Structure and evolution of the volcanic rift zone at Ponta de São Lourenço, eastern Madeira

Abstract

Ponta de São Lourenço is the deeply eroded eastern end of Madeira’s east–west trending rift zone, located near the geometric intersection of the Madeira rift axis with that of the Desertas Islands to the southeast. It dominantly consists of basaltic pyroclastic deposits from Strombolian and phreatomagmatic eruptions, lava flows, and a dike swarm. Main differences compared to highly productive rift zones such as in Hawai’i are a lower dike intensity (50–60 dikes/km) and the lack of a shallow magma reservoir or summit caldera. ⁴⁰Ar/³⁹Ar age determinations show that volcanic activity at Ponta de São Lourenço lasted from >5.2 to 4 Ma (early Madeira rift phase) and from 2.4 to 0.9 Ma (late Madeira rift phase), with a hiatus dividing the stratigraphy into lower and upper units. Toward the east, the distribution of eruptive centers becomes diffuse, and the rift axis bends to parallel the Desertas ridge. The bending may have resulted from mutual gravitational influence of the Madeira and Desertas volcanic edifices. We propose that Ponta de São Lourenço represents a type example for the interior of a fading rift arm on oceanic volcanoes, with modern analogues being the terminations of the rift zones at La Palma and El Hierro (Canary Islands). There is no evidence for Ponta de São Lourenço representing a former central volcano that interconnected and fed the Madeira and Desertas rifts. Our results suggest a subdivision of volcanic rift zones into (1) a highly productive endmember characterized by a central volcano with a shallow magma chamber feeding one or more rift arms, and (2) a less productive endmember characterized by rifts fed from deep-seated magma reservoirs rather than from a central volcano, as is the case for Ponta de São Lourenço.

Appendix 1: 40Ar/39Ar dating methodology

After removing altered crusts, selected pieces of samples were crushed and sieved, and groundmass grains were separated by hand-picking from the 0.25–0.5 mm fraction to avoid alteration phases. The separates were washed in distilled water in an ultrasonic bath and subsequently dried at <50°C. Cd-shielded sample irradiations were carried out using the FRG2 reactor at the GKSS Research Center (Geesthacht) for 7 days. Irradiated groundmass samples were loaded into aluminium trays with multiple pans holding typically between 5 and 10 mg of material, and baked and pumped within a sample chamber fitted with a quartz glass window.

Analyses were carried out using an Ar-ion laser (principal wavelengths 488 and 514 nm) combined with a MAP-216 mass spectrometer at IFM-GEOMAR. The mass spectrometer is equipped with a Baur-Signer-type ion source, a Johnston electron multiplier, and a gas extraction system connected to an ultra-clean gas cleanup line (∼600 cc; Zr–Al getters; liquid nitrogen cold trap). By incrementally increasing laser power output from 0.15 to 25 W, samples were heated from about 300 to ≫1,800°C until complete fusion and partial evaporation of residual silicate melt spheres occurred. Scanning across samples in pre-set patterns with a defocused laser beam evenly heated the material. Automated step-heating was followed by final fusion at ∼25 W with a focused beam to ensure complete degassing (m/z 40 and m/z 39 signals at blank level). Figure 5 does not show temperatures on the heating steps because the temperatures were not measured precisely during the laser step-heating analyses.

Ion beam currents were measured with an electron multiplier at m/z = 36 to 40 and half-mass baselines with an 7.5 digit integrating HP multimeter. Peak heights were regressed to inlet time, peak decay typically being less than 10% during the analyses. Average extraction line blanks are determined as 2 × 10¹⁷ mol at m/z = 36 and 4 × 10¹⁶ mol at m/z = 40. Mass discrimination was monitored using air-fused zero age basaltic glass and pipette air samples (1.0083 ± 0.0006 amu). Correction factors for interfering neutron reactions on Ca and K were determined from co-irradiated CaF₂ and K₂SO₄ crystals ($^{{{{\text{36}}} \mathord{\left/ {\vphantom {{{\text{36}}} {39}}} \right. \kern-\nulldelimiterspace} {39}}} {\text{Ca = 0}}{\text{.445}} \pm {\text{0}}{\text{.005}}$, $^{{{{\text{37}}} \mathord{\left/ {\vphantom {{{\text{37}}} {39}}} \right. \kern-\nulldelimiterspace} {39}}} {\text{Ca = 1006}} \pm {\text{7}}$, ^38/39K = 0.0168, $^{{{{\text{40}}} \mathord{\left/ {\vphantom {{{\text{40}}} {39}}} \right. \kern-\nulldelimiterspace} {39}}} {\text{K = 0}}{\text{.004}} \pm {\text{0}}{\text{.002}}$).

⁴⁰Ar/³⁹Ar ages were measured relative to the flux monitor standard TCR sanidine (27.92 Ma; Duffield and Dalrymple 1990) and primary standard SB-3 biotite (162.9 Ma; Lanphere and Dalrymple 2000), uncertainties for the J-values being estimated as ±0.08% (1σ). For incremental heating plateau ages all consecutive temperature steps that overlapped within 2σ were included. Incremental heating plateau ages and single-particle mean apparent ages were calculated as weighted mean (Taylor 1982) and initial ⁴⁰Ar/³⁶Ar isotope ratios and isochron ages as least squares fit with correlated errors (York 1969) applying decay constants of Steiger and Jäger (1977). Plateau steps are in equilibrium with air by definition (assuming an initial ⁴⁰Ar/³⁶Ar of 295.5); note that the ⁴⁰Ar/³⁶Ar intercepts given in Table 1 are not all within error of air because they are based on all heating steps.