Accurately dating phenocrysts in Holocene volcanic rocks poses many challenges but is critical to placing magmatic processes that occur prior to eruption into a temporal frame-work. We dated alkali feldspar (i.e., orthoclase Or10 to Or46) crystals in four young phonolites from the Teide–Pico Viejo volcanic complex, Tenerife (Spain), using (226Ra)/(230Th) isotopes. Partition coefficients of Ra (DRa) and DRa/DBa of feldspars were predicted using an approach based on the lattice strain model, which yielded crystallization ages that overlap or predate known eruption ages for the Lavas Negras (ca. 1 ka), Montaña Blanca (ca. 2 ka), Arenas Blancas (ca. 2–4 ka), and Teide H (ca. 6 ka) phonolites. Crystallization of feldspar may occur up to the time of eruption, with >8 ka crystals also present, possibly suggesting extended magma differentiation times. However, feldspars yielding finite (226Ra)/(230Th) ages are mostly in equilibrium with the groundmass, unlike >8 ka crystals, which were therefore identified as antecrysts/xenocrysts. The 87Sr/86Sr ratios of feldspars indicate that crystallization predated late-stage assimilation, affecting 87Sr/86Sr ratios of some melts. The (226Ra)/(230Th) ages also constrain the tempo of phonolite magma evolution on Tenerife. Integration of (226Ra)/(230Th) ages with feldspar major elements, trace elements, and isotopes provides a powerful means for investigating crystallization histories using a dominant mineral that controls the overall magmatic evolution of phonolites on thousand-year time scales.

Accurately dating crystals in rocks from active volcanoes has been an elusive goal. Most methods provide an eruption age (e.g., 40Ar/39Ar dating) that is useful in reconstructing the eruption history of a volcano (Lanphere et al., 2007; Ramos et al., 2016; Preece et al., 2018) but offers limited pre-eruption crystal-lization and magmatic information. In contrast, feldspar (226Ra)/(230Th) crystallization ages provide (1) eruption age constraints, (2) time frames connected to magma differentiation reflected in feldspar Ca, Ba, and Sr variations, and (3) links to magmatic sources when correlated to Sr and Pb isotopes in the same feldspars (Ramos et al., 2019). The unmatched ability to determine such constraints at a single-crystal level using a dominant mineral such as feldspar can elucidate the time scales over which magmas crystallize, evolve, and reside prior to eruption. The volcanically active island of Tenerife in the Canary Islands, Spain (Fig. 1), home to Las Cañadas caldera and the Teide–Pico Viejo (T-PV) complex, provides an opportunity to use feldspar (226Ra)/(230Th) ages to evaluate Holocene phonolitic magma evolution. The origins of recent T-PV phonolitic eruptions are enigmatic, and feldspar ages, major and trace elements, and isotopes can be used to identify the processes affecting magmas, potential magmatic sources, and the time scales of magmatic differentiation and storage. The timing of phonolite evolution is controversial given that most uranium-series isotopes reflect disequilibrium (Turner et al., 2017), but (226Ra)/(230Th) ages better constrain these time scales.

The island of Tenerife results from the Canary hotspot, which appeared at ca. 70 Ma (Geldmacher et al., 2001). The volcanic evolution of Tenerife involved construction of a basaltic shield complex starting at ca. 12 Ma, followed by formation of a central mafic-phonolitic volcanic complex and the Las Cañadas edifice from ca. 3.8 Ma onward, including several basanite to phonolite magmatic cycles (Fuster et al., 1968; Ancochea et al., 1990; Cas et al., 2022).

The Las Cañadas edifice is composed of the post-shield, volcanic Lower Group (3.5–1.59 Ma) overlain by the Upper Group (1.59–0.175 Ma), which consists of the Ucanca (1.59–1.18 Ma), Guajara (0.85–0.57 Ma), and Diego Hernández (0.37 to ca. 0.18 Ma) formations (Huertas et al., 2002; Brown et al., 2003; Edgar et al., 2007). The Upper Group mafic to felsic volcanic cycles ended with voluminous (5–15 km3) pyroclastic eruptions that produced numerous east-to-west–migrating caldera collapses (Martí et al., 1994; Cas et al., 2022). The most recent, ca. 0.18 Ma, resulted in the present form of Las Cañadas caldera, a 16 × 9 km depression in the center of Tenerife (Fig. 1). Younger, predominately effusive activity within the caldera formed two stratovolcanoes, Teide and Pico Viejo, as well as Montaña Blanca and other satellite vents (Ablay et al., 1995, 1998), which represent more explosive volcanism (García et al., 2012; Dorado et al., 2021). Along with monogenetic basaltic volcanism outside the caldera, the T-PV complex is the focus of most of the current activity on Tenerife, with 13 identified eruptions in the past ~2 k.y. (Geyer and Martí, 2010; Di Roberto et al., 2016).

We collected four phonolitic lavas (Fig. 1) from Las Cañadas caldera (Table 1). Lavas were processed to obtain clean groundmass and feldspar crystals/fragments (Figs. S1 and S2 in the Supplemental Material1) using procedures described in the Supplemental Material. Major elements of alkali feldspar were analyzed by microprobe (Table S5; Fig. S3), and partition coefficients DCa, DSr, and DBa were determined using the concentrations measured from groundmass and feldspar (Table S1); these were then used to construct Onuma curves (Figs. S4–S7) to predict DRa and DRa/DBa of feldspars (Onuma et al., 1968). Groundmass and feldspar Ra concentrations were then age corrected to determine the age at which the calculated feldspar DRa/DBa was recorded (Tables S2 and S4; Fig. S8); this was interpreted as the crystallization age of the feldspar. Additionally, we measured Sr and Pb isotopes of groundmass and individual feldspar crystals/fragments/separates (Table S3).

Alkali feldspars in Tenerife phonolites reflect a range of ages, major and trace elements, and isotope ratios (Figs. 2 and 3) that constrain the timing of petrogenetic processes. We place these feldspar variations in the context of (226Ra)/(230Th) ages for T-PV eruptions.

Lavas Negras

The Lavas Negras lava flow, which has a 14C eruption age of 1150 ± 140 yr B.P. (Carracedo et al., 2007) and a K/Ar age of 800 ± 300 a (Quidelleur et al., 2001), hosts feldspars that vary from orthoclase Or16 to Or31, with many having higher Or cores (Fig. 2A). Lavas Negras feldspars reflect two populations. One population has an average age of ca. 1.3 ka (n = 4; Fig. 3A) with individual ages overlapping the eruption age, consistent with crystallization occurring up to the time of eruption with little crystal residence time. Ba and Sr concentrations vary widely (1716–2849 ppm and 200–502 ppm, respectively). A second population does not yield parabolic Onuma curves for divalent cations, likely due to compositional zoning, and so crystallization ages were not calculated because the assumption of crystal-melt equilibrium cannot be valid. These feldspars have overlapping but different Ba and Sr concentrations (1126–5503 ppm and 88–374 ppm, respectively) as compared to the first population of feldspars. The 87Sr/86Sr ratios (~0.7032 versuŝ0.7031) and 208Pb/206Pb ratios (~2.002) of groundmass are similar to both feldspar populations (Fig. 3), consistent with closed-system differentiation.

Montaña Blanca

Feldspars from the 2025 ± 40 a (Ablay et al., 1995) Montaña Blanca lava have similar compositions to those from Lavas Negras (Fig. 2B), with only slightly higher Or (Or20 to Or33). Single feldspars and the feldspar separate have Ba concentrations of ~5000–7000 ppm and Sr of ~300–460 ppm. All single feldspars/fragments >1 mg were however “dead” to radium (i.e., contained no measurable radium as a result of 226Ra decay in the absence of parental 230Th, which is incompatible in feldspar) and older than 8 ka (Fig. 3B). The feldspar separate, which is composed of ~20 individual <1 mg crystals, yielded an age of 5331 ± 533 a. This younger age likely results from the presence of both old (>8 ka) >1 mg and young (<8 ka) <1 mg feld-spars. Mineral separate results indicate larger feldspars are likely antecrystic/xenocrystic compared to younger, smaller phenocryst ~1 mg in mass. Ages of feldspar in Montaña Blanca high-light the importance of dating both single-crystal and feldspar separates in lavas with mixed feld-spar populations.

Montaña Blanca antecrystic or xenocrystic feldspars are characterized by less radiogenic 87Sr/86Sr (0.7031 vs. 0.7047) and higher 208Pb/206Pb ratios (>2.005 versus ~2.003) compared to accompanying groundmass, confirming that feldspar crystallization predated the open-system effects that affected the melt. In addition, variable isotope ratios between the groundmass and feldspar undermined crystalmelt equilibrium.

Arenas Blancas

The Arenas Blancas lava has not been dated but directly underlies the ca. 2 ka Montaña Blanca flow (Fig. 1; Ablay and Martí, 2000). Feldspars in this phonolite fall into at least two compositional populations, one with Or < 25 and a second with Or25–32 (Fig. 2C). The first population has lower Or, higher CaO (1.4 wt%), and radium “dead” antecrystic or xenocrystic feldspars. Ba and Sr concentrations are >5000 ppm and >318 ppm, respectively. The second feldspar population with higher Or has lower CaO (~0.96 wt%) and uniform Ba and Sr concentrations (~2200 and ~125 ppm, respectively). These feldspars yielded ages of ca. 4 ka (Fig. 3), and they provide an upper age limit for the Arenas Blancas eruption. The presence of >8 ka feldspar, however, indicates that the Arenas Blancas magma remobilized preexisting antecrysts/xenocrysts.

Arenas Blancas groundmass is characterized by slightly higher 87Sr/86Sr ratios (0.7035 versus 0.7032) than feldspars, confirming latestage assimilation after ca. 4 ka. To test whether added Sr from assimilation could affect DRa and feldspar ages, the effects of adding 25% and 50% of total groundmass Sr (i.e., 2 or 4 ppm of 8 ppm total) from assimilation were estimated (see the Supplemental Material). Calculated ages, however, varied by ~10% for both, which is the error for ca. 4 ka feldspars, and so Ra/Th feldspar ages are assumed to be robust despite minor assimilation. In contrast to Sr, phenocryst 208Pb/206Pb ratios overlap the groundmass.

Teide H

Teide H, located along the northern flank of Teide outside Las Cañadas caldera proper, is the oldest phonolite considered here. Teide H feldspars reflect at least two compositions, Or > 36 and Or15–29. The lower Or population is also shifted toward higher CaO concentrations. The feldspar separate was radium “dead,” and so single feldspar crystals were not analyzed. As a result, the only age constraint for this flow comes from an ~6000 yr maximum melt (226Ra)/(230Th) age (Brown, 2021). Teide H feldspars have similar 87Sr/86Sr ratios to the groundmass (0.7032 versus 0.7033) but more radiogenic 208Pb/206Pb (>2.005 versus 2.002), indicating that Teide H feldspars are likely antecrystic or xenocrystic and do not reflect crystal-melt equilibrium.

Feldspar ages from Tenerife phonolites illuminate magmatic and petrogenetic characteristics of phonolitic volcanism and are consistent with preexisting age constraints. Such ages provide a direct, unequivocal means of distinguishing phenocrysts from antecrysts/xenocrysts and therefore offer petrogenetic insights when accompanied by chemical profiles and Sr and Pb isotopes of the same crystals.

Three of the phonolites host >8 ka feldspars that are several thousands of years older than their eruption ages. In the Arenas Blancas lava, these antecrysts/xenocrysts are accompanied by phenocrysts in Pb isotopic equilibrium with the groundmass, indicating a maximum residence time of 2 k.y. In contrast, apparent phenocrysts in Montana Blanca are out of Pb isotope equilibrium with the groundmass (Table S3). The youngest phonolite (Lavas Negras) hosts anorthoclase phenocrysts that formed within a few hundred years of eruption, plus a zoned crystal population of unknown age. Whole rocks and most crystals yielding finite Ra ages have similar Pb isotope ratios (Fig. 3), suggesting derivation from a common reservoir feeding Teide phonolite eruptions during the past few thousand years. In Arenas Blancas and Montaña Blanca, the youngest feldspars are in equilibrium with the groundmass, whereas the >8 ka crystals span a range of Pb isotope ratios (Tables S2 and S3). Pb isotope ratios are variable (208Pb/206Pb = 1.969–2.030) in older Tenerife rocks (Palacz and Wolff, 1989; Simonsen et al., 2000). Therefore, the most parsimonious explanation is that the older crystals are xenocrysts introduced during assimilation of older rocks. Alternatively, they could be crystals derived from an isotopically variable, crystal-rich mush reservoir. The Ra-bearing Lavas Negras feldspars also span a range in Pb isotope ratios, consistent with the presence of an assimilant that affected the melt that crystallized feldspar much younger than 8 ka.

For T-PV phonolites, Sr and Pb isotopes in feldspar phenocrysts and host groundmass were generally in equilibrium prior to latestage assimilation. Groundmass 87Sr/86Sr and 143Nd/144Nd are near-constant at ~0.7031 and ~0.51288, respectively, consistent across the whole island (Simonsen et al., 2000), with the exception of high 87Sr/86Sr in some Sr-poor phonolites. Elevated 87Sr/86Sr ratios are likely due to seawater contamination (Palacz and Wolff, 1989), either directly or through assimilation of seawater-altered rock, although feldspars in the rocks remained unaffected (Fig. 3; Table S3). Seawater has low Pb contents and thus would have had limited impact on 208Pb/206Pb during alteration. Feldspar crystals, whether phenocrystic or xenocrystic, retained 87Sr/86Sr of ~0.7031–0.7032 (Fig. 3) but variable 208Pb/206Pb, with xenocrysts commonly retaining higher 208Pb/206Pb. These older remobilized feldspars retained different Pb signatures to their host melts, which likely originated from inheritance from feldspar crystallizing in magmas with variable 208Pb/206Pb values.

The (226Ra)/(230Th) ages of individual feldspar crystals in young (<8 ka) Tenerife phonolites are consistent with previously determined eruption ages and stratigraphic constraints, better constrain the ages of undated lavas, and serve to identify xenocryst/antecryst populations. Lavas Negras hosts the youngest crystals (ca. 1.0 ka to 1.4 ka) that formed up to the time of eruption without the effects of late-stage assimilation. These feldspars have limited Sr and Pb isotopic variations that support crystal-melt equilibrium. In contrast, alkali feldspars from Montaña Blanca are >8 ka and likely xenocrystic, although a separate yielding a younger age (ca. 5.4 ka) indicates the presence of younger (<8 ka) crystals <1 mg in mass. Arenas Blancas phonolite hosts a mixed population of older (>8 ka) antecrysts or xenocrysts and younger (ca. 4 ka) phenocrysts, which are clearly identified in both major and trace elements, while the separate from the oldest Teide H phonolite is composed of feldspar >8 ka with Pb isotopes indicating a likely antecrystic or xenocrystic origin. The 208Pb/206Pb ratios indicate that phonolitic lavas were also associated with a limited range of parental magmas, while 87Sr/86Sr ratios are uniform and similar to most lavas erupted on Tenerife unless affected by late-stage assimilation. The contrast between feldspars with finite (226Ra)/(230Th) ages and Ra-dead crystals that are out of equilibrium with the groundmass serves to unequivocally distinguish phenocrysts from antecrysts/xenocrysts and places direct constraints on the tempo of magmatic evolution. Time scales of ~1000 years are consistent with calculated evolution times for phonolites (Wolff, 2017) and somewhat shorter than those for rhyolitic systems (Bachmann and Bergantz, 2004). Integration of (226Ra)/(230Th) ages with feldspar major elements, trace elements, and isotopes provides a powerful pathway for investigating crystallization histories using a dominant mineral phase that controls the overall magmatic evolution of Tenerife phonolites on thousand-year time scales.

This research was funded by the New Mexico State University Johnson Mass Spectrometry Laboratory (Las Cruces, New Mexico) and a Geological Society of America Lipman Grant awarded to Brown. Dorado was supported by a Spanish FPU (Training programme for Academic Staff) grant (FPU18/02572), a mobility grant (EST19/00297) from the Ministry of Universities of Spain, and the European Commission European Volcano Early Warning System (EVE) (ref. DG ECHO H2020 826292). We also express our appreciation to Scott Boroughs for technical assistance, and Graham Edwards and Mark Stelton for thoughtful reviews of the manuscript.

1Supplemental Material. Samples, preparation, and methods; feldspar age variations resulting from late-stage assimilation; Figures S1–S4; and Tables S1–S4. Please visit https://doi.org/10.1130/GEOL.S.20044358 to access the supplemental material, and contact editing@geosociety.org with any questions.
Gold Open Access: This paper is published under the terms of the CC-BY license.