Abstract

Postcaldera (≤ 180 ka) peralkaline trachytes and rhyolites from Socorro Island, Mexico, have relatively restricted ranges of (230Th)/(232Th)o (0.99−1.14) and (238U)/(232Th) (0.91−0.98). Most of the samples exhibit 238U−230Th disequilibria, with initial (230Th) enrichments of up to 21%. In general, (230Th)/(232Th)o values are lower in the rhyolites than in the trachytes. (230Th)/(232Th)o of postcaldera alkalic basalts may also have a relatively restricted range (0.94−0.96), although ratios as low as 0.84 and as high as 1.04 are permitted by uncertainties in some of the ages. These values are also lower than those of the trachytes. Sr and Nd isotopic ratios of the basaltic and silicic suites have fairly restricted ranges that collectively overlap. Previous work suggests that trachytic magmas form by partial melting of basaltic basement, but closed-system partial melting of a source with Th isotope ratios like those of the postcaldera basalts would probably produce trachytic melts with lower rather than higher (230Th)/(232Th)o. Open-system processes affecting melts from such basement are also ruled out because of the lack of a reasonable contaminant that could increase the (230Th)/(232Th) of the trachytic magmas. Postcaldera trachytic magmas probably derive from basaltic basement characterized by Th isotope ratios similar to those of the silicic rocks, and moderate degrees of melting of such material can account for the Th/U of the trachytes. Differences in (230Th)/(232Th)o between the trachytes and rhyolites delimit maximum residence times for rhyolitic magmas of 40−50 kyr, but it is likely that residence times are shorter. This coupled with correlations between Sr and Nd isotopes and indices of differentiation within the silicic suite suggest that rhyolites are related to trachytes by assimilation–fractional crystallization. The most likely assimilant is basement that formed during the silicic phases of magmatism on Socorro. Our results indicate that volcanic basement may be an important component in the genesis of evolved ocean island magmas.

Introduction

The chemistry of evolved ocean island magmas provides insight into differentiation processes that occur within oceanic magma chambers and the genetic relationship such magmas have with the more abundant ocean island basalts. Thorium isotopes are especially powerful tracers of processes of magma genesis because the short half-life of 230Th allows identification of mechanisms not easily resolved with more conventional isotopic systems (e.g. Sigmarsson et al., 1992a). Particularly when combined with major and trace element and other isotopic data, Th isotope signatures have revealed complex petrogenetic histories for ocean island magmas (e.g. Hemond et al., 1988; Sigmarsson et al., 1992a, 1992b; Widom et al., 1992). In addition, U–Th disequilibria may yield constraints on magma residence times.

Socorro Island, Mexico, is the only ocean island in the Pacific with a relatively large volume of silicic peralkaline volcanic rocks. In this paper, we report new Th isotopic and U–Th abundance data for Socorro lavas which are postcaldera (≤ 180 ka) in age, to investigate the petrogenetic link between trachytes and rhyolites as well as between trachytes and contemporaneous mafic magmas. A simple closed-system relationship between trachytic and mafic postcaldera magmas is precluded by the Th isotope data. Coupled with variations in major and trace element abundances which suggest that trachytic magmas were probably produced by partial melting of basaltic basement (Bohrson & Reid, 1997), these new data show that this basement, which presumably formed during the volumetrically dominant precaldera basaltic stage of magmatism on Socorro, had Th isotope ratios similar to those of the postcaldera silicic magmas. Although not recognized in an earlier study, Sr and Nd isotope ratios in the silicic postcaldera suite generally covary with indices of differentiation, consistent with the interpretation that rhyolites are related to trachytes by assimilation–fractional crystallization (AFC). Considerations that build on the Th isotopic data suggest that the most likely assimilant is basement formed during the silicic stages of magmatism on Socorro. Partial melting of basaltic basement and assimilation of silicic basement documented in this study suggest that deep- and shallow-level recycling of crustal material may be important but not easily recognized processes in the origin of evolved ocean island magmas.

Background

Geologic setting and eruption history of Socorro Island

Socorro Island, Mexico, is located in the eastern Pacific Ocean at the northern end of the Mathematicians Ridge, a mid-ocean ridge spreading center that was largely abandoned at ∼3.5 Ma when activity shifted to the East Pacific Rise (Mammerickx et al., 1988). The eruptive history of Socorro Island has been divided into pre-, syn-, and postcaldera phases (Bryan, 1976; Bohrson et al., 1996). The dominant volume of the volcanic edifice (∼2400 km3), which is submarine, erupted before the formation of a small caldera partially preserved near the summit of the island and is probably basaltic in composition. Subaerially exposed precaldera basalt is present only as a few thin flows at the base of a sea cliff on the eastern side of the island (Bryan, 1976). Most subaerially exposed pre- and syncaldera deposits are silicic peralkaline ignimbrites that erupted episodically between ∼540 and 370 ka (Bohrson et al., 1996). Repose intervals between these eruptions had fairly regular durations of 30–35 kyr. Together, the ignimbrites have a minimum volume of 8 km3. Postcaldera silicic peralkaline activity began by ∼180 ka and continued to at least 15 ka, forming ≥ 0.8 km3 of lava flows and domes that dominate the northern, western and southern quadrants of the island. Postcaldera alkaline basaltic lava flows and cones erupted contemporaneously with these silicic magmas but are largely restricted to the southeastern quadrant of the island; approximately 20 vent complexes compose a ≥ 0.1 km3 volcanic field.

Geochemical characteristics and petrogenetic histories of silicic and basaltic rocks

This study focuses on silicic and mafic postcaldera lavas from Socorro Island. Like the pre- and syncaldera silicic rocks, the postcaldera silicic ones are mildly to strongly peralkaline [peralkalinity index, i.e. molar (Na2O + K2O)/Al2O3, of 1.1−2.2] trachytes and rhyolites. Typically, they are Qz (quartz), Ac (acmite) and Ns (Na-metasilicate) normative. Phenocryst abundances are in the range 0−15 vol. %, with alkali feldspar ≫ sodic pyroxene ± fayalite ± Fe–Ti oxides ± aenigmatite. Groundmass ranges from fresh glass to holocrystalline assemblages dominated by alkali feldspar. In general, major and trace element trends in the silicic rocks are consistent with derivation of the rhyolites from the trachytes by fractional crystallization of alkali feldspar ≫ Na pyroxene + fayalite + ilmenite + apatite (maximum amount of crystallization 80%; Bohrson & Reid, 1997). Based on experimentally determined phase equilibria (Bailey et al., 1975), the predominance of alkali feldspar in the mode and in the inferred fractionating assemblage suggests that the silicic magmas differentiated in a shallow-level magma reservoir (≤ 1 kbar). Acid-labile radiogenic Sr provides evidence for post-eruptive alteration in most of the silicic rocks, probably through interaction with hydrothermal fluids that have Sr isotope signatures similar to that of seawater. One postcaldera sample has a negative Ce anomaly as well as rare earth element (REE) abundances in excess of those predicted by fractional crystallization; as in the several pre- and syncaldera samples in which these characteristics are observed, assimilation of small amounts of hydrothermal sediments is indicated (Bohrson & Reid, 1997).

The petrogenetic history of the mafic suite, which comprises alkalic basalts, hawaiites and mugearites, has been described by Bohrson & Reid, (1995). Most of the samples (Group 1; see Table 1 for additional information) are sparsely phyric to aphyric. For the few phyric samples, phenocryst abundances range up to 25%, with plagioclase ≫ olivine ≈ clinopyroxene, or plagioclase ≫ olivine. Holocrystalline groundmass is composed of plagioclase + clinopyroxene + olivine + Fe–Ti oxides ± apatite. Major and trace element and Sr, Nd, and Pb isotopic data indicate that the suite is dominantly related by fractional crystallization of compositionally homogeneous parental alkalic basalts. A subset of the basalt samples has negative Ce anomalies and enrichments in P2O5 and REE that are a consequence of magmatic contamination by hydrothermal sediments (Group 2) or apatite formed as a by-product of fractional crystallization of Socorro magmas (Group 3). Acid-labile radiogenic Sr provides evidence for minor post-eruptive fluid-rock interaction.

Results

Seven unleached silicic and four unleached mafic rocks that are generally representative of the elemental and Sr, Nd, and Pb isotopic ranges for all subaerially exposed rocks from Socorro (Bohrson & Reid, 1995, 1997) were analyzed by thermal ionization mass spectrometry for Th isotope ratios and for U and Th abundances. Most of the samples were also characterized for (234U)/(238U) by thermal ionization mass spectrometry. Th isotopic and U and Th concentration data were reproduced for each sample from separate dissolutions. For 8 of 11 samples, the difference between replicate (230Th)/(232Th) analyses is that expected at the 2σ level solely from in-run statistics [mean square weighted deviation (MSWD ≤3.8; Wendt & Carl, 1991]. Reproducibility for the remaining three samples is larger but standard deviations on the weighted means (≤ 0.31%) are small with respect to the overall range in Th isotope ratios. Reproducibility for Th/U is generally better than 2%.

Compositional and isotopic data for the samples (Bohrson & Reid, 1995, 1997) and a brief description of analytical techniques are presented in Table 1. Because the rocks show evidence for post-eruptive fluid-rock interaction, the magmatic Sr isotope signatures of the silicic rocks are best represented by 87Sr/86Sr of age-corrected, acid-leached alkali feldspar, and the magmatic 87Sr/86Sr values of the basalts are best represented by those of acid-leached whole rocks.

Among the postcaldera silicic samples analyzed for Th isotopes is the only one (91-17) characterized by relative enrichments in REE and negative Ce anomalies (Enr group, Table 1); all other silicic postcaldera samples have REE abundances consistent with fractional crystallization (Ref group, Table 1). None of the relatively REE depleted samples identified by Bohrson & Reid (1997) is included in this study because they are all older than ∼370 ka. High-precision 40Ar/39Ar ages are available for six of the silicic samples, and the age of the seventh sample (90-101) is estimated from its stratigraphic relationship with dated samples (Bohrson et al., 1996). (234U)/(238U) of five postcaldera trachytes and rhyolites are within 1.1% of secular equilibrium (parentheses denote activity ratios). (230Th)/(232Th) are 0.95−1.07 and (238U)/(232Th) are 0.91–0.98. Thorium activity ratios corrected for post-eruptive in situ decay, (230Th)/(232Th)o, range from 0.99 to 1.14. Most of the samples are enriched in 230Th with respect to its ultimate parent, 238U, with initial 230Th enrichments ranging up to 21%; the exception is the REE-enriched sample, 91-17, which plots on the equiline (Fig. 1). U and Th concentrations of the silicic rocks increase with differentiation but Th/U ratios are independent of fractionation and span a fairly narrow range, particularly if sample 91-17 is excluded. In general, (230Th)/(232Th)o and (230Th)/(238U)o are higher in the trachytes than in the rhyolites.

Of the four alkalic basalts analyzed, one sample (90-120) is from a volcanic cone that overlies a lacustrine deposit dated at 5040 ± 460 years (14C, Farmer et al., 1993); stratigraphic relations suggest that the other three were erupted between 150 and 70 ka (Bohrson et al., 1996). Three basalts were analyzed for (234U)/(238U), and all are within 1.4% of secular equilibrium. Present-day (230Th)/(232Th) and (238U)/(232Th) for two of the alkalic basalts are somewhat lower than those of the silicic rocks. Of these, the ≤5 ka basalt (sample 90-120) has a well-constrained (230Th)/(232Th)o of 0.96 and an initial 230Th enrichment of 6%. The other sample is close to secular equilibrium [(230Th)/(238U)o = 1.02−1.03] and has a slightly lower initial Th isotope ratio [(230Th)/(232Th)o = 0.94] than that of the ≤5 ka basalt. The remaining two basalts have very low (238U)/(232Th) and exhibit pronounced disequilibria (230Th enrichments of at least 12%), but (230Th)/(232Th)o values permitted by the age range of 70−150 ka bracket those of the other two basalts. U and Th concentrations do not covary simply with indices of differentiation such as Zr concentration. Th isotope signatures for the basaltic and silicic suites are like those observed in oceanic basalts with similar 143Nd/144Nd (Fig. 2).

Genesis of Trachytic Magmas by Recycling of Deep-Level Basaltic Basement: Constraints from U–Th Data

The broadly similar ranges in Sr and Nd (Table 1) as well as Pb isotopic ratios (Bohrson & Reid, 1997) suggest a cogenetic relationship between the postcaldera alkalic basaltic and silicic rocks. In detail, however, the postcaldera trachytes have somewhat lower 87Sr/86Sr and, on average, higher 143Nd/144Nd than the postcaldera basalts (Table 1). Moreover, gaps in SiO2 and K2O coupled with virtually identical abundances of incompatible trace elements such as Zr, Nb and Th between postcaldera alkalic basalts and peralkaline trachytes (Fig. 3) also indicate that the two suites of lavas are not simply related by closed-system fractional crystallization. Major element models, using postcaldera alkalic basalts as parental magmas, consistently yield trachytic daughter products impoverished in K2O compared with observed abundances (Bohrson & Reid, 1997). The Th isotope data corroborate this interpretation because the trachytes have significantly (> 12%) higher (230Th)/(232Th)o and (230Th)/(238U)o than basalts with similar (238U)/(232Th): if anything, in situ decay during differentiation should have decreased (230Th)/(232Th) in the trachytes relative to the parental magmas.

Experimental data indicate that intermediate composition magmas can be generated by partial melting of basalt [see references in Bohrson & Reid, (1997)], and therefore the most reasonable alternative to fractional crystallization is partial melting. Modal equilibrium melting calculations illustrate that moderate degrees of melting (5−15%) of a mafic source with a composition like that of the postcaldera basalts ± cumulates can produce magmas with appropriate incompatible trace element (e.g. Zr, Nb, Th) and K2O abundances (Fig. 3.).

Partial melting of basement with (230Th)/(232Th) like that of the postcaldera alkalic basalts

The postcaldera basalts are broadly similar in major and trace element composition to subaerially exposed precaldera basalts (Bryan, 1976), which suggests that the U–Th characteristics of the postcaldera basalts may also be good analogs for those of the basement proposed to be the source of the trachytic melts. Closed-system partial melting of basaltic basement with Th isotope characteristics like those of the postcaldera basalts would probably produce trachytic magmas with Th isotope ratios ≤0.93, based on the (238U)/(232Th) of the postcaldera basalts and the assumption that the basement is at secular equilibrium. Such ratios are significantly lower than the (230Th)/(232Th) of the trachytes (> 1.10). A possible caveat would be that if melting times are long with respect to the half-life to 230Th, 230Th-enriched melts could acquire somewhat higher (230Th)/(232Th) than their source; however, such a process of dynamic partial melting is unlikely to be applicable for the formation of trachytic melts on Socorro. Thus, unless the silicic rocks have experienced post-eruptive alteration with respect to Th and/or U, generation of the trachytic magmas by melting of relatively low (230Th)/(232Th) basement would require contamination of the trachytic melts by an assimilant with (230Th)/(232Th) ≥ 1.14, the highest value obtained for a trachyte (Fig. 1).

Post-eruptive alteration of the silicic rocks by a seawater-dominated fluid, evident in acid-labile radiogenic Sr (Bohrson & Reid, 1997), potentially could have also increased the Th isotope ratios of the trachytic rocks. Despite an extremely high (230Th)/(232Th) in seawater (∼4−300, Roy-Barman et al., 1996), the low abundance of Th (picograms/gram) necessitates a large water-to-rock ratio (W/R) to significantly change (230Th)/(232Th). For example, a change from 0.95 (basalt) to 1.11 (trachyte) requires a W/R of >2000, assuming an average seawater Th isotope ratio of 100 and a Th concentration of 0.008 ppb (the highest reported for either seawater or mid-ocean ridge hydrothermal fluids; Chen et al., 1986; Chen, 1987). Such a high W/R should result in whole-rock 87Sr/86Sr equal to 0.709 (assuming the fluid has the Sr abundance and isotope composition of seawater) (trend A, Fig. 4), but ratios in unleached samples are all <0.707 (Fig. 4 ). For both whole-rock 87Sr/86Sr and (230Th)/(232Th)o to be the result of post-eruptive fluid interaction, the Th concentration of the fluid would have to be three orders of magnitude greater than abundances typically observed in marine or terrestrial hydrothermal fluids (Sturchio et al., 1986; Chen, 1987). An increase in (230Th)/(232Th) by post-eruptive alteration of the trachytic rocks is therefore very unlikely.

Fig. 1.

(230Th)/(232Th) vs (238U)/(232Th) for silicic and alkalic basaltic postcaldera lavas from Socorro Island. (230Th)/(232Th) values for the silicic rocks are corrected to time of eruption, (230Th)/(232Th)o; range of initial (230Th)/(232Th) for the basaltic rocks permitted by age-corrections of 70−150 kyr shown except for sample 90−120, which is ≤5 kyr old. Field for East Pacific Rise mid–ocean ridge basalt (EPR MORB) (21°N, Newman et al., 1983 13°N, Ben Othman & Allègre, 1990) shown for comparison.

Fig. 1.

(230Th)/(232Th) vs (238U)/(232Th) for silicic and alkalic basaltic postcaldera lavas from Socorro Island. (230Th)/(232Th) values for the silicic rocks are corrected to time of eruption, (230Th)/(232Th)o; range of initial (230Th)/(232Th) for the basaltic rocks permitted by age-corrections of 70−150 kyr shown except for sample 90−120, which is ≤5 kyr old. Field for East Pacific Rise mid–ocean ridge basalt (EPR MORB) (21°N, Newman et al., 1983 13°N, Ben Othman & Allègre, 1990) shown for comparison.

Table 1.

Isotopic and chemical data for postcaldera lavas from Socorro Island, Mexico

Sample Groupa Age (ka)b SiO2(wt %) Zr (ppm) Sr (ppm)c U (ppm) Th (ppm) 232Th/238U(Th/U) 232Th/230Th × 104 
Silicic          
90-122 Ref 62 63.89 465 34.76 1.58 5.12 3.348 (3.240) 18.00 ± 9 
         17.91 ± 24 
90-29 Ref 68 64.64 494 27.72 1.76 5.61 3.297 (3.191) 17.51 ± 5 
         18.02 ± 8 
90-91 Ref 15 64.12 506 72.97 1.85 6.18 3.455 (3.343) 17.31 ± 7 
         17.37 ± 10 
91-17 Enr 182 66.56 829 8.81 3.19 9.84 3.189 (3.086) 18.84 ± 7 
         18.93 ± 8 
90-99 Ref 41 67.17 1301 2.55 4.98 15.91 3.301 (3.195) 17.67 ± 9 
         17.60 ± 7 
90-101 Ref 30 (?) 67.06 1337 — 5.22 16.65 3.294 (3.188) 17.85 ± 9 
         17.90 ± 22 
90-94 Ref 152 66.71 1543 2.20 6.11 20.22 3.422 (3.311) 19.53 ± 8 
         19.64 ± 8 
Basaltic          
91-29 ≥70 [150] 48.52 203 — 0.70 2.30 3.389 (3.281) 19.97 ± 9 
         19.90 ± 10 
91-55 ≥70 [150] 48.19 228 — 0.55 2.54 4.789 (4.633) 24.67 ± 10 
         25.14 ± 25 
91-4 ≥70 [150] 48.03 272 — 0.84 3.67 4.532 (4.385) 24.30 ± 17 
         23.73 ± 31 
90-120 ≤5 52.49 328 — 1.11 3.71 3.458 (3.345) 19.42 ± 9 
         19.41± 23 
Sample Groupa Age (ka)b SiO2(wt %) Zr (ppm) Sr (ppm)c U (ppm) Th (ppm) 232Th/238U(Th/U) 232Th/230Th × 104 
Silicic          
90-122 Ref 62 63.89 465 34.76 1.58 5.12 3.348 (3.240) 18.00 ± 9 
         17.91 ± 24 
90-29 Ref 68 64.64 494 27.72 1.76 5.61 3.297 (3.191) 17.51 ± 5 
         18.02 ± 8 
90-91 Ref 15 64.12 506 72.97 1.85 6.18 3.455 (3.343) 17.31 ± 7 
         17.37 ± 10 
91-17 Enr 182 66.56 829 8.81 3.19 9.84 3.189 (3.086) 18.84 ± 7 
         18.93 ± 8 
90-99 Ref 41 67.17 1301 2.55 4.98 15.91 3.301 (3.195) 17.67 ± 9 
         17.60 ± 7 
90-101 Ref 30 (?) 67.06 1337 — 5.22 16.65 3.294 (3.188) 17.85 ± 9 
         17.90 ± 22 
90-94 Ref 152 66.71 1543 2.20 6.11 20.22 3.422 (3.311) 19.53 ± 8 
         19.64 ± 8 
Basaltic          
91-29 ≥70 [150] 48.52 203 — 0.70 2.30 3.389 (3.281) 19.97 ± 9 
         19.90 ± 10 
91-55 ≥70 [150] 48.19 228 — 0.55 2.54 4.789 (4.633) 24.67 ± 10 
         25.14 ± 25 
91-4 ≥70 [150] 48.03 272 — 0.84 3.67 4.532 (4.385) 24.30 ± 17 
         23.73 ± 31 
90-120 ≤5 52.49 328 — 1.11 3.71 3.458 (3.345) 19.42 ± 9 
         19.41± 23 
Sample (238U)/(232Th)d (230Th)/(232Th)d (230Th)/(232Th)0d,(230Th)/(238U)0d,(234U)/(238U)f 143Nd/144Nd 87Sr/86Srg 
90-122 0.937 1.033 ± 5 1.107 1.181 — 0.51295 0.70304 
90-29 0.952 1.053 ± 3 1.140 1.198 0.995 0.51294 0.70306 
90-91 0.908 1.072 ± 4 1.097 1.208 — 0.51295 0.70306 
91-17 0.984 0.984 ± 3 0.986 1.002 0.989 0.51294 0.70313 
90-99 0.950 1.054 ± 3 1.102 1.160 0.999 0.51291 0.70323 
90-101 0.952 1.041 ± 5 1.069 1.123 0.995 0.51293 — 
90-94 0.916 0.949 ± 3 1.049 1.146 1.003 0.51291 0.70319 
91-29 0.925 0.932 ± 3 0.939 [0.953] 1.015 [1.031] 0.994 0.51294 — 
91-55 0.655 0.751 ± 3 0.838 [1.038] 1.280 [1.584] 0.986 0.51295 0.70309 
91-4 0.692 0.769 ± 5 0.839 [0.998] 1.212 [1.442] 0.991 0.51292 0.70311 
90-120 0.907 0.957 ± 4 0.959 1.058 — 0.51292 0.70310 
Sample (238U)/(232Th)d (230Th)/(232Th)d (230Th)/(232Th)0d,(230Th)/(238U)0d,(234U)/(238U)f 143Nd/144Nd 87Sr/86Srg 
90-122 0.937 1.033 ± 5 1.107 1.181 — 0.51295 0.70304 
90-29 0.952 1.053 ± 3 1.140 1.198 0.995 0.51294 0.70306 
90-91 0.908 1.072 ± 4 1.097 1.208 — 0.51295 0.70306 
91-17 0.984 0.984 ± 3 0.986 1.002 0.989 0.51294 0.70313 
90-99 0.950 1.054 ± 3 1.102 1.160 0.999 0.51291 0.70323 
90-101 0.952 1.041 ± 5 1.069 1.123 0.995 0.51293 — 
90-94 0.916 0.949 ± 3 1.049 1.146 1.003 0.51291 0.70319 
91-29 0.925 0.932 ± 3 0.939 [0.953] 1.015 [1.031] 0.994 0.51294 — 
91-55 0.655 0.751 ± 3 0.838 [1.038] 1.280 [1.584] 0.986 0.51295 0.70309 
91-4 0.692 0.769 ± 5 0.839 [0.998] 1.212 [1.442] 0.991 0.51292 0.70311 
90-120 0.907 0.957 ± 4 0.959 1.058 — 0.51292 0.70310 

SiO2, Zr, Sr, 143Nd/144Nd and 87Sr/86Sr data presented by Bohrson & Reid (1995, 1997). Th isotope, U isotope and U–Th concentration data generated at UCLA. U and Th concentration data by isotope dilution; elements separated by anion exchange chromatography. Thorium isotopic composition analyzed as metal using a dual collector array composed of Faraday cup and ion counter. Reported uncertainties are 2σ. 232Th/230Th ± 2σ obtained during this study on a standard solution and international reference samples for chemical analysis are (1) UC Santa Cruz Th standard: 170587 ± 0.26% (n = 35) [compare values of (17.03 ± 9) × 104 and 17.05 × 104 reported by Reid (1995)]; (2) AGV1 199040 ± 0.20% (n = 5); (3) JB1 332302 ± 0.96% (n = 5) [compare values of 20.06 × 104 and 33.54 × 104, respectively, reported by Reid (1995)].

a

Group refers to designations defined in previous work. For silicic samples: Ref, samples have relative REE abundances that can be explained by fractional crystallization (FC); Enr, samples have negative Ce anomalies and abundances of REE greater than those predicted by FC (Bohrson & Reid, 1997). For basaltic samples: 1, samples have REE, Y, P2O5 consistent with FC; 2, samples have negative Ce anomalies and abundances of REE, Y, P2O5 in excess of those predicted by FC; 3, samples lack negative Ce anomalies and have abundances of P2O5 and middle REE in excess of those predicted by FC (Bohrson & Reid, 1995).

b

Age constraints for silicic lavas from 40Ar/39Ar technique on alkali feldspar separates (Bohrson et al., 1996), except age of 90-101, which is estimated from stratigraphic relations. Age constraints for basaltic lavas estimated from stratigraphic relations, except age of 90-120, which is constrained using 14C age of lacustrine deposit underlying vent (Farmer et al., 1993).

c

Sr abundances are isotope dilution analyses of acid-leached silicic peralkaline whole rocks.

d

Activity ratios denoted by parentheses and calculated using the decay constants λ232 = 4.9475 × 10−11 yr−1, λ230 = 9.1952 × 10−6 yr−1, λ238 = 1.55125 × 10−10 yr−1, and using the weighted mean of replicate 232Th/230Th.

e

Initial ratios calculated using reported ages. For 91-29, 91-55, and 91-4 (230Th)/(232Th)o and (230Th/238U)o corrected to 70 ka, values in brackets are for corrections to 150 ka.

f

234U/238U corrected for mass discrimination to 238U/235U of 137.88. λ234 = 2.835 × 10−6 yr−1.(234U)/(238U) for USGS reference samples for chemical analysis, which should all be in equilibrium, are 1.002 and 0.999 for BCR1 and AGV1, respectively.

g

Sr isotope ratios are, for silicic compositions, those of age-corrected, acid-leached alkali feldspar, and for basaltic compositions, those of acid-leached whole rocks.

Possible magmatic contaminants with high (230Th)/(232Th) are fresh (Fig. 1) or hydrothermally altered East Pacific Rise mid-ocean ridge basalt (EPR MORB) and hydrothermally altered basaltic basement associated with Socorro. Fresh EPR MORB is an unlikely contaminant. For combined AFC (DePaolo, 1981), the large amount of fresh MORB required to account for the change in Th isotope ratios from 0.95 to 1.11 {r (rate of mass assimilated/rate of mass crystallized) = 0.75; MORB model parameters (230Th)/(232Th) ≤ 1.45; [Th] ≤ 1 ppm; Newman et al., 1983; Ben Othman & Allègre, 1990} would produce MORB-like Sr (trend B, Fig. 4) and Nd isotope signatures (∼0.5131−0.5132 and ∼0.7024−0.7025, respectively; White et al., 1987) in the trachytic magmas; clearly this has not occurred. Because seawater has a high (230Th)/(232Th), the Th isotope signature of hydrothermally altered MORB might be higher than that of fresh MORB. However, as in the case of post-eruptive alteration discussed above, the low abundance of Th in seawater precludes any significant change in (230Th)/(232Th) solely by Th exchange at reasonable W/R. For example, average hydrothermally altered MORB, which has a Sr isotope ratio of 0.70475 (Staudigel et al., 1995), corresponds to a W/R of ∼9. For this W/R, the Th isotope ratio of altered MORB will only be slightly higher than fresh MORB (e.g. from 1.45 to 1.46), and the Nd isotope ratio will remain unchanged. As in the case of fresh MORB, a large amount of assimilated hydrothermally altered MORB is required, and the trachytic magmas therefore should be characterized by MORB-like 143Nd/144Nd. Hydrothermal alteration commonly leads to U gains in altered MORB (Albarède & Michard, 1986; Staudigel et al., 1995) that could potentially generate elevated Th isotope ratios by in situ decay. Even given the potentially high (230Th)/(232Th) for old, altered MORB, mass balance constraints largely preclude the trachytic magmas acquiring their higher Th isotope ratios from assimilation of such material; for hydrothermally altered MORB in secular equilibrium with (238U)/(232Th) of 9.5 {[U] = 0.275, [Th] = 0.088; Super Composite of Staudigel et al. (1995)}, the moderate rates of assimilation required to achieve the Th isotope signatures of the trachytes would yield 87Sr/86Sr much more radiogenic than the magmatic ratios (trend C, Fig. 4), assuming the 87Sr/86Sr of the assimilant is represented by that of average hydrothermally altered MORB. In addition, assimilation of such high (238U)/(232Th) material would probably generate melts with significantly higher (238U)/(232Th) than observed in the trachytes. Assimilation of hydrothermally altered basaltic basement associated with the Socorro edifice might also increase magmatic Th isotope ratios. Because of the relatively high Sr/U ratio of seawater, the change in Th/U required to increase the Th isotope ratio of Socorro basaltic basement by U enrichment and in situ decay would be very likely to yield basement with relatively radiogenic 87Sr/86Sr. As in the case of altered MORB, assimilation of such material would yield trachytic magmas with higher Sr isotope ratios than observed. In summary, there is no reasonable mechanism by which the trachytes can be derived from basement with Th isotope characteristics like those of the postcaldera alkalic basalts.

Fig. 2.

(230Th)/(232Th) vs 143Nd/144Nd for silicic and alkalic basaltic rocks from Socorro as well as a number of other ocean islands. Gray band represents the ’mantle array‘ (Gill & Condomines, 1992). Comparison data from Newman et al., (1983), Macdougall & Lugmair, (1986), Ben Othman & Allègre, (1990), Sigmarsson et al., (1991), Gill & Condomines, (1992, and references therein), Hemond et al., (1994), Turner et al., (1997), and Widom et al., (1997).

Fig. 2.

(230Th)/(232Th) vs 143Nd/144Nd for silicic and alkalic basaltic rocks from Socorro as well as a number of other ocean islands. Gray band represents the ’mantle array‘ (Gill & Condomines, 1992). Comparison data from Newman et al., (1983), Macdougall & Lugmair, (1986), Ben Othman & Allègre, (1990), Sigmarsson et al., (1991), Gill & Condomines, (1992, and references therein), Hemond et al., (1994), Turner et al., (1997), and Widom et al., (1997).

Partial melting of basement with (230Th)/(232Th) values similar to those of the silicic peralkaline postcaldera rocks

Because it is difficult to reconcile the relatively high (230Th)/(232Th) of the trachytes with melting of a source with (230Th)/(232Th) values as low as those of the postcaldera basalts, it is likely that the Th isotope signature of the basaltic basement is similar to or higher than those of the postcaldera trachytes from Socorro. Although the differences are small, it is notable that the Sr and Nd isotope signatures of the postcaldera trachytes and postcaldera basalts are also distinct (Table 1). The Th/U and Th isotope ratios in silicic rocks that erupted over more than 150 kyr constitute a relatively narrow range, which suggests that the process of partial melting is fairly reproducible and samples a source that is apparently relatively homogeneous with respect to Th isotopes over the length scale of melting.

If such high (230Th)/(232Th) basement is appropriate, then partial melting of a source that has Th/U in secular equilibrium with the (230Th)/(232Th) of the trachytes should reasonably produce the Th/U (3.19−3.34, exclusive of 91-17) and therefore the (230Th) enrichments (up to 21%) of postcaldera trachytes. Bulk U and Th partition coefficients calculated from mineral–melt distribution coefficients (Mahood & Stimac, 1990) for a phase assemblage in equilibrium with the peralkaline trachytes [modal abundances and assemblage constrained from quantitative fractional crystallization models (Bohrson & Reid, 1997)] yield DTh/DU of 0.4 (DTh = 0.004; DU = 0.01); with these partition coefficients, 5−10% equilibrium melting (Bohrson & Reid, 1997) of a source that is in secular equilibrium with (230Th)/(232Th) ≈ 1.1 generates Th/U ratios of 2.9−3.1 and (230Th) enrichments of up to 11% (Fig. 5); these values will not be significantly changed by the amount of fractional crystallization inferred to have affected the trachytes. Decreasing DTh/DU to 0.33, which is similar to the ratio estimated for silicic rocks from Hekla (DTh = 0.005, DU = 0.015; Sigmarsson et al., 1992a), expands the Th/U range to 3.28 and increases the (230Th) enrichments to 19%. Most of the range in Th/U and (230Th) enrichments therefore may be explained by partial melting of basement with (230Th)/(232Th) and (238U)/(232Th) of ≈ 1.1 (Fig. 5). Melting of a source having (230Th)/(232Th) higher than those of the trachytes (> 1.1) would require still smaller degrees of partial melting and/or more extreme DTh/DU values than those presented here.

Fig. 3.

Modal equilibrium partial melting models of mafic sources for selected elements and oxides. Results for three hypothetical sources are illustrated. Curves 1 (continuous) and 2 (long–dash): compositions are those estimated for cumulates derived by fractionation of alkalic basalt from Socorro (Bohrson & Reid, 1995). Curve 3: composition is that of the least differentiated alkali basalt (Bohrson & Reid, 1995) sampled on Socorro Island. Bulk partition coefficients used in partial melting calculations (reported in figure) estimated from published values (Mahood & Stimac, 1990; Furman et al., 1991). Numbers in small boxes represent bulk compositions of starting materials. Numbers next to tick marks on curves are percent partial melt; some labels are omitted for clarity. Most rhyolites plot at >800 ppm Zr. Basalt and peralkaline trachyte and rhyolite fields represent compositional ranges for extrusive rocks from Socorro; cumulate field represents a range of hypothetical mafic cumulate compositions estimated for Socorro. The curves that terminate in (a) do so at 0.1% melt. Gray curves in (b) illustrate fractional crystallization models using the same bulk partition coefficients as in the partial melting models; dots represent 50, 80, 90 and 95% fractional crystallization, and one on each curve is labeled for clarity (italics). Symbols in (c) same as in Fig. 1.

Fig. 3.

Modal equilibrium partial melting models of mafic sources for selected elements and oxides. Results for three hypothetical sources are illustrated. Curves 1 (continuous) and 2 (long–dash): compositions are those estimated for cumulates derived by fractionation of alkalic basalt from Socorro (Bohrson & Reid, 1995). Curve 3: composition is that of the least differentiated alkali basalt (Bohrson & Reid, 1995) sampled on Socorro Island. Bulk partition coefficients used in partial melting calculations (reported in figure) estimated from published values (Mahood & Stimac, 1990; Furman et al., 1991). Numbers in small boxes represent bulk compositions of starting materials. Numbers next to tick marks on curves are percent partial melt; some labels are omitted for clarity. Most rhyolites plot at >800 ppm Zr. Basalt and peralkaline trachyte and rhyolite fields represent compositional ranges for extrusive rocks from Socorro; cumulate field represents a range of hypothetical mafic cumulate compositions estimated for Socorro. The curves that terminate in (a) do so at 0.1% melt. Gray curves in (b) illustrate fractional crystallization models using the same bulk partition coefficients as in the partial melting models; dots represent 50, 80, 90 and 95% fractional crystallization, and one on each curve is labeled for clarity (italics). Symbols in (c) same as in Fig. 1.

Fig. 4.

Models illustrating (230Th)/(232Th) vs 87Sr/86Sr for (a) post-eruptive alteration where tick marks represent water-to-rock (W/R) ratios, (b) AFC involving fresh MORB as contaminant, and (c) AFC involving MORB hydrothermally altered by U addition as contaminant. Th-Sr isotopic range for EPR MORB also shown. Arrows for altered MORB and seawater mark the 87Sr/86Sr values for these contaminants whereas (230Th)/(232Th) is off-scale and is noted in parentheses. Silicic and basaltic denote fields for Socorro Island postcaldera extrusive rocks; 87Sr/86Sr of basaltic rocks are acid-leached whole rocks, those of the silicic rocks are age-corrected, acid-leached feldspars. (230Th)/(232Th) are age-corrected whole-rock values. (See text for additional details.)

Fig. 4.

Models illustrating (230Th)/(232Th) vs 87Sr/86Sr for (a) post-eruptive alteration where tick marks represent water-to-rock (W/R) ratios, (b) AFC involving fresh MORB as contaminant, and (c) AFC involving MORB hydrothermally altered by U addition as contaminant. Th-Sr isotopic range for EPR MORB also shown. Arrows for altered MORB and seawater mark the 87Sr/86Sr values for these contaminants whereas (230Th)/(232Th) is off-scale and is noted in parentheses. Silicic and basaltic denote fields for Socorro Island postcaldera extrusive rocks; 87Sr/86Sr of basaltic rocks are acid-leached whole rocks, those of the silicic rocks are age-corrected, acid-leached feldspars. (230Th)/(232Th) are age-corrected whole-rock values. (See text for additional details.)

Isotopic and Elemental Variations within the Silicic Suite: Evidence for Recycling of Shallow-Level Basement

Most of the silicic lavas exhibit significant initial (230Th) enrichments. The exception is sample 91-17, which is close to being in secular equilibrium (Fig. 1), represents the oldest postcaldera lava sampled, and is the only one which has REE characteristics (e.g. negative Ce anomaly) that have been attributed to assimilation of small amounts of hydrothermal sediments. The lack of significant disequilibria in this sample may therefore reflect its unique petrogenetic history among postcaldera silicic samples. Hydrothermal sediments have a wide range of Th/U but most commonly are relatively U rich (e.g. Shimmield & Price, 1988; German et al., 1993; Mills et al., 1993; Cocherie et al., 1994), which may account for the slightly higher (238U)/(232Th) of 91-17 compared with the other silicic postcaldera rocks. In addition, 91-17 could have been derived from the same batch of magma as that responsible for pre- and syncaldera deposits with the same unusual REE characteristics, in which case, the lower (230Th)/(232Th) is expected by in situ decay since formation of those magmas >370 kyr ago.

Although the uncertainties in (230Th)/(232Th)o of individual trachytes and rhyolites overlap, there is, exclusive of sample 91-17, a general trend towards decreasing (230Th)/(232Th)o and (230Th)/(238U)o with indices of differentiation such as increasing Th concentration (Fig. 6). Because the postcaldera silicic lavas erupted over a period of >150 kyr, the magmas may have had somewhat different petrogenetic histories, but the differences in (230Th)/(232Th)o between the trachytes and rhyolites, coupled with evidence that the suite is related by fractional crystallization (Bohrson & Reid, 1997), could reflect magma residence times for the rhyolites of 40−50 kyr. As illustrated in an earlier section, it is unlikely that assimilation could have raised the (230Th)/(232Th) of the rhyolites, and consequently, residence times inferred in this fashion are maxima. These residence times are comparable with the repose periods (30−35 kyr) between silicic pre- and syncaldera eruptions on Socorro (Bohrson et al., 1996), but the similarity may be fortuitous. Whereas the post-, syn- and precaldera silicic magma chambers are demonstrably compositionally zoned (Bohrson et al., 1996), the total volume of the postcaldera phase is estimated to be an order-of-magnitude less than the combined volume of the pre- and syncaldera phases (0.8 vs ≥ 8 km3, respectively), despite similar eruptive durations [180 kyr vs 170 kyr, respectively (Bohrson et al., 1996)]. In general, eruption volume and apparent residence times of zoned silicic magma bodies are positively correlated (Trial & Spera, 1990), which suggests that the actual residence times of the postcaldera rhyolites may be appreciably shorter than those of the pre- and syncaldera ones. Therefore, the difference in Th isotope signatures between postcaldera trachytes and rhyolites may not solely reflect the effect of in situ decay.

Fig. 5.

(230Th)/(232Th) vs (238U)/(232Th) illustrating Th/U and (230Th) enrichments for modal equilibrium partial melts generated from a basaltic source in secular equilibrium with (230Th)/(232Th) ≈ 1.1. Labeled tick marks represent percent partial melting for DTh/DU = 0.4 and 0.33. Symbols same as in Fig. 1.

Fig. 5.

(230Th)/(232Th) vs (238U)/(232Th) illustrating Th/U and (230Th) enrichments for modal equilibrium partial melts generated from a basaltic source in secular equilibrium with (230Th)/(232Th) ≈ 1.1. Labeled tick marks represent percent partial melting for DTh/DU = 0.4 and 0.33. Symbols same as in Fig. 1.

Within the postcaldera silicic suite from Socorro, 143Nd/144Nd decreases whereas 87Sr/86Sr increases with differentiation; similar trends are not evident in the pre- and syncaldera silicic rocks. The range in Sr isotope ratios observed in the silicic postcaldera lavas may reflect, at least in part, in situ decay of 87Rb that occurred in the silicic magma chamber during formation and residence of the rhyolites. Most of the silicic magmas probably had 87Rb/86Sr ≤ 250 [estimated from isotope dilution Sr and Rb data of acid-leached whole-rock analyses (Bohrson & Reid, 1997)], and therefore, for a maximum magma residence time of 40 kyr, the 87Sr/86Sr of rhyolitic magmas could have increased by no more than ∼0.0001. Even this fairly generous estimate of the effect of in situ decay is still less than the maximum difference in Sr isotope ratios, ∼0.0002 (Table 1), and therefore some assimilation of material characterized by relatively radiogenic Sr isotope signatures is probably required. If, as seems likely, rhyolite magma residence times are shorter than implied by the range in Th isotope signatures of the silicic suite, then the Th isotope signatures are also at least partly influenced by open-system processes. Consequently, the systematic variations in Sr, Th and Nd isotopes with differentiation imply a process of AFC. Available assimilants include silicic basement as well as those described above (fresh and altered MORB and mafic basement formed in association with magmatism on Socorro); probable assimilants exclude MORB because the Nd isotope ratios of the rhyolites are lower rather than higher than those of the trachytes.

Representative AFC models for derivation of the rhyolites from the trachytes have been constrained using Th isotope ratios estimated for silicic and basaltic basement at secular equilibrium (Fig. 5) and Th concentrations observed in the postcaldera samples. It is evident from the general similarity of Th/U among trachytes and rhyolites that the assimilant must be dominated by components with Th/U similar to those of the trachytes. Th/U values for all silicic rocks from Socorro (Bohrson & Reid, 1997) suggest that derivative basement has (230Th)/(232Th) in secular equilibrium with (238U)/(232Th) of ∼0.9−1.0. The two postcaldera basalts analyzed in this study that are near secular equilibrium also have relatively low (230Th)/(232Th) as well as Th/U like those of the silicic rocks, possibly suggesting assimilation of relatively young basaltic basement with (230Th)/(232Th) of ∼0.90. Basement with relatively low (230Th)/(232Th), derived from postcaldera mafic and all phases of silicic magmatism, therefore might be expected at the shallow levels where the silicic magma chamber is inferred to reside beneath Socorro. For the purpose of delimiting its potential effect, the entire range of (230Th)/(232Th)o is assumed to be the result of assimilation. The best-fit models to the Th data yield r of 0.2 for silicic basement and 0.45 for basaltic basement (Fig. 6a). These r values can be used to infer Sr and Nd isotope ratios of the assimilants by fitting AFC curves to the observed trends (Fig. 6b, c). Sr and Nd concentrations in the assimilants are also taken to be those observed in Socorro lavas (Bohrson & Reid, 1995; Bohrson & Reid, 1997). The Nd isotope ratios obtained for model silicic and basaltic basement (143Nd/144Nd = 0.51285 and 0.51288, respectively) are within uncertainty of those for the two oldest dated silicic precaldera ignimbrites (143Nd/144Nd = 0.51290−0.51287; Bohrson & Reid, 1997), which demonstrates the existence of precaldera silicic basement with relatively unradiogenic Nd isotope signatures. The Sr isotope compositions obtained for both the model silicic basement (87Sr/86Sr = 0.7042) and the model basaltic basement (87Sr/86Sr = 0.70323) are higher than magmatic ratios obtained for Socorro lavas. These values will be somewhat lower if in situ decay during magma residence has played an important role in the observed range of 87Sr/86Sr. Nevertheless, the relatively radiogenic Sr isotope compositions inferred from the AFC calculations suggest that silicic and possibly basaltic basement may have undergone hydrothermal alteration. The low Sr concentration of the silicic rocks (Table 1) would make associated basement particularly sensitive to this effect. An additional source of radiogenic Sr for silicic basement might be in situ decay because relatively high 87Rb/86Sr (up to 250) could have permitted significant increases in 87Sr/86Sr (0.0011) in 300 kyr. The likelihood that pre- and syncaldera silicic basement had the relatively radiogenic Sr signatures inferred from the model AFC calculations, coupled with the lower melting temperature of such basement relative to basaltic basement, suggests it is the more likely assimilant. Moreover, it seems probable that such silicic basement will be a more important constituent of the shallow crust beneath Socorro than mafic basement related to the postcaldera basaltic phase of eruption.

Fig. 6.

Th, Nd, and Sr isotopic ratios vs Th abundance in silicic postcaldera samples from Socorro Island. Correlation coefficients (R) for negative data trends in Th and Nd isotopes vs Th are 0.84 (excluding 91−17) and 0.90, respectively, and for positive data trend in Sr isotope vs Th is 0.92, suggesting a high probability that the data are correlated. Uncertainties in (a) represent a combination of uncertainties in isotopic measurements and Ar ages and are maxima; for (b) and (c), uncertainties are estimated reproducibility. Curves illustrate AFC trends for model silicic (continuous line) and basaltic basement (dashed line). Tick marks (thick, silicic; thin, basaltic) represent proportion of residual liquid. r values are best-fit results for the Th isotope data (see text for discussion). Basaltic basement associated with postcaldera mafic magmatism is assumed to have a Th isotope ratio ∼0.91 and to be at secular equilibrium. Th/U ratios of postcaldera silicic rocks are assumed to be representative of pre- and syncaldera rocks, and associated basement is therefore assumed to have (230Th)/(232Th) of 0.95 and be at secular equilibrium. Other model parameters are: magma [Th] = 5 ppm, (230Th)/(232Th) = 1.11; [Nd] = 50 ppm, 143Nd/144Nd = 0.51295; [Sr] = 13 ppm, 87Sr/86Sr = 0.7030; silicic basement [Th] = 16 ppm; [Nd]; = 120 ppm, 143Nd/144Nd = 0.51285; [Sr] = 4 ppm, 87Sr/86Sr = 0.7042; basaltic basement [Th]; = 4 ppm; [Nd]; = 50 ppm, 143[Nd]/144Nd = 0.51288; [Sr] = 400 ppm, 87Sr/86Sr = 0.70323. Symbols same as in Fig. 1.

Fig. 6.

Th, Nd, and Sr isotopic ratios vs Th abundance in silicic postcaldera samples from Socorro Island. Correlation coefficients (R) for negative data trends in Th and Nd isotopes vs Th are 0.84 (excluding 91−17) and 0.90, respectively, and for positive data trend in Sr isotope vs Th is 0.92, suggesting a high probability that the data are correlated. Uncertainties in (a) represent a combination of uncertainties in isotopic measurements and Ar ages and are maxima; for (b) and (c), uncertainties are estimated reproducibility. Curves illustrate AFC trends for model silicic (continuous line) and basaltic basement (dashed line). Tick marks (thick, silicic; thin, basaltic) represent proportion of residual liquid. r values are best-fit results for the Th isotope data (see text for discussion). Basaltic basement associated with postcaldera mafic magmatism is assumed to have a Th isotope ratio ∼0.91 and to be at secular equilibrium. Th/U ratios of postcaldera silicic rocks are assumed to be representative of pre- and syncaldera rocks, and associated basement is therefore assumed to have (230Th)/(232Th) of 0.95 and be at secular equilibrium. Other model parameters are: magma [Th] = 5 ppm, (230Th)/(232Th) = 1.11; [Nd] = 50 ppm, 143Nd/144Nd = 0.51295; [Sr] = 13 ppm, 87Sr/86Sr = 0.7030; silicic basement [Th] = 16 ppm; [Nd]; = 120 ppm, 143Nd/144Nd = 0.51285; [Sr] = 4 ppm, 87Sr/86Sr = 0.7042; basaltic basement [Th]; = 4 ppm; [Nd]; = 50 ppm, 143[Nd]/144Nd = 0.51288; [Sr] = 400 ppm, 87Sr/86Sr = 0.70323. Symbols same as in Fig. 1.

Smaller r values would be required if the difference in Th isotope ratios between trachytes and rhyolites is the result of in situ decay as well as assimilation. This, in turn, requires more extreme Sr and Nd isotopic values for the basement. Nd isotope values lower than those obtained for the model silicic basement have not been observed in volcanic rocks from Socorro, which suggests that the lower (230Th)/(232Th)o in the rhyolites is due largely to assimilation, and therefore that magmatic residence times are relatively short.

The absence of recognizable AFC trends among pre- and syncaldera trachytes and rhyolites from Socorro may reflect the lack of resolvable isotopic contrast between magma and assimilant. Distinctions between 87Sr/86Sr and (230Th)/(232Th) of the proposed assimilant and the postcaldera magmas may be partly the result of aging of the basement. The Sr isotope ratio of silicic basement apparently became significantly more radiogenic because of its high Rb/Sr and susceptibility to seawater-rock interaction. The Th isotope ratio of this basement became less radiogenic because (230Th) enrichments led to lower (230Th)/(232Th) by in situ decay. The development of these isotopic distinctions, which potentially were absent during the pre- and syncaldera phases, may account for the recognizable record of shallow basement assimilation in the postcaldera silicic rocks. This underscores the potential difficulty of diagnosing the process of volcanic basement assimilation in the ocean basins because, in most cases, significant isotopic distinctions are probably lacking.

Th Isotopic Variations within the Basaltic Suite: Evidence for Mantle Heterogeneity and Open-System Processes?

The two postcaldera basalts that have (230Th)/(238U)o close to secular equilibrium have (230Th)/(232Th)o of ∼0.95, which is at least 15% lower than that inferred for the mafic source of the trachytic magmas. It is possible that the same initial (230Th)/(232Th) applies to the other two basalts, but large 230Th enrichments make inferences about their initial Th isotope ratios very sensitive to uncertainties in the eruption ages. The apparent variation in Th isotope ratios between postcaldera basalts and the mafic source of the trachytic magmas is similar in magnitude to that which characterizes mantle associated with the East Pacific Rise (e.g. Ben Othman & Allègre, 1990), and therefore the range may reflect mantle heterogeneity. It is also possible, although less likely, that the Th isotope ratios of the postcaldera basalts could be the result of long residence times and/or assimilation of material with low (230Th)/(232Th). The residence times required for these basalts to have been formed from mantle with Th isotope characteristics like those inferred for the source of the trachytes appear long given that there were ∼20 basaltic eruptions during an 80 kyr period: the average repose interval of 4 kyr is much shorter than the apparent residence times (∼150 kyr). Availability of low (230Th)/(232Th) basement is probably limited to silicic basement, and a large mass of assimilant is required (e.g. r ≥ 0.5) to lower the Th isotope ratios to those of the postcaldera basalts.

The most surprising characteristic of the postcaldera basalts is the large range in (238U)/(232Th) overall and the extremely low (238U)/(232Th) obtained for two of them. For these two samples, this cannot be an analytical artifact because present-day (230Th)/(232Th) values reflect decay in a low (238U)/(232Th) environment. Based on less accurate ICP-MS data (Bohrson & Reid, 1995), the range in Th/U obtained in this study appears to be like that of the suite of postcaldera basalts as a whole. In addition, the range in Th/U does not correspond in any simple way to other chemical characteristics that have been attributed to open-system processes. U/Zr ratios in these two samples are lower than from those of the lower Th/U basalts, which considered together with the low present-day (230Th)/(232Th), might suggest U loss from these basalts near time of eruption.

Summary

Silicic postcaldera rocks from Socorro Island, Mexico, have relatively restricted ranges of (230Th)/(232Th)o and initial (230Th) enrichments. The range of (230Th)/(232Th)o of the postcaldera basalts also may be relatively restricted but is lower than that of the trachytic rocks; initial (230Th) enrichments vary widely. Postcaldera trachytic magmas form by partial melting of basaltic basement, but closed-system partial melting of a source with Th isotope characteristics like those of the postcaldera basalts is precluded: neither assimilation of a high Th isotope-component nor post-eruptive alteration provides a satisfactory explanation for the relatively high (230Th)/(232Th) of the trachytes. Instead, the source of the trachytic melts, which is basaltic basement that presumably formed during the voluminous precaldera basaltic phase of eruption on Socorro, is probably characterized by Th isotope ratios similar to those observed for the trachytic rocks. Partial melting of such basement in secular equilibrium with (230Th)/(232Th) ≈ 1.1 can generate the range of Th/U observed in the silicic rocks. In general, rhyolites have lower (230Th)/(232Th)o than trachytes do, which could reflect assimilation and/or in situ decay. Although maximum residence times of rhyolitic magmas are delimited to be 40−50 kyr, the possibility of shorter residence times, coupled with systematic variations in Nd and Sr isotopes with indices of differentiation, indicates that the rhyolites are related to the trachytes by AFC. Based upon constraints from Sr, Nd and Th isotope characteristics, the most likely assimilant is volcanic basement that formed during the silicic phases of magmatism on Socorro. The record of basement assimilation on Socorro is, in part, evident because aging of silicic basement apparently allowed development of isotopic distinctions between contaminant and magma. This might suggest that ocean islands characterized by silicic magmatism may be the best locations at which to explore the process of assimilation on ocean islands. The prominent roles that deep- and shallow-level basement play in the petrogenetic history of evolved magmas from Socorro suggest that recycling of crustal material may be important in the genesis of evolved ocean island magmas.

Acknowledgements

This study was supported by postdoctoral fellowships from the University of California Office of the President and the National Science Foundation (to W.A.B.) and NSF grants (to M.R.R.). We thank Frank Ramos for invaluable assistance in the laboratory. This study benefited from discussions with Jon Davidson, Peter Holden, Kurt Knesel, Frank Ramos, and Frank Spera. Valuable reviews by John Lassiter, Peter Michael, and an anonymous reviewer are greatly appreciated. The first author thanks Heather Trim and Larry Jacobson for their generous hospitality while she generated Th isotope data at UCLA. This is Institute for Crustal Studies Contribution 0267−43CM.

References

Albarède
F.
Michard
A.
Transfer of continental Mg, S, O, and U to the mantle through hydrothermal alteration of the oceanic crust
Chemical Geology
 , 
1986
, vol. 
57
 (pg. 
1
-
15
)
Bailey
D. K.
Cooper
J. P.
Knight
J. L.
Anhydrous melting and crystallization of peralkaline obsidians
Bulletin of Volcanology
 , 
1975
, vol. 
38
 (pg. 
653
-
665
)
Ben Othman
D.
Allègre
C. J.
U–Th isotopic systematics at 13°N east Pacific Ridge segment
Earth and Planetary Science Letters
 , 
1990
, vol. 
98
 (pg. 
129
-
137
)
Bohrson
W. A.
Reid
M. R.
Petrogenesis of alkaline basalts from Socorro Island, Mexico: trace element evidence for contamination of ocean island basalt in the shallow ocean crust
Journal of Geophysical Research
 , 
1995
, vol. 
100
 (pg. 
24555
-
24576
)
Bohrson
W. A.
Reid
M. R.
Genesis of silicic peralkaline volcanic rocks in an ocean island setting by crustal melting and open-system processes: Socorro Island, Mexico
Journal of Petrology
 , 
1997
, vol. 
38
 (pg. 
1137
-
1166
)
Bohrson
W. A.
Reid
M. R.
Grunder
A. L.
Heizler
M. T.
Harrison
T. M.
Lee
J.
Prolonged history of silicic peralkaline volcanism in the eastern Pacific Ocean
Journal of Geophysical Research
 , 
1996
, vol. 
101
 (pg. 
11457
-
11474
)
Bryan
W. B.
Aoki
H.
Iizuka
S.
A basalt–pantellerite association from Isla Socorro Islas Revillagigedo Mexico
Volcanoes and Tectonosphere
 , 
1976
Tokyo
Tokai University Press
(pg. 
75
-
91
)
Chen
J. H.
U, Th, and Pb isotopes in hot springs on the Juan de Fuca ridge
Journal of Geophysical Research
 , 
1987
, vol. 
92
 (pg. 
11411
-
11415
)
Chen
J. H.
Edwards
R. L.
Wasserburg
G. J.
238U234U, and232Th in seawater
Earth and Planetary Science Letters
 , 
1986
, vol. 
80
 (pg. 
241
-
251
)
Cocherie
A.
Calvez
J. Y.
Oudin-Dunlop
E.
Hydrothermal activity as recorded by Red Sea sediments: Sr–Nd isotopes and REE signatures
Marine Geology
 , 
1994
, vol. 
118
 (pg. 
291
-
302
)
DePaolo
D. J.
Trace element and isotopic effects of combined wallrock assimilation and fractional crystallization
Earth and Planetary Science Letters
 , 
1981
, vol. 
53
 (pg. 
189
-
202
)
Farmer
J. D.
Farmer
M. C.
Berger
R.
Radiocarbon ages of lacustrine deposits in volcanic sequences of the Lomas Coloradas area, Socorro Island, Mexico
Radiocarbon
 , 
1993
, vol. 
35
 (pg. 
253
-
262
)
Furman
T.
Frey
F. A.
Park
K.-H.
Chemical constraints on the petrogenesis of mildly alkaline lavas from Vestmannaeyjar, Iceland: the Eldfell (1973) and Surtsey (1963–1967) eruptions
Contributions to Mineralogy and Petrology
 , 
1991
, vol. 
109
 (pg. 
19
-
37
)
German
C. R.
Higgs
N. C.
Thomson
J.
Mills
R.
Elderfield
H.
Blusztajn
J.
Fleer
A. P.
Bacon
M. P.
A geochemical study of metalliferous sediment from the TAG hydrothermal mound, 26°08′N, Mid-Atlantic Ridge
Journal of Geophysical Research
 , 
1993
, vol. 
98
 (pg. 
9683
-
9692
)
Gill
J.
Condomines
M.
Short-lived radioactivity and magma genesis
Science
 , 
1992
, vol. 
257
 (pg. 
1368
-
1376
)
Hemond
C.
Condomines
M.
Fourcade
S.
Allègre
C. J.
Oskarsson
N.
Javoy
M.
Thorium, strontium and oxygen isotopic geochemistry in recent tholeiites from Iceland: crustal influence on mantle-derived magmas
Earth and Planetary Science Letters
 , 
1988
, vol. 
87
 (pg. 
273
-
285
)
Hemond
C.
Hofmann
A. W.
Heusser
G.
Condomines
M.
Raczek
I.
Rhodes
M.
U–Th–Ra systematics in Kilauea and Mauna Loa basalts, Hawaii
Chemical Geology
 , 
1994
, vol. 
116
 (pg. 
163
-
180
)
Macdougall
J. D.
Lugmair
G. W.
Sr and Nd isotopes in basalts from the East Pacific Rise: significance for mantle heterogeneity
Earth and Planetary Science Letters
 , 
1986
, vol. 
77
 (pg. 
273
-
284
)
Mahood
G. A.
Stimac
J. A.
Trace-element partitioning in pantellerites and trachytes
Geochimica et Cosmochimica Acta
 , 
1990
, vol. 
54
 (pg. 
2257
-
2276
)
Mammerickx
J.
Naar
D. F.
Tyce
R. L.
The Mathematician Paleoplate
Journal of Geophysical Research
 , 
1988
, vol. 
93
 (pg. 
3025
-
3040
)
Mills
R.
Elderfield
H.
Thomson
J.
A dual origin for the hydrothermal component in a metalliferous sediment core from the Mid-Atlantic Ridge
Journal of Geophysical Research
 , 
1993
, vol. 
98
 (pg. 
9671
-
9681
)
Newman
S.
Finkel
R. C.
Macdougall
J. D.
230Th–238U disequilibrium systematics in oceanic tholeiites from 21°N on the East Pacific Rise
Earth and Planetary Science Letters
 , 
1983
, vol. 
65
 (pg. 
17
-
33
)
Reid
M. R.
Processes of mantle enrichments and magmatic differentiation in the eastern Snake River Plain: Th isotope evidence
Earth and Planetary Science Letters
 , 
1995
, vol. 
131
 (pg. 
239
-
254
)
Roy-Barman
M.
Chen
J. H.
Wasserburg
G. J.
230Th–232Th systematics in the central Pacific Ocean: the sources and the fates of thorium
Earth and Planetary Science Letters
 , 
1996
, vol. 
139
 (pg. 
351
-
363
)
Shimmield
G. B.
Price
N. B.
The scavenging of U230Th and231Pa during pulsed hydrothermal activity at 20°S, East Pacific Rise
Geochimica et Cosmochimica Acta
 , 
1988
, vol. 
52
 (pg. 
669
-
677
)
Sigmarsson
O.
Hemond
C.
Condomines
M.
Fourcade
S.
Oskarsson
N.
Origin of silicic magma in Iceland revealed by Th isotopes
Geology
 , 
1991
, vol. 
19
 (pg. 
621
-
624
)
Sigmarsson
O.
Condomines
M.
Fourcade
S.
A detailed Th, Sr and O isotope study of Hekla: differentiation processes in an Icelandic volcano
Contribution to Mineralogy and Petrology
 , 
1992
, vol. 
112
 (pg. 
20
-
34
)
Sigmarsson
O.
Condomines
M.
Ibarrola
E.
238U–230Th radioactive disequilibria in historic lavas from the Canary Islands and genetic implications
Journal of Volcanology and Geothermal Research
 , 
1992
, vol. 
54
 (pg. 
145
-
156
)
Staudigel
H.
Davies
G. R.
Hart
S. R.
Marchant
K. M.
Smith
B. M.
Large scale isotopic Sr, Nd, and O isotopic anatomy of altered oceanic crust: DSDP/ODP site 417/418
Earth and Planetary Science Letters
 , 
1995
, vol. 
130
 (pg. 
168
-
185
)
Sturchio
N. C.
Muehlenbachs
K.
Seitz
M. G.
Element redistribution during hydrothermal alteration of rhyolite in an active geothermal system: Yellowstone drill cores Y-7 and Y-8
Geochimica et Cosmochimica Acta
 , 
1986
, vol. 
50
 (pg. 
1619
-
1631
)
Trial
A. F.
Spera
F. J.
Mechanisms for the generation of compositional heterogeneities in magma chambers
Geological Society of America Bulletin
 , 
1990
, vol. 
102
 (pg. 
353
-
367
)
Turner
S.
Hawkesworth
C.
Rogers
N.
King
P.
U–Th isotope disequilibria and ocean island basalt generation in the Azores
Chemical Geology
 , 
1997
, vol. 
139
 (pg. 
145
-
164
)
Wendt
I.
Carl
C.
The statistical distribution of the mean squared weighted deviation
Chemical Geology
 , 
1991
, vol. 
86
 (pg. 
275
-
285
)
White
W. M.
Hofmann
A. W.
Puchelt
H.
Isotope geochemistry of Pacific mid-ocean ridge basalt
Journal of Geophysical Research
 , 
1987
, vol. 
92
 (pg. 
4881
-
4893
)
Widom
E.
Schmincke
H.-U.
Gill
J. B.
Processes and timescales in the evolution of a chemically zoned trachyte: Fogo A, Sao Miguel, Azores
Contributions to Mineralogy and Petrology
 , 
1992
, vol. 
111
 (pg. 
311
-
328
)
Widom
E.
Carlson
R. W.
Gill
J. B.
Schmincke
H.-U
Th–Sr–Nd–Pb isotope and trace element evidence for the origin of the Sao Miguel, Azores, enriched mantle
Chemical Geology
 , 
1997
, vol. 
140
 (pg. 
49
-
68
)