JOURNAL OF PETROLOGY

Journal of Petrology, 2015, Vol. 56, No. 8, 1547–1584 doi: 10.1093/petrology/egv045 Advance Access Publication Date: 11 September 2015 Original Article

Shield to Rejuvenated Stage Volcanism on Kauai and Niihau, Hawaiian Islands Brian L. Cousens1* and David A. Clague2 1

Ottawa–Carleton Geoscience Centre, Department of Earth Sciences, Carleton University, 1125 Colonel By Drive, Ottawa, ON K1S 5B6, Canada and 2Monterey Bay Aquarium Research Institute, 7700 Sandholdt Road, Moss Landing, CA 95039, USA *Corresponding author. Telephone: 613-520-2660 ext. 4436. Fax: 613-520-5613. E-mail: [email protected] Received July 16, 2014; Accepted July 23, 2015

ABSTRACT We report and interpret new geochemical and Pb–Sr–Nd isotopic data from 325 samples of shield, late-shield, postshield, and rejuvenated stage lavas from Kauai and Niihau, the two most northwesterly islands in the Hawaiian island chain. Kauai is unique in the Hawaiian chain in that it exhibits a near-continuous geochemical transition from shield to postshield to rejuvenated stage volcanism between 44 and 36 Ma and has been continuously active over 6 Myr. From c. 57 to 43 Ma, the shield stage of both islands produced tholeiitic basalts typical of other Hawaiian shield volcanoes. The Niihau basalts are more evolved and have high Gd/Yb compared with Kauai, indicating a higher residual garnet content in the source. Both Kauai and Niihau shield basalts have Kea-like trace element ratios, but isotopic ratios are transitional between Kea- and Loa-like compositions. The geochemical similarity of the two shields indicates that mantle sources in different regions of the plume source were similar, and that the 1 (SD Electronic Appendices 11 and 12). Samples 75-NII-1, 2, and 4, from the Ka’eo Hill plug in the central part of the island, are rich in MgO (13%) and Ni (690–1436 ppm) compared with the two dykes

Journal of Petrology, 2015, Vol. 56, No. 8

from the NW coast of the island (833 and 614% MgO, 214–136 ppm Ni). Ka’eo samples also include abundant xenocrystic olivine, and the abnormally high Ni abundances suggest that the olivine xenocrysts may include excess Ni compared with olivine phenocrysts (Fig. 5a). In addition to their more alkaline character (higher K2O and TiO2), CaO/Al2O3 and Sc are somewhat lower in late-shield rocks (especially Ka’eo Hill samples) compared with Niihau and Kauai shield basalts, consistent with minor clinopyroxene fractionation (Figs 4a and 5a). Both onland and submarine late-shield lavas are very similar compositionally. Incompatible element patterns (SD Electronic Appendix 13, Panel C) and ratios (Fig. 6) for the submarine lavas are almost indistinguishable from onland late-shield intrusions. Samples T321-R6 and R7 are slightly enriched in the most incompatible elements and depleted in the less incompatible elements, such that their patterns cross those of the other late-shield lavas (SD Electronic Appendix 13, Panel C). The late-shield lavas are isotopically indistinguishable in Pb, Sr and Nd from the Niihau shield basalts (Fig. 7a and b). The exception is sample T321-R6, which has the lowest Pb isotope ratios (206Pb/204 Pb ¼ 1805) of all shield and late-shield rocks from Niihau. Late-shield rocks were included within the Paniau Basalt by Langenheim & Clague (1987) based on the indistinguishable ages of the tholeiitic and transitional basalts, and the similarities in chemistry between the two lava groups at Niihau are consistent with this interpretation.

Rejuvenated stage Kauai Rejuvenated stage lavas of the Koloa Volcanics on Kauai include alkalic basalts, basanites, nephelinites and melilitites (Fig. 3b). Included in Fig. 3a are three basanite samples that Garcia et al. (2010) assigned to the postshield stage on Kauai even though the rocks have Koloa-like chemistry. Our Koloa samples rarely have K2O/P2O5 < 1 (SD Electronic Appendix 12), typical of Koloa samples analyzed in previous studies. Fe2O3t decreases (not shown) and Al2O3 increases with decreasing MgO, but other major elements show little to no correlation with MgO (Fig. 4b). Although CaO/ Al2O3 broadly decreases with decreasing MgO, Sc abundances show a scattered increase with decreasing MgO (Fig. 5b), ruling out clinopyroxene fractionation. Ni and Cr abundances decrease with decreasing MgO, whereas the abundances of most incompatible elements such as Sr, Nb and the LREE (e.g. Ce) show a diffuse, highly scattered decrease with decreasing MgO (Fig. 5b). Normalized incompatible element patterns for our new analyses of Koloa lavas are shown in SD Electronic Appendix 13, Panel D. Although the shapes of the patterns are all similar, the slopes vary and patterns commonly cross in the MREE range. All lavas show positive

1561

spikes at Ba, Nd, and less commonly Sr, and troughs at Th and U, K, Zr and Hf, and Ti. La/Smpmn ranges from 24 to 42, and Gd/Ybpmn from 26 to 6, demonstrating variable LREE enrichment and HREE depletion in Koloa lavas (Fig. 6). The observed depletion in U and Th compared with adjacent elements Ba and Nb is similar to that seen in postshield rocks from Kauai. Our isotopic analyses of onland Kauai Koloa lavas mirror those of previous studies (Fig. 7a and b). The 87 Sr/86Sr are lower, and the 143Nd/144Nd are higher those of the Kauai shield lavas. Although their 206Pb/204 Pb ratios (1805–1835) overlap the range shown by Kauai shield lavas, the rejuvenated lavas exhibit a greater range in 208Pb/204Pb and form a data array with a higher slope than the shield data (Fig. 7b). All three analyzed and dated Koloa lavas from the Hanamaulu Town (HTZ) groundwater drill hole (Sherrod et al., 2015) plot within the range of rejuvenated lavas on Kauai. The ROV Tiburon Dive T316 recovered both alkalic basalts and basanites, whereas Dive T324 collected only basanites (SD Electronic Appendix 9). The submarine samples have high MgO that fall within the range of onland lavas, other than two samples (one glass, one whole-rock) with MgO 75 wt % is sample 70-NII-15; this contains more than 15% modal olivine, many crystals of which show similar optical properties to olivine crystals in Kauai picritic basalts and hence are xenocrystic. If samples with > 20% olivine crystals are discounted from the onland Kauai and Niihau datasets, then the range of MgO contents for the Kauai shield lavas is 67–183% and for Niihau is 55–75%. The Niihau shield lavas, unlike those from Kauai, commonly include plagioclase and pyroxene phenocrysts, and the more evolved lavas have lower CaO/Al2O3 and Sc concentrations that indicate clinopyroxene fractionation prior to eruption. Clearly, the mineralogy and major element chemistry show that Niihau shield lavas are more evolved than those from Kauai.

Loa versus Kea affinity: implications for plume zonation, and the longevity of Hawaiian trends The 60%). Xu et al. (2014) also demonstrated that isotopic ratios

1565

correlate with calculated mantle temperatures, with the Kea-type basalts having higher mantle temperatures than Loa-type basalts. The isotope–temperature correlations are supportive of a thermal model, in which the Kea-type magmas are generated at higher temperatures closer to the center of the plume from a more peridotitic source and the Loa-type magmas are generated at lower temperatures from a source richer in recycled components (pyroxenite, sediments) (e.g. Ren et al., 2009). This model predicts that magmas generated along the periphery of the plume, including the last gasps of shield volcanism, should be more Loa-like in composition than magmas generated above the hot core of the plume. The similarity in isotopic and trace element characteristics between West Molokai–Penguin Bank and Kauai–Niihau shield basalts is remarkable (Fig. 10). Both pairs of volcanoes straddle the Loa–Kea dividing line in a 208Pb/204Pb–206Pb/204Pb plot, and fall within the Kea and Loa fields in Sr–Nd isotopic space as well as in plots of diagnostic trace element ratios (e.g. Fig. 9b; Xu et al., 2014, fig. 6c). Like the West Molokai–Penguin Bank lavas, Kauai can be interpreted in terms of the thermal model proposed for the generation of the Kea and Loa magmatism. The shift from more Kea-like to more Loa-like compositions from west to east in the Kaua’a shield lavas corresponded to magma generation from hot to cool sources as the volcano migrated over the plume. However, the Niihau and Kauai shield basalts have essentially the same transitional Kea to Loa characteristics and there is no lateral isotopic gradient between the two shields, unlike that observed from dominantly Kea-like at East Molokai to more Loa-like between West Molokai and Penguin Bank. Shield activity at Niihau and Kauai, and presumably beneath Kaula Island (Ka’ula, Fig. 1) SW of Niihau (Garcia et al., 1986), occurred at the onset of an increase in magmatic flux from the Hawaiian plume (Van Ark & Lin, 2004; Vidal & Bonneville, 2004). Pulsation of the flux over time may be tied to solitary (or conduit) waves travelling in the plume (e.g. Schubert et al., 1989), and in this model the increase in flux at 5 Ma marks the arrival of a solitary wave. Solitary waves allow material from the deep mantle to travel quickly up the rising mantle plume, allowing hot and slightly lower viscosity mantle to be periodically injected into the base of the lithosphere (Schubert et al., 1989). The newly arrived pulse may be responsible for the breadth of magmatic activity across the plume, producing the 200 km wide Kaula–Niihau–Kauai platform (Fig. 1). The higher temperature of the solitary wave material would favor higher degrees of melting, allowing a peridotitic source (i.e. Kea-like) to dominate magma compositions at both Kauai and Niihau. As time progressed and the effect of the solitary wave arrival diminished, lower melting temperature components (i.e. Loa-like) became more important during magma generation beneath eastern Kauai. This idea emphasizes vertical heterogeneity in the plume, as well as variable solidus temperatures,

1566

Journal of Petrology, 2015, Vol. 56, No. 8

(a)

(b)

Fig. 10. (a) 143Nd/144Nd vs 87Sr/86Sr in Kea-type (dashed field; Kilauea, Kohala, Mauna Kea, Haleakala, and West Maui) and Loa-type (dot–dashed field; Mauna Loa, Kahoolawe, Lanai, Koolau) shield basalts (data fields from GEOROC, 2012). Data for Kauai and Niihau shield lavas are from this study and previous work (Feigenson, 1984; Holcomb et al., 1997; Reiners & Nelson, 1998; Mukhopadhyay et al., 2003; Garcia et al., 2010). Kauai lavas are subdivided by member. East Molokai, West Molokai and Penguin Bank data fields are from Xu et al. (2005, 2007, 2014). LO box is the Loihi endmember. (b) 208Pb/204Pb vs 206Pb/204Pb for Kea-type (dashed field) and Loa-type (dot–dash field) shield basalts. Data sources as in (a). For clarity, a separate field for Penguin Bank is not shown and is included with West Molokai. Continuous line is the Kea–Loa dividing line from Abouchami et al. (2005). KEA, Loihi, and Enriched Makapuu Koolau (EMK) components are from Bizimis et al. (2013). Western Kauai lavas (Napali West) are Kealike in isotopic composition, but eastern lavas (Napali East, Haupu) shift to more Loa-like compositions.

which are discussed in detail by Blichert-Toft et al. (2003). Based on the interpretation of curving loci of volcanic centers along the Hawaiian plume track (Fig. 1

inset) (Jackson et al., 1972), Kauai and Niihau should have different geochemical affinities. However, we have shown that both the Kauai and Niihau tholeiites are generally Kea-like with transitional to Loa

Journal of Petrology, 2015, Vol. 56, No. 8

compositional characteristics. Thus, there is no evidence that two parallel Hawaiian ‘trends’ existed during Kauai and Niihau shield construction, as seen in the southeastern Hawaiian islands. The observations from Kauai and Niihau are inconsistent with a long-term bilateral, concentric or radial zonation model (see also Xu et al., 2007; Huang et al., 2013). The radial zonation model could be correct if the plume was extremely broad, such that both Kauai and Niihau sampled the same part of the plume.

Late-shield to postshield stage Niihau late-shield volcanism We suggest that Niihau includes both late-shield and postshield lavas. The late-shield lavas and dykes have K–Ar ages that fall within the range of shield lavas and have geochemical compositions that differ only slightly from shield basalts, similar to what is seen at Kahoolawe (Huang et al., 2013). Late-shield activity includes the onland alkaline intrusions at Ka’eo and the NE coast dykes, as well as mildly alkaline samples from Dives T318, T321 and T322. The overlap of the isotopic data from the late-shield and shield basalts indicates that they shared the same sources. Normalized REE patterns for shield and late-shield rocks form a fanning pattern; the late-shield basalts are LREE enriched compared with most shield basalts and all patterns converge towards the HREE (SD Electronic Appendix 13, panel C). The fanning of the patterns suggests that the shield and late-shield lavas are related by approximately a factor of two difference in the degree of partial melting of a similar source in which garnet is a residual phase. Late-shield lavas from Niihau follow the differentiation trends seen in late-shield tholeiitic to mildly alkalic basalts from Mauna Kea (Hamakua Volcanics) and East Molokai (Frey et al., 1990; Xu et al., 2005). The Hamakua Volcanics include alkalic basalts with steeper REE patterns than interbedded tholeiitic basalts, but the HREE abundances of the two lava types overlap. In a plot of Sr/Ce vs Nb, the Niihau late-shield lavas follow a lowpressure differentiation trend (olivine þ plagioclase) that parallels the Niihau shield basalts but at slightly higher Nb concentrations, overlapping with the Hamakua Volcanics (Fig. 11). The lack of a decrease in CaO/Al2O3 and Sc at low MgO indicates that clinopyroxene was not a major fractionating phase in late-shield lavas from Niihau. Niihau late-shield lavas also share isotopic similarities with Mauna Kea and East Molokai late-shield basalts in that they are shifted to slightly lower Sr and Pb and higher Nd isotope ratios than the shield basalts (Frey et al., 1990; Xu et al., 2005; Hanano et al., 2010). Samples T321-R6 and R7 are exceptions. Both are enriched in K2O, Ba, Th, Nb and the LREE, and have lower Zr/Nb and higher La/Sm and Nb/Y compared with other Niihau late-shield lavas. These two samples appear to be transitional in composition between other

1567

late-shield basalts and the Niihau alkaline submarine suite (Figs 5a and 6).

Niihau postshield alkaline suite The inclusion of the basanites from Dives T319–T321 in the postshield stage is based primarily on the overlap in isotopic composition between these alkaline lavas and the Niihau shield and late-shield lavas, along with the lack of overlap with the rejuvenated stage Kiekie Volcanics lavas. Postshield alkaline suite lavas appear to be confined to small cones that lie atop a ridge extending NW from Niihau that may represent a small rift zone. Postshield volcanism commonly is concentrated along rift zones (e.g. Hualalai, Haleakala), unlike the rejuvenated stage volcanic vents. In addition, samples of late-shield basalts were collected on Dives T319 and T321, but no Kiekie-like basalts were collected on these dives. The alkaline suite basanites are also very different mineralogically from the Kiekie volcanic rocks and have trace element ratios such as La/Sm, Nb/Y, and Zr/ Y that do not overlap compositionally with the Kiekie alkali basalts. Thus, we see no geological and only rare geochemical ties between the postshield alkaline suite and the rejuvenated stage of activity on Niihau. On these grounds we assign the postshield alkaline suite to the postshield stage. The postshield alkaline suite differs significantly from late-stage lavas at Mauna Kea and East Molokai (Fig. 12). SiO2 is lower and total alkalis (especially K2O) are much higher in the Niihau alkaline suite. With decreasing MgO, K2O and P2O5 (not shown) increase more rapidly than in the late-stage shield basalts. CaO/ Al2O3 decreases from 085 to 075 with decreasing MgO, but Sc concentrations vary little and show no covariation with MgO or CaO/Al2O3, ruling out clinopyroxene as an important fractionating phase in the alkaline suite magmas (Fig. 12). Compared with the late-shield basalts from Niihau, the alkaline suite rocks are enriched in the most incompatible trace elements (e.g. Nb). However, primitive mantle normalized REE patterns for the two lava suites largely overlap in the MREE–HREE (Fig. 13). The LREE enrichment, combined with subtle HREE depletion, indicates that the alkaline suite may be related to the late-shield lavas on Niihau by a lower degree of partial melting of the mantle and a higher residual garnet content in the residue. Samples T321-R6 and -R7, the late-shield basalts with an average age of 484 Ma, have REE patterns that are subtly distinct from other late-shield basalts (dashed curves, Fig. 13) and trace element ratios (Fig. 6) that are intermediate between other late-shield and the postshield lavas.

Kauai postshield volcanism Rocks ascribed to the postshield phase of activity on Kauai are varied in composition. In this study, postshield lava flows from the upper parts of the Olokele and Makaweli formations and a dyke cutting the Napali

1568

Journal of Petrology, 2015, Vol. 56, No. 8

17

Intermediate P Kauai Shield Low P

Kauai Postshield

15 G. et. Postshield Niihau Shield

13

Niihau Lateshield

Sr/Ce

Niihau Postshield

11

T321-R6,7

Laupahoehoe

th

9

7 m

Hamakua

5 0

10

20

30

40

50

60

70

80

90

100

Nb (ppm)

Fig. 11. Sr/Ce vs Nb for shield, late-shield and postshield samples from Kauai and Niihau from this study. Also shown are two postshield tholeiites (th) and a mugearite (m) from Garcia et al. (2010) (G. et.). Dashed fields are postshield Hamakua tholeiites and alkali basalts and younger Laupahoehoe hawaiites and mugearites from Mauna Kea (Frey et al., 1990). Sr/Ce decreases during plagioclase fractionation, which is dominantly a shallow-level crystallizing phase (low-P arrow). Laupahoehoe lavas first underwent moderate-pressure fractionation of clinopyroxene and olivine (intermediate-P arrow) raising Nb concentrations in the residual liquids at a constant Sr/Ce, followed by plagioclase crystallization that lowered Sr/Ce. Postshield rocks from Kauai overlap with Laupahoehoe lavas, suggesting that they underwent pyroxene or olivine fractionation at depth prior to plagioclase fractionation. Lavas of the Niihau postshield alkaline suite follow the same systematics as Kauai postshield rocks. Late-shield basalts from Niihau are displaced to slightly higher Nb concentrations but follow the low-pressure, olivine þ plagioclase-dominated fractionation history seen in the shield basalts.

formation basalts range from alkalic basalt to mugearite (Fig. 3a). This range of compositions is much like that seen in the Hamakua and Laupahoehoe Volcanics at Mauna Kea (Frey et al., 1990). The alkalic basalt dyke sample 88KA-3 has 11% MgO and may be (in general terms) a suitable parental magma for the more evolved hawaiite and mugearite. This is supported by the similarity in isotopic compositions amongst the four postshield samples (Fig. 7). The four postshield rocks generally follow the magma evolution trends for Kohala and Mauna Kea postshield basalts to hawaiites (Fig. 12). The positive correlations of CaO/Al2O3 and Sc with MgO are consistent with clinopyroxene fractionation. In Fig. 11, three of the four postshield lavas have high Nb contents and intermediate to low Sr/Ce that fall within the field of Laupahoehoe formation lavas from Mauna Kea, interpreted to indicate a role for initial deep fractionation of clinopyroxene followed by plagioclase (Frey et al., 1990). Data from two mugearites (samples GA-566 and KV04-20) from Garcia et al. (2010) follow the major

element trends of the three evolved postshield lavas from this study. Figure 12 shows that the field for postshield lavas from Haleakala volcano (Hana and Kula Volcanics) also encompasses postshield rocks from Kauai. Most of the highly incompatible elements, such as Rb, Ba, Th, Nb, and the LREE, are more enriched in Kauai postshield lavas compared with Mauna Kea, and tend to more closely follow the Haleakala trend. This is particularly true for sample 88KA-3, the least evolved postshield dyke from this study, which plots near the least-evolved end of the Haleakala field. Garcia et al. (2010) included three basanites in the postshield stage on Kauai, based on ages (392– 358 Ma) that overlap with or closely follow the end of the shield stage of activity on Kauai. Although mugearite sample GA-566 (m in Fig. 14) has the same isotopic composition as the four postshield rocks from this study, the two MgO-rich basanites and an evolved basanite (b in Fig. 14) have much lower 87Sr/86Sr and 206 Pb/204Pb and higher 143Nd/144Nd than Kauai hawaiites, and fall within the field of the Koloa Volcanics rejuvenated stage lavas. Basanite lava sample 86KA-12 (this study), with a K–Ar age of 365 6 003 Ma, was assigned to the Koloa Volcanics based on its chemistry, as were three cobbles (N4, N6, and N8) from a conglomerate unit (Feigenson, 1984) underlying the 365 Ma flow (Clague & Dalrymple, 1988). The basanites are more silica-undersaturated and have lower 87Sr/86Sr than most postshield lavas on other Hawaiian islands, and are best included in the rejuvenated stage (Koloa Volcanics) on Kauai. Other Hawaiian volcanoes with postshield volcanic rocks generally lack undersaturated rocks and include alkalic basalt, abundant hawaiite, and lesser mugearite, benmoreite, and trachyte (Mauna Kea, Kohala, Hualalai, West Maui, West Molokai, East Molokai, and Waianae [summarized by Sherrod et al. (2007)]. An exception is Haleakala, where the earliest postshield lavas are Kula alkalic basalts and hawaiites with rare mugearites and benmoreites, followed after only a 03 Myr hiatus by Hana basanites and rare phono-tephrites (West & Leeman, 1987; Chen et al., 1990; Sherrod et al., 2003). At Haleakala, there is geochemical overlap between the Kula and Hana Volcanics, including isotopic ratios. The most dramatic change in isotopic composition occurs at the Honomanu tholeiite–Kula alkali basalt boundary over less than 01 Myr (see Sherrod et al., 2003, fig. 8), after which 87Sr/86Sr ratios remain more typical of rejuvenated stage volcanic rocks at 07033) compared with lower Sr/Ce Kauai rejuvenated rocks (