The zircon Hf isotope archive of rapidly changing mantle sources in the south Patagonian retro-arc

Rapid changes in geochemical and isotopic signatures of arc-related magmatic products can be used to trace magmatic processes in subduction zones across many scales, from the regional response of magmatism to large-scale geodynamic changes in the subduction system, to the emplacement of single intrusions. In this contribution, we use the south Patagonian subduction system as a natural laboratory to investigate magmatic processes in continental arcs. We use diverse intrusions and dikes from the retro-arc region at 49−51°S to investigate these processes both at the subduction zone scale, and within the exceptionally well-exposed Torres del Paine sheeted intrusion. We present Hf isotope data for zircon from 30 to 12 Ma magmatic units that were emplaced ∼50 km inboard of the main subduction-related batholith. These samples record an ∼18 m.y. period during which the region experienced profound geodynamic changes, resulting in transient migration of arc magmatism into the retro-arc, which then vanished to be replaced by more alkaline retro-arc magmatism. Integrating published whole rock geochemistry, we show that the Hf isotope signatures of these magmatic units directly record their mantle sources, with negligible assimilation of continental crust into magmas during transport through and storage in the crust. This allows us to trace the appearance and disappearance of the subduction component in the retro-arc mantle. Our data show that migration of calc-alkaline magmatism into the retro-arc produced magmas with a more enriched Hf isotope composition that was remarkably consistent over ∼200 km and >4 m.y. (eHf(i) of −1 to +2.5). These signatures record addition of subducted continental crust to the mantle wedge during a period of subduction erosion that was associated with arc migration, and show that only a few m.y. of fluxing by a subduction component is sufficient to leave a distinct Hf isotope imprint on a mantle wedge previously unmodified by subduction. The Torres del Paine laccolith was built up by discrete pulses of magmatism in <200 k.y. Our data show isotopic differences between magmatic batches, and an abrupt shift to more juvenile Hf isotope compositions during the buildup of the youngest part of this magmatic complex, recording the rapid input of new mantle-derived melts during its formation. This rapid rejuvenation occurred within 20 ± 10 k.y., demonstrating that different batches of magmatism within a single intrusive complex can tap geochemically distinct mantle reservoirs on very short timescales of <200 k.y.


Data Repository
Table DR1.Hf isotope compositions of individual zircons measured by solution MC-ICPMS Table DR2.Hf isotope compositions of individual zircons from samples measured by laser ablation MC-ICPMS Table DR3.Hf isotope compositions of individual analyses of natural zircons used as primary standards for laser ablation MC-ICPMS Table DR4: Hf isotope compositions of individual analyses of synthetic zircons obtained from Fisher et al. (2011) and used as secondary standards for laser ablation MC-ICPMS

DR1 Additional sample descriptions
Torres del Paine Intrusive Complex Samples JL438 and JL439 come from the mafic part of the feeder zone.Both are from late stage pegmatitic segregations within gabbronorites, which were the only lithology within the mafic units of the feeder zone that crystallised zircon (Leuthold et al., 2012).Samples JL383, JL385, JL160 and JL165 are hornblende gabbros and (monzo)-diorites from the mafic sill complex collected on Cerro Castillo.The samples come from four distinct magmatic units distinguishable on the basis of field relations, chemistry and mineralogy (Leuthold et al., 2012(Leuthold et al., , 2013)).Samples JM-05-77 and JM-05-80 come from two outcrops ∼7 km apart of the uppermost and oldest TPIC granite, Granite I (Fig. 2, main text).Sample JM-04-66 is a sample of the lowermost Granite III, collected from the sill complex.
∼12 Ma dikes and monzonite from the area surrounding the TPIC Sample 11-TPM-94 is from a small exposure of a monzonitic intrusion in Valle Bader that bears some similarities to parts of the TPIC.It was dated at 12.4 ± 0.2 Ma (Müntener et al., 2018), within error of both the mafic sill complex and the feeder zone of the TPIC.Its whole rock chemistry is similar to that of the TPIC (Müntener et al., 2018).The monzonite is largely hidden under glacial moraine, obscuring any field relations with other units including the TPIC.Samples 11-TPM-32A and 11-TPM-85 are from dikes that cut metasediments of the Cerro Toro formation (Fig. 2, main text).Both have transitional alkaline whole rock geochemistry similar to that of the TPIC (Müntener et al., 2018).11-TPM-32A is a sample from the felsic part of a bimodal dike cutting metasediments of the Cerro Toro formation on the western flank of Monte Almirante Nieto.This sample gave a zircon U-Pb age of 12.8 ± 0.2 Ma (Müntener et al., 2018).A number of such bimodal dikes, which have microgranitic centres and basaltic andesite borders, have been documented (Müntener et al., 2018).At least ten such dikes have been observed crosscutting granites of the TPIC in various localities throughout the area, implying that at least some bimodal dikes postdate these granites, and providing indirect evidence that the analysed sample probably postdates Granite III.However, there is no direct evidence for how the dike sampled by 11-TPM-32A relates to any part of the TPIC.11-TPM-32A has transitional alkaline whole rock geochemistry similar to that of the TPIC (Müntener et al., 2018).The granitic part of these dikes has geochemistry similar to the oldest TPIC granite (Granite I).11-TPM-85 is a microgranitic dike that in some parts is bimodal (as for 11-TPM-32A).It gave a zircon U-Pb age of 12.3 ± 0.2 Ma (Müntener et al., 2018).It cuts metasediments of the Cerro Toro formation several kilometres southeast of 11-TPM-32A.

∼16 Ma calc-alkaline intrusions in the TPIC region
Samples 13-OLV-1, 13-TP2-1, PP-36, 07-JL-204, 09-PR-06, and 11-TPM-61A each represent a different intrusive body ≤10 km distant from the TPIC (Fig. 2, main text), ranging from gabbro to granite in composition.13-OLV-1 comes from the Olvidado Complex, a major complex made up of multiple intrusive units, while all other samples come from small isolated intrusions.The analysed samples gave tightly clustered U-Pb ages ranging from 15.7 ± 0.4 Ma to 16.5 ± 0.2 Ma (Müntener et al., 2018) (Table 1, main text).All of the ∼16 Ma intrusions have typical subduction-related, calc-alkaline chemistry indistinguishable from that of the Southern Patagonian Batholith (Müntener et al., 2018).13-OLV-1 is a monzogranitic sample from the voluminous Olvidado intrusive complex, which also includes monzogabbroic and pyroxenite units.13-TP2-1 is a granodioritic sample from a small exposure of a felsic intrusive body cropping out at Paso John Gardner, which includes granodiorite, quartz syenite and granite units.Sample PP-36 is from an intrusive body exposed on the lower part of the SSE flank of Punta Bariloche.This intrusion is a heterogeneous body dominated by diorites but with intermingled granodiorites and pegmatitic granites in parts, and a granitic upper unit.PP-36 comes from a granitic pegmatite intermingled with diorite.Sample 07-JL-204 is from a separate intrusive body exposed higher up on the SSE flank of Punta Bariloche.This intrusive body is made up of gabbros that evolve upwards into diorites.The two distinct intrusive bodies that PP-36 and 07-JL-204 come from are separated by a layer of metasediments of the Punta Barossa formation.09-PR-06 is a sample of granite from an intrusion exposed at the foot of the west ridge of Cumbre Central.11-TPM-61A samples a dioritic intrusion that crops out in Valle Ascencio, on the lowermost part of the east ridge of Monte Almirante Nieto.
∼30 Ma alkaline gabbro 11-TPM-73A was sampled from the Amarga alkaline gabbro, a small body that crops out ∼15 km east of the TPIC (Fig. 2, main text).Its age of ∼30 Ma (Müntener et al., 2018) is significantly older than other intrusions in the area.It has distinct, alkaline chemistry, with trace element patterns similar to those of typical ocean island basalts (Müntener et al., 2018).

Chaltén Plutonic Complex
Based on detailed field relations, mineralogy and geochemistry, Ramírez de Arellano et al. ( 2012) identified eight major intrusive units within the ultramafic to granitic rocks of the Chaltén Plutonic Complex.In this study, five samples from four of these units were analysed.The studied units are among the youngest of the complex, noting that all dated units (seven of the eight) range in age only from 16.90 ± 0.05 Ma to 16.37 ± 0.02 Ma (Ramírez de Arellano et al., 2012).Units that were studied in this work are here described from oldest to youngest.Descriptions of the other units can be found in Ramírez de Arellano et al. (2012).FR-08-65 samples a gabbronorite from the Laguna de los Tres Mafic Group.This unit, which comprises gabbronorites, two-pyroxene-bearing diorites and quartz diorites, has the most deformed and altered rocks of the Chaltén Plutonic Complex (Ramírez de Arellano et al., 2012).FR-08-51 samples a gabbronorite from the Laguna Sucia Mafic Group, which comprises two mafic series of norites, gabbronorites and hornblende gabbros, with brecciated igneous contacts recording substantial cooling between successive magmatic pulses (Ramírez de Arellano et al., 2012).FR-07-03 samples the Laguna Sucia tonalite, which has brecciated contacts with the Laguna Sucia mafic rocks (Ramírez de Arellano et al., 2012).FR-07-31 and FR-08-141 sample granodiorite and banded granite respectively, both from the Fity Roy Group.The Fitz Roy group is dominated by granodiorite, which crosscuts all other mafic and tonalitic units, with subordinate banded granite (Ramírez de Arellano et al., 2012).The Fitz Roy granodiorite and banded granite are essentially undeformed (Ramírez de Arellano et al., 2012).

Country rock of the Chaltén plutonic complex
The Chaltén plutonic complex was emplaced into strongly deformed Paleozoic-Mesosoic sedimentary and volcanogenic units.The oldest exposed stratigraphic level is the Paleozoic Bahia de Lancha Formation, which comprises deformed semi-pelitic to psammitic rocks (Ramírez de Arellano et al., 2012) and forms part of the Eastern Andes Metamorphic Complex (Augustsson et al., 2006).The overlying El Quemado Complex consists of Jurassic rhyodacitic volcanoclastic rocks, with local conglomeratic units (Ramírez de Arellano et al., 2012).Late Jurassic-Early Cretaceous sandstones of the Springhill Formation locally overlie the El Quemado Formation (Ramírez de Arellano et al., 2012).Stratigraphically above is the Río Mayer Formation, which consists of Cretaceous marine black shales (Ramírez de Arellano et al., 2012).

DR2 Methodology: Additional details Laser ablation MC-ICPMS instrumental
Nine masses were measured in static mode: 171 Yb, 173 Yb, ( 174 Yb+ 174 Hf), 175 Lu, ( 176 Hf+ 176 Yb+ 176 Lu), 177 Hf, 178 Hf, 179 Hf, and 181 Ta. 181 Ta is not used in the data reduction process but changing Hf/Ta was sometimes helpful in identifying contamination by inclusions.Detection was by nine faraday cups, with amplifiers calibrated for gain and baseline at the start of each session.For each analysis a total of 120 cycles of data were collected over 125 s, providing a time-resolved signal of which any part could be selected for integration if inclusions or zones of different 176 Hf/ 177 Hf were encountered, or the zircon drilled through before the end of the analysis.No relationship between the number of cycles integrated and calculated 176 Hf/ 177 Hf was observed for any sample, demonstrating that selecting different analysis lengths does not introduce any bias.

DR3 Source of the crustal component added to the mantle wedge:
subducted sediments versus eroded forearc crust Plausible source lithologies for the crustal component In Southern Patagonia, the crust west of the Andes is made up almost entirely of the South Patagonian batholith (Hervé et al., 2007) and Paleozoic metamorphic complexes (Hervé et al., 2003).To appreciably change the Hf isotope composition of magmatism, the crustal component added to the mantle wedge must have had significantly lower 176 Hf/ 177 Hf than the ambient sub-continental mantle.
Because of the very long half-life of 176 Lu (∼54 Ga Söderlund et al., 2004), extremely long periods of time are required for crust to acquire significantly less radiogenic ε Hf .This and the very low 176 Lu/ 177 Hf of zircon mean that plutonic rocks of the Patagonian batholith would have essentially retained their initial Hf isotope composition since they formed ≤160 Ma.No published Hf isotope data exist for the South Patagonian batholith, but whole rock Nd isotope data show that from ∼100 Ma to the present magmatism had relatively juvenile compositions (ε Nd(i) of +2 to +6 Hervé et al., 2007), unsuitable to represent the crustal component contaminating the mantle wedge.The oldest parts of the batholith have somewhat more crustal whole rock Nd compositions (ε Nd(i) of -6 at ∼140 Ma), but evolved to become steadily less crustal over time (ε Nd(i) of +2 by ∼100 Ma Hervé et al., 2007).Thus even the most unradiogenic (crustal) parts of the batholith do not have dramatically crustal isotope signatures, so that very large volumes of such crust would have to be incorporated into the mantle to significantly change its Hf isotope composition.This would be even further diluted by the large parts of the batholith that have more juvenile isotopic compositions.The batholith is therefore unlikely to be the source of the crustal component in the mantle wedge.A more likely source for the crustal component is late Paleozoic to Mesozoic metamorphic complexes from Southern Patagonia (Hervé et al., 2003).Metasediments from the Eastern Andes metamorphic complex (EAMC) contain a significant proportion of zircons with ages >500 Ma and up to 3500 Ma, often with a dominance of Grenvillian (1000-1200 Ma) ages (Augustsson et al., 2006;Hervé et al., 2003).
The presence of even a small number of such ancient detrital zircons can dramatically lower the bulk ε Hf(i) (Ewing et al., 2014).Detrital zircons from two EAMC metasediments analysed for Hf isotopes gave extremely unradiogenic present-day ε Hf values ranging from -10 to -80 (Augustsson et al., 2006).
A simple arithmetic mean of all of the zircons from these metasediments gives a present day ε Hf of -27.Such metasediments, with their dramatically more crustal Hf signature than the batholith, are the most plausible source of the crustal component introduced to the mantle from 17-12.5 Ma.
South of ∼47 o S, surface exposure seaward of the main divide of the Andes is overwhelmingly dominated by Jurassic-Neogene arc magmatic products of the South Patagonian batholith (Hervé et al., 2007).Late Paleozoic to Mesozoic metamorphic complexes, which are extensively exposed north of 47 o S (Hervé et al., 2003), are inferred to form the middle and lower crust beneath the Torres del Paine region (Fosdick et al., 2011, see their Fig. 3) but crop out only sporadically at the surface at the latitudes of Torres del Paine and El Chaltén.Here the Eastern Andes metamorphic complex, which as discussed above contains ancient zircons with very low ε Hf (t=0Ma) , forms only a small sliver between the batholith and the inboard intrusions (Fosdick et al., 2011;Hervé et al., 2003).Another Paleozoic unit, the Duque de York metamorphic complex, crops out in small exposures along the coast at these latitudes, but in contrast to other Paleozoic metasedimentary units, contains only a small proportion of zircons older than ∼300 Ma, and ages >500 Ma are rare (Hervé et al., 2003).These metasediments would, like the batholith, have a Hf isotope signature that is not strongly crustal, and are unlikely to represent the crustal component in the investigated samples.Our data do not allow us to unequivocally distinguish between subducted sediments and eroded forearc crust as the source of the continental component in our 17-12.5Ma magmatic rocks.However, the limited surface exposure of lithologies with highly crustal Hf isotope signatures at 49-51 o S favours subduction erosion of forearc crust as most likely.The small contribution to trench sediments expected from the limited exposure of the Eastern Andes metamorphic complex would be strongly diluted by the large contribution from the mainly juvenile batholith, which dominates surface exposure.In contrast, subduction erosion of continental crust in a forearc position would be expected to entrain mainly Paleozoic metamorphic basement rocks into the mantle.Removal of a significant quantity of forearc crust by subduction erosion south of 47 o S is also suggested by the near-complete lack of metamorphic complexes outboard of the batholith to the south of the present-day triple junction (47 o S), in contrast to the extensive exposures of these north of the triple junction (Hervé et al., 2003, their Fig. 1; cf also Kay et al. (2005)).The present-day complete lack of forearc crust outboard of the batholith south of ∼47 o S is unusual and implies removal of a significant volume of crust at some time in the past.There is independent evidence for a period of subduction erosion of forearc crust associated with continent-wards migration of arc magmatism and uplift at 20-16 Ma at 49 o S (Ramírez de Arellano et al., 2012), and between ∼30 Ma and 12-8 Ma at ∼48-51 o S (Thomson et al., 2001).Thomson et al. (2001) estimated that up to 180 km of forearc crust was removed at 48-51 o S during this period of subduction erosion.Weighted mean 176 Hf/ 177 Hf for each session is plotted as a horizontal line and given as text (95% conf.level); this relative to the nominal value of 0.282160 (Nowell et al., 1998) was used to normalise analyses of unknowns in the same session.The 2 S.D. (also given) was propagated in quadrature onto the errors of unknowns.Data for JMC-475 analyses are available from the first author on request.DR3) are not plotted.Black horizontal lines show the average offset of all primary standards in a given session, which is used to normalise unknowns in the same session.Note that primary standards were not normalised, so it is the agreement of analyses with this average (and not an offset of zero) that is important.Grey panels indicate the 2 s.d. on the average offset. 173Yb/ 177 Hf is approximate as it was calculated using only β Hf , i.e. it is not corrected for inter-element fractionation.

To r r e s d e l P a i n e i n t r u s i v e c o m p l e x C h a l t é n p l u t o n i c c o m p l e x
ε Hf(i) = 1.4 ± 0.5  (Leuthold et al., 2012;Michel et al., 2008;Müntener et al., 2018;Ramírez de Arellano et al., 2012)  For the TPIC, whole rock data were taken from Leuthold et al. (2013), for the same sample analysed for Hf where available; when this was not the case, either an equivalent sample collected from the same locality was used (two mafic sill complex samples), or the average of all analyses of the same lithology (two root zone samples).Whole rock data for the remaining samples are from Müntener et al. (2018) and are for the same samples as analysed for Hf isotopes.Note the lack of any systematic trend towards lower ε Hf(i) with increasing SiO 2 , as would be expected if crustal assimilation was important in controlling the Hf isotope composition.et al., 2013) and calc-alkaline and alkaline intrusions from the Torres del Paine region (Müntener et al., 2018).

South
Gray field shows the range for the South Patagonian Batholith (Hervé et al., 2007).

Figure DR1 :
Figure DR1: 176 Hf/ 177 Hf of individual analyses of 40 ppb JMC-475 standard solution during the three solution MC-ICPMS sessions in which Patagonian samples were analysed.Excluded analyses (n=4) are shown as a dashed line and also had anomalous 178 Hf/ 177 Hf, indicating a problem with the mass bias correction.Weighted mean 176 Hf/ 177 Hf for each session is plotted as a horizontal line and given as text (95% conf.level); this relative to the nominal value of 0.282160(Nowell et al., 1998) was used to normalise analyses of unknowns in the same session.The 2 S.D. (also given) was propagated in quadrature onto the errors of unknowns.Data for JMC-475 analyses are available from the first author on request.

Figure DR3 :
FigureDR3: Offset of measured 176 Hf/ 177 Hf from the nominal value (in ε Hf -units) versus 173 Yb/ 177 Hf for three synthetic zircons doped with REE, obtained fromFisher et al. (2011) and used as secondary standards in six LA-MC-ICPMS sessions (TP1-6). 176Hf/ 177 Hf was normalised to the primary zircon standards in the same session (as for unknowns).Black horizontal lines show zero offset, i.e. perfect agreement with the nominal value.Grey panels indicate the average offset of all secondary standards in the given session, shown at the 2 s.d.level. 173Yb/ 177 Hf is approximate as it was calculated using only β Hf , i.e. it is not corrected for inter-element fractionation.

Figure DR4 :
FigureDR4: Plots of 176 Hf/ 177 Hf versus 173 Yb/ 177 Hf for solution MC-ICPMS measurements of individual zircons from (A) six samples from the Torres del Paine intrusive complex, and (B) five samples from the Chaltén plutonic complex.No relationship between the measured Hf isotope composition and Yb/Hf is observed for accepted analyses (solid symbols), confirming the accurate correction for isobaric interferences from Yb. Conversely, in three analyses excluded for high Yb contents (hollow symbols marked by *), a systematic offset of 176 Hf/ 177 Hf relative to low-Yb analyses from the same sample is observed with increasing 173 Yb/ 177 Hf, indicating that the Yb isobaric interference is not accurately corrected at such high Yb/Hf. 173Yb/ 177 Hf is somewhat approximate ("approx") as it was calculated using only β Hf , i.e. it is not corrected for inter-element fractionation.Errors are omitted from 173 Yb/ 177 Hf for clarity.

Figure DR8 :Figure DR9 :Figure DR10 :
Figure DR8: 176 Hf/ 177 Hf of individual zircons measured by solution MC-ICPMS (2σ errors).Excluded analyses (see TableDR1) are shown in gray.Red horizontal lines plot the weighted mean for each sample; the weighted mean ε Hf(i) is given as text with 95% conf.errors.
FigureDR11: ε Hf(i) of zircon plotted against whole rock SiO 2 for the same units, taken from geochemical data available in the literature.SiO 2 is plotted as anhydrous values.For the Chaltén plutonic complex, whole rock data are for the same samples analysed for Hf isotopes, taken from Ramírez deArellano et al. (2012).For the TPIC, whole rock data were taken fromLeuthold et al. (2013), for the same sample analysed for Hf where available; when this was not the case, either an equivalent sample collected from the same locality was used (two mafic sill complex samples), or the average of all analyses of the same lithology (two root zone samples).Whole rock data for the remaining samples are fromMüntener et al. (2018) and are for the same samples as analysed for Hf isotopes.Note the lack of any systematic trend towards lower ε Hf(i) with increasing SiO 2 , as would be expected if crustal assimilation was important in controlling the Hf isotope composition.

Figure DR12 :
FigureDR12: Th/Nb versus Zr, from published whole rock geochemical data for the mafic TPIC(Leuthold et al., 2013) and calc-alkaline and alkaline intrusions from the Torres del Paine region(Müntener et al., 2018).Gray field shows the range for the South Patagonian Batholith(Hervé et al., 2007).
Fisher et al. (2011)he nominal value (in ε Hf -units) versus 173 Yb/ 177 Hf for three synthetic zircons doped with REE, obtained fromFisher et al. (2011)and used as secondary standards in six LA-MC-ICPMS sessions (TP1-6). 176Hf/ 177 Hf was normalised to the primary zircon standards in the same session (as for unknowns).Black horizontal lines show zero offset, i.e. perfect agreement with the nominal value. Grey paels indicate the average offset of all secondary standards in the given session, shown at the 2 s.d.level.173 Yb/ 177 Hf is approximate as it was calculated using only β Hf , i.e. it is not corrected for inter-element fractionation.
Plots of 176 Hf/ 177 Hf versus 178 Hf/ 177 Hf for solution MC-ICPMS measurements of individual zircons from (A) six samples from the Torres del Paine intrusive complex, and (B) five samples from the Chaltén plutonic complex. 178Hf/ 177 Hf is an invariant ratio which has no isobaric interferences, and therefore tests the effectiveness of the mass bias correction.No systematic relationship between 178 Hf/ 177 Hf and 176 Hf/ 177 Hf is observed between or with samples, indicating that mass bias has been corrected for appropriately in the ratio of interest ( 176 Hf/ 177 Hf).Plots of 176 Hf/ 177 Hf versus 173 Yb/ 177 Hf for samples from the Torres del Paine region analysed by laser ablation MC-ICPMS, demonstrating that there is no relationship between the measured Hf isotope composition and Yb/Hf.This confirms the accurate correction for isobaric interferences from Yb. 173 Yb/ 177 Hf is somewhat approximate ("approx") as it was calculated using only β Hf , i.e. it is not corrected for inter-element fractionation.Error bars are omitted from 173 Yb/ 177 Hf for clarity.Plots of 176 Hf/ 177 Hf versus 178 Hf/ 177 Hf for samples from the Torres del Paine region analysed by laser ablation MC-ICPMS. 178Hf/ 177 Hf is an invariant ratio which has no isobaric interferences, and therefore tests the effectiveness of the mass bias correction. 178Hf/ 177 Hf is always within error between analyses, and no relationship is observed between 176 Hf/ 177 Hf and 178 Hf/ 177 Hf, demonstrating that mass bias has been corrected for appropriately.
is observed for accepted analyses (solid symbols), confirming the accurate correction for isobaric interferences from Yb. Conversely, in three analyses excluded for high Yb contents (hollow symbols marked by *), a systematic offset of 176 Hf/ 177 Hf relative to low-Yb analyses from the same sample is observed with increasing 173 Yb/ 177 Hf, indicating that the Yb isobaric interference is not accurately corrected at such high Yb/Hf. 173Yb/ 177 Hf is somewhat approximate ("approx") as it was calculated using only β Hf , i.e. it is not corrected for inter-element fractionation.Errors are omitted from 173 Yb/ 177 Hf for clarity.