for Unit 3 of the UBT. 60

74 61

75 occurred (Gardner et al., 1986; Stimac, 1996). Secondly, the hornblende dacite pumice could be a hybrid composition resulting from intrusion and mingling of a more mafic magma into a rhyolitic chamber. There may be some evidence for a basaltic source in the Mg-rich pyroxenes and the most calcic zones of some plagioclase crystals. However, textural and mineralogical evidence suggests that the hornblende dacite magma existed as a discrete layer within the Bandelier chamber shortly before eruption (Stimac, 1996). There was a degree of mixing and mingling between the two magmas, and vesiculation by partial quenching of the hotter dacite probably caused a small volume of this material to rise into the roof regions of the chamber, where it was erupted synchronously with the Tsankawi Pumice Bed (Stimac, 1996). Several authors have remarked upon the close geochemical similarity of the hornblende dacite pumices to Cerro Rubio dacite, erupted during late Tschicoma volcanism at Ma (Fig. 2; Heiken et al., 1986; Gardner et al., 1986; Broxton et al., 2007). Based on major elements, there is close geochemical agreement between the two magma compositions (Fig. 10, F, G; Appendix 1). However, the large variation of trace elements precludes any simple derivation and warrants further investigation. The close spatial relationship between the Bandelier magma chamber and Cerro Rubio raises the possibility of a volume of Tschicoma dacite being injected into the Bandelier system and forming the hornblende dacite pumices. It is clear that the dacite pumices do not share any clear geochemical relationship to the Upper Bandelier Tuff. as there is poor agreement between the trace element geochemistry of the hornblende dacite and the UBT (Fig. 10). It may be possible that magmatic recharge was not limited to periods before 62

76 Figure 10. Trace element and major element plots showing relationships among the Upper Bandelier Tuff (UBT), Deer Canyon Rhyolite (DC), Redondo Creek Rhyodacite (RC), Cerro del Medio Rhyolite (CdM), Hornblende dacite pumices (HbDP) and Cerro Rubio dacite (CR). The Bandelier geochemical trend progresses from the early-erupted Tsankawi pumice bed (T) to the late upper ignimbrite units (UBT 4&5). Each subunit is represented by a separate symbol, indicated on the legend. The reader is referred to Appendix 1 for a comprehensive geochemical table. 63

77 64

78 65

79 eruption, and may have continued throughout the period examined in this study. Warren et. al (1997, 2007) showed that anorthoclase-rich samples of UBT unit 4 have interesting geochemistry. Notable spikes in Ti, Sr, Ba, and Zr within this unit depart from their individual increasing trends (see Table 1 in Warren et al., 2007). An injection of more primitive magma may have created this geochemical trend. Thus, the prominent anorthoclase, as well as plagioclase armoured by anorthoclase within UBT unit 4 could be the product of mixing between rhyolite and a more primitive magma, distinct from the hornblende dacite injected before the UBT eruption (Warren et al., 2007). 5.6 Pre-eruptive thermal regime of the UBT The abrupt appearance and occurrence of zoned quartz crystals within the mid-tolate ignimbrite units is intriguing. Extensive mapping of the UBT outflow sheets has identified that units 1g, 1v and 2 comprise ~ 80-90% of the entire volume of the 400 km 3 ignimbrite sequence (F. Goff, pers. comm.). Our complementary titanium-in-quartz measurements suggest that a large volume of the ignimbrite was thus erupted from a portion of the Bandelier chamber which had been at a relatively uniform temperature of C (Table 3). However, given that a significant proportion of these quartz grains show evidence of partial resorption, these temperatures hence are reflective of conditions prevalent before the heating event commenced. The occurrence of zoned quartz crystals within units 3-5 suggests that the deeper portion of the chamber was periodically subject to thermal disequilibrium, and the high-temperature outer rims are evidence of growth from a hotter magma. 66

80 Based on the geochemistry of alkali feldspar phenocrysts, the model of Caress (1995, 1996) proposes that the Bandelier magma chamber was composed of two main horizons. The upper portion of the chamber contained the magma that erupted as the Tsankawi Pumice Bed and ignimbrite units 1 and 2 of the UBT, representing collectively over 80 % of the erupted volume. Within the upper portion of the magma chamber, unzoned sanidine with Or was the dominant alkali feldspar. The lower level of the chamber is represented by the late-erupted units 3-5, and the increased geochemical variation in alkali feldspar chemistry (Or <28 - Or 44 ) is a reflection of a more dynamic, convecting region. Appendix 5 details a preliminary assessment of CL zoning and geochemistry of feldspar from the UBT and Deer Canyon Rhyolite. Results indicates that feldspar phenocrysts from the early-erupted units are invariably unzoned, both under CL and geochemically (Or ). The appearance of dominantly zoned alkali feldspar, corresponding with quartz, occurs within unit 3 of the UBT. The rims of alkali feldspar from unit 3 show an increased compositional range of Or (Appendix 5). 5.7 Resurgence: origin of the post-ubt magmas Deer Canyon Rhyolite Following caldera collapse, the Bandelier magma chamber was in a state of disequilibrium. Our Ti-in-quartz results from the Deer Canyon Rhyolite suggest that these lavas probably originated from distinct regions of the chamber following collapse. Firstly, there is marked contrast in zoning patterns of both quartz and feldspar between lavas erupted on the western resurgent dome (sample #13, porphyritic with zoned quartz and 67

81 Fig. 11. Mosaic of CL images of feldspar crystals from the hornblende dacite pumice, units of the UBT and resurgence-related lavas. The dashed horizontal line denotes the occurrence of zoned quartz crystals. This commences in UBT unit 3 and continues through units 4 and 5 and the porphyritic Deer Canyon Rhyolite. (A-C) Sieve-texture plagioclase cores mantled by sanidine overgrowths (anti-rapakivi texture) from the Redondo Creek Rhyodacite. (D, E) Sanidine crystals from the porphyritic Deer Canyon Rhyolite. (F) Three feldspars exhibiting synneusis texture from the crystal-poor Deer Canyon Rhyolite. (G, H) Sanidine phenocrysts from unit 3 of the UBT, showing patchy and concentric zoning patterns. (I) A sieve-textured plagioclase core with an overgrowth rim of alkali feldspar from UBT unit 4. (J) Complex, concentrically-zoned sanidine from UBT unit 5. (K) Plagioclase phenocryst from a hornblende dacite pumice from UBT unit 1g. (L) Unzoned sanidine crystal from the Tsankawi Pumice Bed. (M) Zoned plagioclase crystal surrounded by a thin rim of alkali feldspar from UBT Unit 1g. (N) Unzoned sanidine crystal from UBT unit 2. 68

82 69

83 feldspar) and those from central dome locations (sample #15, less-abundant, smaller, unzoned quartz and feldspar). In fact, the Deer Canyon Rhyolite occurs as a fine-grained, crystal-poor lava on the eastern resurgent dome, contrasting with its extremely porphyritic nature within western regions (Goff et al., 2007). Gibler (2004) shows that the easternmost flow lobes are relatively crystal-poor (2-6 % phenocrysts), containing little to no quartz. However, no relationship between major and trace element chemistry, and spatial location of the Deer Canyon Rhyolite on the resurgent dome is documented. A pertinent question is whether the Deer Canyon eruptions sampled two distinct reservoirs, as suggested by Goff et al. (2007) and Gibler (2004), or if the heterogeneity is the result of new magma intruding locally into a single chamber. A two-reservoir hypothesis is difficult to reconcile with the similar mineralogy and geochemistry of the Deer Canyon Rhyolite and the later-erupted UBT (Fig. 10) (Spell et al., 1993; Phillips, 2004). The Deer Canyon lavas appear to be depleted in Ba, Sr and Zr relative to the lateerupted UBT unit 4 and 5, although most samples have undergone some degree of alteration (Fig. 10; Phillips, 2004; Goff et al., 2007). The titanium concentrations of 1) cores of zoned quartz from the porphyritic rhyolite and 2) unzoned quartz from the crystal-poor rhyolite, are similar to core concentrations of UBT quartz (Fig. 7). This suggests that all these quartz crystals were subjected to similar thermal conditions. To generate the heterogeneity in quartz zoning and crystal content, a possible mechanism may involve intrusion of new magma into a residual crystal mush, of a composition similar to the last erupted unit of the UBT. Intrusion may have been localised, leading to extensive quartz resorption in certain areas and thus creating the crystal-poor or aphyric nature of the central and eastern lavas. 70

84 Detailed intracaldera drilling has revealed that caldera collapse was both piecemeal and trapdoor in nature, with maximum subsidence to the east where the intracaldera UBT is thickest (Goff, 1983; Heiken et al., 1986; Self et al., 1986). The crystal-poor nature of the Deer Canyon lavas in the east may therefore be related to the increased subsidence of the caldera roof blocks, which probably sank into the crystal mush, inducing convective mixing. The occurrence of synneusis texture in the eastern lavas (sample #15) suggests a low-energy environment of sufficient fluidity to permit crystal entrainment (Vance, 1969; Vernon, 2004). A plausible process for development of this texture is expulsion of crystal-poor melt from a mush zone which had previously been heated as suggested by the occurrence of widespread resorption textures (Gibler, 2004; see Appendix 5, Fig. A5.3). Several studies have shown that, under application of downward pressure, it is feasible to expel significant volumes of crystal-poor rhyolite and facilitate crystal attachment, from a mush zone of rhyodacitic composition (Bachmann & Bergantz, 2004; Girard & Stix, 2009). This process may have been important during formation of the crystal-poor Deer Canyon Rhyolite. The porphyritic Deer Canyon Rhyolite erupted onto the western resurgent dome may be derived from more crystal-rich horizons in the residual Bandelier magma chamber. Conditions favoured crystal growth, and temperatures were probably similar to the end of the UBT eruption, since there is close agreement in the values and the range of Ti concentrations of rims of quartz of UBT units 3-5 and the porphyritic Deer Canyon (Fig. 7). In summary, the mineralogical diversity of the Deer Canyon Rhyolite may be a product of several processes: 1) interaction with new, hot, primitive magma; 2) variable 71

85 caldera collapse; 3) expulsion of small-volume magma batches from a crystal mush Redondo Creek Rhyodacite The processes driving resurgence were first discussed in the seminal work by Smith & Bailey (1968), one scenario involving new input of magma at depth. The occurrence of the hornblende dacite pumices within the initial eruptive products of the UBT could indicate that resurgence may be a response to injection of magma of similar chemistry. The mineralogy exhibited by the Redondo Creek Rhyodacite is unusual and distinct from that of the UBT and the Deer Canyon Rhyolite. The lack of quartz, as well as the presence of anti-rapikivi feldspars (sieve-textured plagioclase cores with sanidine rims), and abundant biotite and Fe-Ti oxides, suggests a complex relationship (Fig. 12). Smith & Bailey (1968) hypothesised that the Redondo Creek Rhyodacite represents the product of chamber overturn following withdrawal and eruption of the UBT. This unit shows transitional trace element chemistry between that of the UBT and the hornblende dacite pumice (Fig 10). Based on this evidence, a key question is whether the Redondo Creek Rhyodacite represents mixing of two distinct sources or a new input of relatively primitive magma. The first scenario is supported by our geochemical plots, where the Redondo Creek Rhyodacite could be derived from mixing of residual UBT and renewed input of Tschicoma-like, hornblende dacite. However, several lines of reasoning complicate this process. First, the composition of the UBT parental magma is unclear, as the scatter of incompatible elements within the late-erupted UBT units suggests that these also may represent a partially hybridised magma (Fig. 10). Stimac (1996) suggests that the hornblende dacite pumice is the product of mixing between a more mafic member 72

86 Figure 12. Photomicrographs of minerals and textures seen in the Redondo Creek Rhyodacite and hornblende dacite pumice of the UBT. (A) biotite (Bt), clinopyroxene (CPx) and plagioclase (Plg) within a glassy matrix, (B) a sieve-textured plagioclase core with sanidine rim overgrowth, (C) a sieve textured plagioclase crystal, (D) a hornblende dacite pumice from the plinian Tsankawi Pumice Bed of the UBT. The left of the image shows part of a 1 cm sieve-textured K-feldspar. This crystal is set within a porphyritic, finely vesicular glassy matrix.(a) and (D) are under plane-polarised light, (B) and (C) are under crossed-polarised light. Scale bars are 0.5 mm. 73

87 74

88 (possibly andesitic) and rhyolite of composition similar to UBT units 2 and 3 which exhibit less geochemical heterogeneity than do later ignimbrite units. To generate a composition similar to the Redondo Creek Rhyodacite, a simple mixing model between the hornblende dacite pumice and rhyolite of UBT units 2 and 3 composition generally works for Ba but not for Sr. This discrepancy is also indicated by Sr vs Nb and Ba vs Nb plots (Fig. 13). In the case of Sr, to derive the Redondo Creek Rhyodacite, the model requires 80 % incorporation of UBT rhyolite. With Ba, however, only 40 % rhyolite is required. The UBT endmember may in fact be a more primitive composition, similar to Deer Canyon Rhyolite, which would improve some but not all linear mixing trends. Thus, simple mixing of hornblende dacite and UBT rhyolite may not be the controlling factor in the formation of the Redondo Creek Rhyodacite. An alternative scenario is that the Redondo Creek Rhyodacite itself represents a distinct magma Cerro del Medio Rhyolite It appears unlikely that the Cerro del Medio Rhyolite is sourced from the Bandelier chamber following eruption of the UBT and resurgence-related units, based on geochemical and isotopic data (Fig. 9.; Spell et al., 1993 and references therein). There may be several ways in which the Cerro del Medio lavas were generated. Our titanium data from unzoned quartz crystals of the del Medio complex indicates that crystallisation temperatures were ~ 100 C hotter than those prevailing before eruption of the UBT. Gardner et al. (2007) estimate the cumulative volume of Cerro del Medio to be < 5 km 3, and thus the underlying magma chamber (one distinct from the Bandelier) may have been relatively small, perhaps on the order of km 3. Based on our CL imaging 75

89 Figure 13. Simple mixing models of (A) Sr-Nb and (B) Ba-Nb between hornblende dacite pumices (HbDP) and subunits 2, 3 and 5 of the UBT to form the Redondo Creek Rhyodacite (RC). The most primitive HbDP (P) was used as one endmember, and the most evolved sample of UBT 2, 3 and 5, respectively, as the other endmember. Red numbers indicate the percentage incorporation of UBT rhyolite. 76

90 77

91 and Ti-in-quartz data, this chamber was possibly isothermal (> 800 C). The unzoned but hot nature of quartz crystals may be indicative of a relatively short residence time in a hot magma, which did not permit the development of zoning in quartz. Alternatively, the Cerro del Medio lavas may be products of remelting of a crystal mush at depth (Gibler, 2004). This mush zone may represent residual UBT, or perhaps a hybrid derived from mixing of new magmas during resurgence. The latter scenario could explain the isotopic trends identified by Spell et al. (1993). Evidence is so far unsupportive of a direct relationship to the UBT, although many studies have illustrated that reactivation of a crystal mush is an important mechanism in post-caldera volcanism (e.g. Hildreth & Wilson, 2007; Bachmann & Bergantz, 2008; Girard & Stix, 2009, 2010). 6. Conclusions Based on the results of this work and a synthesis of other studies, we propose the following four-stage model for the thermal evolution of the Bandelier magmatic system from Ma to ~ 1.11 Ma (Fig. 14). This ~ 150 ka period encompasses eruption of the Tshirege (Upper) Member of the Bandelier Tuff and collapse of the Valles caldera, resurgence of the caldera and associated volcanism, and eruption of the initial ring fracture rhyolite into the caldera moat following termination of resurgence. By application of the TitaniQ titanium-in-quartz geothermometer (Wark & Watson, 2006), combined with cathodoluminescence imaging, we have discovered a magmatic recharge event prior to eruption of the Upper Bandelier Tuff (UBT). Furthermore, recharge probably occurred throughout the UBT eruption and during resurgence of the caldera, 78

92 either as a continual process or as a series of discrete events. This is manifested by the compositional heterogeneity of late UBT units as well as lavas erupted during resurgence, in particular the Redondo Creek Rhyodacite. 1) Although the ultimate source of the recharging magma is debatable, this hotter, more primitive magma appears to be dacitic in composition and underwent a limited amount of mixing and mingling with rhyolite magma of the UBT, forming a discrete layer in the chamber (Fig. 14, Stage 1, A). Recharge of the Bandelier system commenced < 10,000 years before eruption. 2) The heat provided by this magma led to widespread crystal resorption. With regard to quartz, this resorption event left rounded cores of low titanium concentration. The intruding hornblende dacite magma vesiculated and was partially quenched, forming a magmatic foam which subsequently rose into the roof region of the Bandelier chamber (Fig. 14, Stage 1, B). This process probably induced overpressure in the chamber and triggered the UBT eruption (Fig. 14, Stage 1, C). 3) The cataclysmic eruption commenced with a plinian eruption depositing the Tsankawi Pumice Bed along with the bulk of the hornblende dacite pumice. Similar to the Tsankawi, the subsequent ignimbrite units 1g, 1v and 2 of the UBT contain predominantly rounded and unzoned quartz crystals of low titanium concentration. These units comprise % of the total 400 km 3 erupted volume. This suggests that the UBT magma chamber was largely static and isothermal, at temperatures of C, prior to its heating from intrusion of the recharging magma. 79

93 4) The lower portion of the chamber, containing the late-erupted UBT ignimbrite units 3-5, was affected by continual intrusion of hotter, more primitive magma. High-temperature rims on quartz indicate that these crystals were subject to temperatures at least ~ 100 C hotter than those recorded by core Ti concentrations. In fact, temperatures must have been raised significantly to induce resorption of the core regions before rim growth. Convection induced by hot recharging magma and caldera collapse led to development of multiple growth zones on quartz and compositional heterogeneity in alkali feldspar within the lateerupted UBT units (Fig. 14, Stage 2). 5) Following the UBT eruptions, the chamber was in a state of temperature/pressure disequilibrium and potentially underpressured. New magma may have been intruded into a residual crystal mush of a composition similar to the late-erupted UBT, as manifested by the Deer Canyon Rhyolite, which shows thermal heterogeneity, based on Ti-in-quartz data. Increased subsidence of the caldera floor to the east may have led to greater interaction of the intruding magma with the crystal mush, leading to its partial melting and eruption of crystal-poor Deer Canyon Rhyolite onto the central and eastern resurgent dome (Fig. 14, Stage 3, A). Conversely, the western porphyritic lavas exhibit complexly zoned quartz crystals and may have been derived from a cooler or more crystal-rich portion of the chamber that underwent less interaction with replenishing magma (Fig. 14, Stage 3, B). 6) The Redondo Creek Rhyodacite represents the most primitive silicic material associated with the upper Bandelier eruptions. We theorize that intrusion of this 80

94 magma into the base of the residual chamber caused resurgence. Independent temperature estimates show that this magma was hot (810 C) and may have partially melted the residual UBT crystal mush (Fig. 14, Stage 3, C). 7) The Cerro del Medio Rhyolite quartz crystals are unzoned and exhibit elevated titanium concentrations (> 40 ppm) in comparison to cores of UBT and Deer Canyon Rhyolite quartz. This suggests elevated temperatures but isothermal conditions. The Cerro del Medio complex shows little geochemical and isotopic affinity to the UBT but may have been erupted as a series of small volume, crystal-poor to aphyric rhyolite batches from the UBT crystal mush (Fig. 14, Stage 4). 81

95 Fig. 14. Four-stage thermal evolution model of the Bandelier magmatic system during collapse and resurgence of the Valles caldera. Stage 1: Magmatic recharge commences < 10,000 years prior to eruption of the Upper Bandelier Tuff, initially heating the entire chamber and causing widespread crystal resorption. Following this, the Bandelier chamber is composed of a voluminous upper and largely static layer at isothermal conditions, and a lower convecting layer experiencing thermal disruption. A) Intrusion of relatively mafic magma into the Bandelier chamber forms a discrete horizon of hybrid composition. This undergoes a limited amount of mingling and mixing with the surrounding rhyolitic magma of composition similar to units 1 and 2 of the UBT. B) Partial quenching of the hybrid magma forms the hornblende dacite magma. Vesiculation induced by this process provides buoyancy to the hornblende dacite, which rises into the roof region of the Bandelier chamber, residing with rhyolitic magma with a composition of the Tsankawi Pumice Bed. C) The release of volatiles through partial quenching of the hornblende dacite leads to overpressure in the chamber and eruption of the UBT initially from a central vent. Stage 2: Caldera collapse and formation of the Valles caldera along ring faults. Maximum subsidence of roof blocks is to the east, forming a piecemeal, trapdoor-style collapse and thick deposits of intracaldera ignimbrite. The late-erupted UBT units 3-5 contain quartz crystals with complex zoning and high-temperature rims. These are the 82

96 product of hotter temperatures and convective mixing induced by the subsiding roof blocks. New magma may have been drawn into the chamber during caldera collapse, creating the geochemical heterogeneity in late-erupted units. Stage 3: Resurgence of the Valles caldera. Maximum resurgence of ~1000 m is confined to the central area of the caldera. The thermal heterogeneity exhibited by the Deer Canyon Rhyolite is a product of differing degrees of interaction between a replenishing magma and a residual UBT crystal mush (A, B). The crystal-poor to aphyric lavas on the eastern resurgent dome may have been derived from a partially-melted crystal mush, whereas the porphyritic lavas in the west were sourced from a cooler part of the chamber. The Redondo Creek Rhyodacite may represent a hybrid mix of replenishing magma and residual UBT (C), or a distinct magma in itself (D). A temperature gradient for the Bandelier magma chamber is indeterminable during caldera resurgence, thus we have indicated a lighter colour for the magma, only to distinguish it from the crystal mush zone. Stage 4: Ring fracture volcanism following termination of resurgence. The foci of volcanism in the Valles caldera shifted to the easternmost parts of the caldera where the Cerro del Medio Rhyolite complex was erupted. The sparsely porphyritic to aphyric nature of these lavas suggests they were sourced from hot, small-volume rhyolite batches expelled from the crystal mush. 83

97 84

98 85

99 86

100 87

101 References Acocella, V. (2010). Evaluating fracture patterns within a resurgent caldera: Campi Flegrei, Italy. Bulletin of Volcanology. 72, Aldrich Jr, M. J., Chapin, C. E., and Laughlin, A. W. (1986). Stress History and Tectonic Development of the Rio Grande Rift, New Mexico. Journal of Geophysical Research. 91, Annen, C. (2009). From plutons to magma chambers: Thermal constraints on the accumulation of eruptible silicic magma in the upper crust. Earth and Planetary Science Letters. 284, Bachmann, O., and Bergantz, G. W. (2004). On the origin of crystal-poor rhyolites: extracted from batholithic crystal mushes. Journal of Petrology. 45, Bachmann, O., and Bergantz, G.W. (2008). Deciphering magma chamber dynamics from styles of compositional zoning in large silicic ash flow sheets. Reviews in Mineralogy & Geochemistry. 69, Bailey, R.A., Smith, R.L., and Ross, C.S. (1969). Stratigraphic nomenclature of volcanic rocks in the Jemez Mountains, New Mexico. U.S. Geol. Surv. Bull P,

102 Balsley, S.D. (1988). The petrology and geochemistry of the Tshirege Member of the Bandelier Tuff, Jemez Mountains Volcanic Field, New Mexico, U.S.A. M. Sc. thesis, University of Texas at Arlington. 188 pp. Broxton, D. E., and Rogers, M. A. (2007). Comparison of two systems of nomenclature for the Tshirege Member, Bandelier Tuff, Central Pajarito Plateau, New Mexico. In Kues, B.S., Kelley, S.A., and Lueth, V.W., eds. Geology of the Jemez Region II. New Mexico Geological Society 58 th Annual Field Conference Guidebook. Socorro, New Mexico Geological Society, p Broxton, D., Woldegabriel, G., Peters, L., Budahn, J., and Luedemann, G. (2007). Pliocene volcanic rocks of the Tschicoma formation, east-central Jemez Volcanic Field: chemistry, petrography, and age constraints. In Kues, B.S., Kelley, S.A., and Lueth, V.W., eds. Geology of the Jemez Region II. New Mexico Geological Society 58 th Annual Field Conference Guidebook. Socorro, New Mexico Geological Society, p Broxton, D.E., Heiken, G.H., Chipera, S.J., and Byers, F.M.Jr. (1995a). Stratigraphy, petrography and mineralogy of Bandelier Tuff and Cerro Toledo deposits. In Broxton, D.E., and Eller, P.G., (eds). (1995). Earth Science Investigations for Environmental Restoration - Los Alamos National Laboratory Technical Area 21. LANL Report LA MS. 118 p. 89

103 Broxton, D.E., Vaniman, D., Byers, F.M. Jr., Chipera, S.J., Kluk, E.C., and Warren, R.G. (1995b). Stratigraphy, mineralogy and chemistry of bedrock tuffs at Pajarito Mesa. In Broxton, D.E., Eller, P.G. (eds). (1995). Geological site characterization for the proposed mixed waste disposal facility, Los Alamos National Laboratory. LANL Report LA MS. Campbell, M. E., Hanson, J. B., Minarik, W. G., and Stix, J. (2009). Thermal History of the Bandelier Magmatic System: Evidence for Magmatic Injection and Recharge at 1.61 Ma as Revealed by Cathodoluminescence and Titanium Geothermometry. Journal of Geology. 117 (5), Caress, M.E. (1995). Alkali feldspars in the Tshirege Member of the Bandelier Tuff: Systematic vertical and lateral distribution of feldspar compositions and their implications. Ph. D. thesis, University of California, Santa Barbara. 151 pp. Caress, M.E. (1996). Zonation of alkali feldspar compositions in the Tshirege Member of the Bandelier Tuff in Pueblo Canyon, near Los Alamos, New Mexico: New Mexico Geological Society, 47th Field Conference, Guidebook, p Cherniak, D.J., Watson, E.B., and Wark, D.A. (2006). Ti diffusion in quartz. Chemical Geology. 236,

104 Cole, J.W., Milner, D.M., and Spinks, K.D. (2005). Calderas and caldera structures: a review. Earth Science Reviews. 69, Deer, W.A., Howie, R.A., Zussman, J. (1992). An introduction to the rock-forming minerals. Second edition. Longman Scientific & Technical. Harlow, England. 696 pp. Dennen, W. H. (1964). Impurities in quartz. Geological Society of America Bulletin. 75, Dennen, W. H. (1966). Stoichiometric substitution in natural quartz. Geochimica et Cosmochimica Acta. 30, Dethier, D.P., Kampf, S.K., Sawyer, D.A., and Budahn, J.R. (2007). Chemical zonation in the Upper Bandelier Tuff: evidence from the chemistry of distal outcrops, Puye quadrangle. In Kues, B.S., Kelley, S.A., and Lueth, V.W., eds. Geology of the Jemez Region II. New Mexico Geological Society 58 th Annual Field Conference Guidebook. Socorro, New Mexico Geological Society, p Downie, N.M, and Heath, R.W. (1970). Basic statistical methods (3 rd edition). Harper & Row, New York. 356 pp. 91

105 Gardner, J.N., Goff, F., Garcia, S., and Hagan, R.C. (1986). Stratigraphic relations and lithologic variations in the Jemez volcanic field, New Mexico. Journal of Geophysical Research. 91, Gardner, J., and Goff, F. (1996). Geology of the northern Valles caldera and Toledo embayment, New Mexico. In Goff, F., Kues, B., Rogers, M., McFadden, L., and Gardner, G., eds. The Jemez Mountains Region. N.M. Geol. Soc. 47th Annual Field Conference, p Gardner, J. N., Goff, F., Kelley, S., and Jacobs, E. (2010). Rhyolites and associated eposits of the Valles-Toledo caldera complex. New Mexico Geology, 32, no. 1. Gardner, J.N., Sandoval, M.M., Goff, F., Phillips, E., and Dickens, A. (2007). Geology of the Cerro del Medio rhyolite center, Valles caldera, New Mexico. In Kues, B.S., Kelley, S.A., and Lueth, V.W., eds. Geology of the Jemez Region II. New Mexico Geological Society 58 th Annual Field Conference Guidebook. Socorro, New Mexico Geological Society, p Gibler, K. (2004). Postcollapse volcanism in the Valles caldera, New Mexico: the transition from large volume explosive to small volume effusive eruptions. M. Sc. thesis. University of Nevada, Las Vegas. 195 p. 92

106 Ginibre, C., Wörner, G., and Kronz, A. (2007). Crystal zoning as an archive for magma evolution. Elements. 3, Girard, G., and Stix, J. (2010). Rapid extraction of discrete magma batches from a large differentiating magma chamber; the Central Plateau Member rhyolites, Yellowstone Caldera, Wyoming. Contributions to Mineralogy & Petrology. 160, Girard, G., and Stix, J. (2009). Buoyant replenishment in silicic magma reservoirs: Experimental approach and implications for magma dynamics, crystal mush remobilization, and eruption. Journal of Geophysical Research. 114, B doi: /2008jb Gladney, E.S., and Bower, N.W. (1985). Determination of elemental composition of NBS 278 and NBS 688 via neutron activation and x-ray fluorescence. Geostandards Newsletter. 9, no. 2, Goff, F. (1983). Subsurface structure of Valles Caldera: A resurgent cauldron in northern New Mexico. Geological Society of America Abstract Programs. 15, 381. Goff, F. (2010). The Valles Caldera: New Mexico's supervolcano. New Mexico Earth Matters. 10 (1). 93

107 Goff, F., and Gardner, J. N. (2004). Late Cenozoic geochronology of volcanism and mineralization in the Jemez Mountains and Valles caldera, north central New Mexico. In Mack, G. H., and Giles, K. A., eds. The Geology of New Mexico A Geologic History. New Mexico Geological Society Special Volume 11. Socorro, New Mexico Geological Society, p Goff, F., Warren, R.G., Goff, C.J.,Whiteis, J., Kluk, E., and Counce, D. (2007). Comments on the geology, petrography, and chemistry of rocks within the resurgent dome area, Valles caldera, New Mexico. In Kues, B.S., Kelley, S.A., and Lueth, V.W., eds. Geology of the Jemez Region II. New Mexico Geological Society 58 th Annual Field Conference Guidebook. Socorro, New Mexico Geological Society, p Götze, J., Plötze, M., and Habermann, D. (2001). Origin, spectral characteristics and practical applications of the cathodoluminescence (CL) of quartz a review. Mineralogy and Petrology. 71, Griggs, R. L. (1964). Geology and groundwater resources of the Los Alamos area, New Mexico: U. S. Geological Survey Water-Supply Paper 1735, 107 p. Hayden, L.A., and Watson, E. B. (2007). Rutile saturation in hydrous siliceous melts and its bearing on Ti-thermometry of quartz and zircon. Earth and Planetary Science Letters. 258,

108 Heiken, G., Goff, F., Gardner, J. N., and Baldridge, W. S. (1990). The Valles/Toledo caldera complex, Jemez Volcanic Field, New Mexico. Annu. Rev. Earth Planet. Sci. 18, Heiken, G., Goff, F., Stix, J., Tamanyu, S., Shafiqullah, M., Garcia, S., Hagan, R. (1986). Intracaldera volcanic activity, Toledo caldera and embayment, Jemez Mountains, New Mexico. Journal of Geophysical Research. 91, Hervig, R.L., and Dunbar, N.W. (1992). Cause of chemical zoning in the Bishop (California) and Bandelier (New Mexico) magma chambers. Earth and Planetary Science Letters. 111, Hildreth, W.S. (1981). Gradients in silicic magma chambers: implications for lithospheric magmatism. Journal of Geophysical Research, 86, B11, Hildreth, W.S., and Wilson, C.J.N. (2007). Compositional zoning in the Bishop Tuff. Journal of Petrology. 48, Jarosewich, E., Nelen, J.A., and Norberg, J.A. (1980). Reference samples for electron microprobe analysis. Geostandards Newsletter. 4,

109 Kennedy, B., Stix, J., Lavallée, Y., and Longpré, M-A. (2004). Controls on caldera structure: Results from analogue sandbox modeling. Geological Society of America Bulletin. 116, Krier, D., Caporuscio, F., Gardner, J., and Lavine, A. (1998). Stratigraphy and geologic structure at the Chemical and Metallurgy (CMR) Building, Technical Area 3, Los Alamos National Laboratory. Report LA MS. 32 p. Larsen, R. B., Henderson, I., Ihlen, P. M., and Jacamon, F. (2004). Distribution and petrogenetic behaviour of trace elements in granitic pegmatite quartz from South Norway. Contributions to Mineralogy and Petrology. 147, Lewis, C.J., Lavine, A., Reneau, S.L., Gardner, J.N., Channell, R., Criswell, C.W. (2002). Geology of the Western Part of Los Alamos National Laboratory (TA-3 to TA-16), Rio Grande Rift, New Mexico. Report LA MS. 98 p. Lipman, P.W. (1984). The roots of ash flow calderas in western North America: windows into the tops of granitic batholiths. Journal of Geophysical Research. 89, Lipman, P.W. (2007). Incremental assembly and prolonged consolidation of Cordilleran magma chambers: Evidence from the Southern Rocky Mountain volcanic field. Geosphere. 3,

110 Müller, A., Wiedenbeck, M., Van Den Kerkhof, A. M., Kronz, A., and Simon, K. (2003). Trace elements in quartz a combined electron microprobe, secondary ion mass spectroscopy, laser-ablation ICP-MS, and cathodoluminescence study. European Journal of Mineralogy. 15, Peppard, B.T., Steele, I.M., Davis, A.M., Wallace, P. J., and Anderson, A.T. (2001). Zoned quartz phenocrysts from the rhyolitic Bishop Tuff. American Mineralogist. 86, Phillips, E.H. (2004). Collapse and resurgence of the Valles caldera, Jemez Mountains, New Mexico: 40 Ar/ 39 Ar age constraints on the timing and duration of resurgence and ages of megabreccia blocks. M. Sc. thesis. New Mexico Institute of Mining and Technology, Socorro, New Mexico. 200 pp. Phillips, E.H., Goff, F., Kyle, P.R., McIntosh, W.C., Dunbar, N.W., and Gardner, J.N. (2007). The 40 Ar/ 39 Ar age constraints on the duration of resurgence at the Valles caldera, New Mexico. Journal of Geophysical Research. 112, B08201, doi: /2006jb Roche, O., Druitt, T.H., and Merle, O. (2000). Experimental study of caldera formation. Journal of Geophysical Research. 105,

111 Rowe, M.C., Wolff, J.A., Gardner, J.N., Ramos, F.C., Teasdale, R., and Heikoop, C.E. (2007). Development of a continental volcanic field: petrogenesis of pre-caldera intermediate and silicic rocks and origin of the Bandelier magmas, Jemez Mountains (New Mexico, U.S.A). Journal of Petrology. 48, Ryerson, F.J., and Watson, E.B. (1987). Rutile saturation in magmas: implications for Ti Nb Ta depletion in island-arc basalts. Earth & Planetary Science Letters. 86, Self, S., Goff, F., Gardner, J. N., Wright, J. V. and Kite, W. M. (1986). Explosive rhyolitic volcanism in the Jemez Mountains: vent locations, caldera development and relation to regional structure. Journal of Geophysical Research. 91, Self, S., Lipman, P. W. (1989). Large ignimbrites and caldera-forming eruptions. IAVCEI Working Group on Explosive Volcanism and Its Products. Guidebook, Field Workshop (11WA) in Jemez Mountains, New Mexico and San Juan Mountains, Colorado. Shane, P., Smith, V. C., and Nairn, I. (2008). Millennial timescale resolution of rhyolite magma recharge at Tarawera volcano: insights from quartz chemistry and melt inclusions. Contributions to Mineralogy and Petrology. 156, Smith, R.L. (1979). Ash-flow magmatism. In Chapin, C. E., and Elson W. E., eds. Ash- Flow Tuffs. Geological Society of American Special Paper. 180, p

112 Smith, R.L., and Bailey, R.A. (1966). The Bandelier Tuff: A study of ash-flow eruption cycles from zoned magma chambers. Bulletin of Volcanology. 29, Smith, R.L., and Bailey, R.A. (1968). Resurgent cauldrons. Geological Society of America Memoir. 116, Smith, R.L., Bailey, R.A., and Ross, C.S. (1970). Geologic map of the Jemez Mountains, New Mexico. U.S, Geol. Surv., Misc. Geol. Invest., Map I-571. Smith, V., Shane, P., Nairn, I. (2010). Insights into silicic melt generation using plagioclase, quartz and melt inclusions from the caldera-forming Rotoiti eruption, Taupo volcanic zone, New Zealand. Contributions to Mineralogy and Petrology. doi: /s Snyder, D. (2000). Thermal effects of the intrusion of basaltic magma into a more silicic magma chamber and implications for eruption triggering. Earth & Planetary Science Letters. 175, Spell, T.L., Harrison, T.M., and Wolff, J.A. (1990). 40 Ar/ 39 Ar dating of the Bandelier Tuffs and San Diego Canyon ignimbrites, Jemez Mountains, New Mexico: Temporal constraints on magmatic evolution. Journal of Volcanology and Geothermal Research. 43,

113 Spell, T.L., Kyle, P.R., Thirlwall, M.F., and Campbell, A.R. (1993). Isotopic and Geochemical Constraints on the Origin and Evolution of Postcollapse Rhyolites in the Valles Caldera, New Mexico. Journal of Geophysical Research. 98, Spell, T.L., McDougall, I., and Doulgeris, A.P. (1996). Cerro Toledo Rhyolite, Jemez Volcanic Field, New Mexico: 40Ar/39Ar geochronology of eruptions between two caldera-forming events. Geological Society of America Bulletin. 108, Spera, F.J., and Crisp, J.A. (1981). Eruption volume, periodicity, and caldera area: relationships and inferences on development of compositional zonation in silicic magma chambers. Journal of Volcanology and Geothermal Research. 11, Stimac, J. A. (1996). Hornblende-dacite pumice in the Tshirege member of the Bandelier Tuff: Implications for magma chamber and eruptive processes. In Goff, F., Kues, B. S., Rogers, M.A., McFadden, L.D., and Gardner, J.N., eds. New Mexico Geological Society Fall Field Conference Guidebook 47, Jemez Mountains Region. Socorro, New Mexico Geological Society, p Stimac, J., Hickmott, D., Abell, R., Larocque, A.C.L., Broxton, D., Gardner, J., Chipera, S., Wolff, J., and Gauerke, E. (1996). Redistribution of Pb and other volatile trace metals during eruption, devitrification, and vapor-phase crystallization of the Bandelier Tuff, New Mexico. Journal of Volcanology and Geothermal Research. 73,

114 Stix, J. F. (1989). Physical and chemical fractionation processes in subaerial and subaqueous pyroclastic rocks. Ph.D. thesis. University of Toronto, 365 pp. Stix, J., Gauthier, G., and Ludden, J.N. (1995). A critical looks at quantitative laserablation ICP-MS analysis of natural and synthetic glasses. The Canadian Mineralogist. 33, Stix, J., Goff, F., Gorton, M. P., Heiken, G., and Garcia, S. R. (1988). Restoration of compositional zonation in the Bandelier silicic magma chamber between two caldera forming eruptions: geochemistry and origin of the Cerro Toledo Rhyolite, Jemez Mountains, New Mexico. Journal of Geophysical Research. 93(B6): Stix, J., and Gorton, M.P. (1990a). Changes in Silicic Melt Structure Between the Two Bandelier Caldera-Forming Eruptions, New Mexico, USA: Evidence from Zirconium and Light Rare Earth Elements. Journal of Petrology. 31, Stix, J., and Gorton, M.P. (1990b). Variations in trace element partition coefficients in sanidine in the Cerro Toledo Rhyolite, Jemez Mountains, New Mexico: Effects of composition, temperature, and volatiles. Geochemica et Cosmochimica Acta. 54, Streck, M.J. (2008). Mineral textures and zoning as evidence for open system processes. Reviews in Mineralogy and Geochemistry. 69,

115 Vance, J.A. (1969). On synneusis. Contributions to Mineralogy & Petrology. 24, Vasquez, J.A., Kyriazis, S.F., Reid, M.R., Sehler, R.C., and Ramos, F.C. (2009). Thermochemical evolution of young rhyolites at Yellowstone: Evidence for a cooling but periodically replenished postcaldera magam reservoir. Journal of Volcanology and Geothermal Research. 188, Vernon, R.H. (2004). A practical guide to rock microstructure. Cambridge University Press. Cambridge, UK. 594 pp. Wark, D.A., Hildreth, W., Spear, F.S., Cherniak, D.J., and Watson, E.B. (2007). Preeruption recharge of the Bishop magma system. Geology. 35, Wark, D.A., and Watson, E.B. (2006). TitaniQ: a titanium-in-quartz geothermometer. Contributions to Mineralogy and Petrology. 152, Warren, R.G., Goff, F., Kluk, E.C., and Budahn, J.R. (2007). Petrography, chemistry, and mineral compositions for subunits of the Tshirege member, Bandelier Tuff within the Valles caldera and Pajarito Plateau. In Kues, B.S., Kelley, S.A., and Lueth, V.W., eds. Geology of the Jemez Region II. New Mexico Geological Society 58 th Annual Field Conference Guidebook. Socorro, New Mexico Geological Society, p

116 Warren, R.G., McDonald, E.V., and Ryti, R.T. (1997). Baseline geochemistry of soil and bedrock Tshirege Member of the Bandelier Tuff at MDA-P. Los Alamos National Laboratory, Report LA MS, 89 p. Warshaw, C. M., and Smith, R. L. (1988). Pyroxenes and fayalites in the Bandelier Tuff, New Mexico: temperatures and comparison with other rhyolites. American Mineralogist 73, Wiebe, R. A., Wark, D. A., and Hawkins, D. P. (2007). Insights from quartz cathodoluminescence zoning into crystallization of the Vinalhaven granite, coastal Maine. Contributions to Mineralogy and Petrology. 154, Wolff, J. A., Balsley, S. D., and Gregory, R. T. (2002). Oxygen isotope disequilibrium between quartz and sanidine from the Bandelier Tuff, New Mexico, consistent with a short residence time of phenocrysts in rhyolitic magma. Journal of Volcanology and Geothermal Research. 116, Wolff, J.A., and Gardner, J.N. (1995). Is the Valles caldera entering a new cycle of activity? Geology. 23 (5), p

117 Appendix 1. Compilation of bulk XRF analyses and ages for rocks from various formations and units of the Jemez Mountains Volcanic Field, New Mexico, USA. 104

118 105

119 106

120 107

121 108

122 109

123 110

124 111

125 112

126 Appendix 2. Titanium concentrations in quartz obtained by LA-ICP-MS and corresponding zoning types following classifications of Peppard et al. (2001). 113

127 114

128 115

129 116

130 117

131 118

132 119

133 120

134 121

135 122

136 Appendix 3. Un-normalised XRF bulk analyses of rocks from various formations and units of the Jemez Mountains Volcanic Field, New Mexico, USA. 123

137 124

138 125

139 126

140 127

141 128

142 129

143 130

144 Appendix 4. Standardisation data for LA-ICP-MS analysis of Ti-in-quartz, XRF analysis of bulk rock and glass, and EMP Ti-in-glass analysis. 131

145 132

146 133

147 134

148 135

149 Appendix 5. A preliminary study of cathodoluminescene zoning and geochemistry of feldspar in the Upper Bandelier Tuff and resurgence units Several studies have documented compositional variation in alkali feldspar within the UBT with regard to stratigraphic position and thus eruptive age (Caress, 1995, 1996; Phillips, 2004; Warren et al., 2007). Alkali feldspars erupted within the Tsankawi Pumice Bed and Units 1g, 1v and 2 of the ignimbrite are sanidines with Or There is also a significant component of plagioclase derived from disaggregation of the hornblende dacite pumices erupted synchronously with these units (Stimac, 1996). An increase in compositional variation of alkali feldspar (Or <28 Or 44 ) occurs from UBT unit 3 to unit 5. Concurrent with this increased compositional range, there is an increase in 1) the proportion of unidirectionally and complexly zoned grains (geochemical zonation), 2) the maximum compositional range found within individual phenocrysts and 3) the proportion of phenocrysts exhibiting resorption textures (Caress, 1995). Results from this study identify an abrupt occurrence of zoned quartz phenocrysts commencing within UBT unit 3 and continuing for units 4 and 5. An important observation is that the results from the quartz zoning study agree very closely with the feldspar geochemical trends within the UBT, in terms of the stratigraphic position at which the feldspar zoning is first observed. 5.1 Feldspar cathodoluminescence zoning patterns A preliminary assessment of feldspar cathodoluminescence (CL) zoning was undertaken to further assess the compositional trends documented by Caress (1995, 1996) 136

150 and Phillips (2004). Subunits 2-5 of the UBT, the Deer Canyon Rhyolite and the Redondo Creek Rhyodacite were analysed using the same procedure as documented in Section 3. Figure 11 shows representative images of feldspars from these stratigraphic units. In the absence of quantitative analysis of alkali feldspar from the Tsankawi Pumice Bed and UBT unit 1g, caution must be applied regarding speculation on magmatic processes prevalent within this horizon of the Bandelier chamber. However, we have good reason to believe that the trends exhibited by feldspar of UBT unit 2 will be similar to those of the Tsankawi Pumice Bed, since the CL zoning patterns and titanium concentrations of both units are very similar (Fig. 6; Fig. 7). In addition, studies by Caress (1995, 1996) and Phillips (2004) have indicated that the Tsankawi Pumice Bed contains geochemically unzoned feldspar. Three alkali feldspar phenocrysts recovered from the Tsankawi Pumice Bed for this study were unzoned under CL Feldspar zoning classifications In general, there are two main types of zoning identified in this preliminary study which show similarity to zoning patterns found in sanidine crystals from the Laacher See eruption (Ginibre et al., 2004). Firstly, there is a component of crystals which show zones following a concentric growth pattern (C-type, Fig A5.1). The core regions are often rounded and exhibit a dark red CL signal, progressing to lighter pink outer zones. Zone boundaries are sharply defined and zones of intermediate CL intensity are present, although some crystals show oscillatory zoning which varies between dark and bright zones. CR-type zoning denotes a crystal exhibiting concentric zoning but with an extensively-resorbed core region (Fig. A5.1). 137

151 Appendix 5. Figure A5.1. Feldspar cathodoluminescene (CL) zoning pattern classifications, following the criteria of Ginibre et al. (2004). (A) Unzoned feldspar (UZ). (B) Zoned feldspar crystals exhibiting C-type, concentric zoning. CR-type is a further subdivision based on the occurrence of a resorbed core region. (C) Zoned feldspar crystals exhibiting P-type, patchy zoning. PR-type represents a P-type zoned crystal with a resorbed core region. 138

152 139

153 P-type zoning is more localised whereby zones are 'patchy' and amorphous, showing no directional conformity. Dark red CL constitutes a core region, which is often resorbed (PR-type). Overgrowths of patchy zones with lighter CL colours are common, with some exhibiting concentric zoning (C-type) features. The patches are on the order of μm in size. No evidence of exsolution lamallae was found at the visible scale, in agreement with Caress (1995). 5.2 CL zoning patterns results Alkali feldspar from the Tsankawi Pumice Bed and UBT unit 2 is predominantly unzoned and exhibits red CL colours (Fig. A5.2; Fig. 11, L). Within UBT unit 1g, a significant proportion of the feldspar phenocrysts are plagioclase, which may be derived from the hornblende dacite pumice. These show complex internal zonation, often conforming to a concentric pattern and exhibiting yellow-green CL colours. A thin rim of alkali feldspar with a red CL colour encompasses most plagioclase grains (Fig. 11, M). Concurrent with the occurrence of dominantly zoned quartz phenocrysts, there is a notable presence of zoned alkali feldspars at UBT unit 3 (Fig. A5.2). As with the quartz crystals, a small component of the crystals are unzoned (5-17 %). There is also no dominant zoning type throughout these units, a trend which is also noted with quartz (Fig, 6). The relationships between internal zones are more complex than for quartz. The crystal edges of most feldspar crystals are also poorly-defined and resorbed. Some internal zonation is visible optically, but there is a component which is only apparent using CL analysis. This zoning trend is repeatable throughout the late-erupted units 3-5 (Fig. A5.2). 140

154 Appendix 5. Figure A5.2. Feldspar cathodoluminescene (CL) zoning pattern abundances, with reference to stratigraphic position. Subclassifications follow the criteria of Ginibre et al. (2004) (see Fig. A5.1 for representative images of each zoning type). Int. = intracaldera setting, Ext. = extracaldera setting. An insufficient number of alkali feldspar crystals were obtained from the Tsankawi Pumice Bed and UBT unit 1g to be included in quantitative analysis. 141

155 142

156 The proportions of zoned and unzoned crystals within each UBT subunit are remarkably similar for both feldspar (Fig A5.2) and quartz (Fig. 6). Firstly, there is a lack of CL zonation in 88 % of phenocrysts from unit 2 of the UBT, similar to quartz crystals (90 %). Concurrent with the occurrence of an abundance of zoned quartz, feldspar becomes dominantly zoned in UBT unit 3. Again, the proportion of zoned crystals of both quartz and feldspar are very similar, 85 % and 88 %, respectively. For the remaining UBT units 4 and 5, feldspar crystals are predominantly zoned under CL. In addition, there is a small component (5-15 %) of unzoned feldspar crystals, a trend which is also exhibited by quartz (Fig. 6). The CL zoning patterns of alkali feldspar from the Deer Canyon Rhyolite also show similarity to that observed within quartz crystals from the same unit (see section 4.1.3). There is heterogeneity in CL zoning of both quartz and feldspar between spatially separate samples of the Deer Canyon Rhyolite, one from a porphyritic lava erupted onto the western resurgent dome and the other from a crystal-poor lava from the central resurgent dome. Within the porphyritic sample, the proportion of zoned crystals of quartz and feldspar are similar, 92 % for quartz and 82 % for feldspar. The crystal-poor sample exhibits 78 % unzoned feldspar crystals compared to 95 % unzoned quartz crystals (Fig. A5.2, Fig. 6). The zoned feldspar component is dominantly C-type and concentrically zoned. In this sample, it is common to find feldspar and quartz exhibiting synneusis texture (Appendix 5, Fig. A5.3). In general, the feldspar crystals are larger than quartz. The Redondo Creek Rhyodacite exhibits feldspar with interesting morphology and CL zoning. The occurrence of antirapakivi feldspars is observed, whereby plagioclase cores are commonly mantled by sanidine rims (Fig. 12., B). Furthermore, plagioclase 143

157 Figure A5.3. Synneusis texture of feldspar and quartz crystals in Deer Canyon Rhyolite from the central resurgent dome, viewed under CL. All crystals were identified using a petrologic microscope. (A) Multiple unzoned feldspar crystals, scale bar 0.5 mm. (B) Unzoned and rounded quartz crystals, scale bar 0.5 mm. (C) Unzoned feldspar crystals, scale bar 1 mm. 144

158 145