April 2017 LIP of the Month

The Eastern rift branch of East Africa: an LIP-style continental rift exhibiting massive magma and volatile production

James D. Muirhead¹, Cynthia J. Ebinger², Simon A. Kattenhorn³, Hyunwoo Lee⁴, Tobias P. Fischer⁴, Steven Roecker⁵, Sara Jaye Oliva²

¹Syracuse University, New York, USA

(James Muirhead jmuirhead@ymail.com; https://jamesdmuirhead.wordpress.com/)

²Tulane University, Louisiana, USA

³University of Alaska Anchorage, Alaska, USA

⁴University of New Mexico, New Mexico, USA

⁵Rensselaer Polytechnic Institute, New York, USA

The following is based on a series of recent publications that include, but are not limited to, Lee et al. (2016), Muirhead et al. (2016), Lee et al. (2017), Roecker et al. (2017), and Weinstein et al. (in review). For full details, please refer to these papers.

1. Introduction

Continental rift zones are sites of lithospheric thinning and heating, commonly accompanied by magmatism and volatile transfer to the lithosphere and atmosphere. Some rifts evolve to seafloor spreading, with the heavily intruded and stretched passive margin increasing the surface area and volume of continental plates. Quantifying the flux of magma and volatiles during rifting is thus fundamental to our understanding of the growth of continental lithosphere and the deep carbon cycle.

Although some continental rifts lack the magma production rates of large igneous provinces (LIPs), the Eastern rift branch of the East African Rift System (EARS) fits under the category of an LIP (Ernst, 2014), and hence is consider here to be an “LIP-style continental rift”. The Kenya Rift sector alone has eruptive volumes in excess of 100,000 km³ (Guth, 2015) and has developed since ~25 Ma in an intra-plate continental setting (Ebinger et al., 2000), albeit within a nascent divergent margin. Although intruded magma volumes for the Kenya Rift are poorly constrained, they are likely significant even during earliest rift stages (first few million years), as illustrated by the multi-disciplinary CRAFTI (continental rifting in Africa: fluid-tectonic involvement) project described below.

Quantifying the fluxes of magma and volatiles during the earliest stages of rifting can help to address outstanding fundamental questions regarding how continental rifts initiate and evolve to seafloor spreading. Force balance analyses indicate that the break up of thick, strong lithosphere in East Africa requires more than tectonic forces (Bott et al., 1991), and thus continental rifting is inferred to be assisted by buoyancy forces and heat transfer from ascending magmatic intrusions (Buck, 2004; Bialas et al., 2010), and possibly from fluid-driven lithospheric weakening via metasomatism (e.g., Ebinger et al., 1997; Vauchez et al., 2005; Lindenfeld et al., 2012). However, to date no study has quantified the flux of magma and related volatiles into early-stage rifts in the Eastern rift branch, or demonstrated the importance of these rift components for strain localization processes during incipient rifting.

The CRAFTI project in the Natron-Magadi basin (Kenya-Tanzania border) has produced a complementary, multi-disciplinary dataset that allows us to probe the inner workings of continental rifts in the Eastern rift branch. Through this dataset we have constrained, for the first time, a model for the involvement magma and magmatic volatiles across a ~40 km-deep crustal section of incipient continental rifting. The results of geophysical studies constrain crustal structure and strain patterns, and mantle studies are underway (Plasman et al., 2017; Roecker et al., 2017). Below we summarize major findings from recently published geological, geophysical, and geochemical studies in the region. From these studies, we can make the following broad conclusions regarding early-stage rifting in the Eastern rift branch:

• Early-stage rifting (<7 Ma) in the Eastern rift branch is accompanied by magma accumulation at the crust mantle boundary (Plasman et al., 2017). These magmas rise to ca. 12 km depth at the Natron and Magadi basins, with significant volumes emplaced in the lower crust beneath rifted basins, as well as below uplifted horsts between basins (Roecker et al., 2017; Weinstein et al., in review).
• These magmas are generated in the sub-continental lithosphere, and their emplacement at crustal levels contributes to crustal weakening through heating and hydraulic fracturing of surrounding rocks (Lee et al., 2017). The magmatic intrusions also release massive volumes of magmatic CO₂ along border and intra-rift faults at large distances from volcanoes (Lee et al., 2016).
• The intrusion of massive volumes of magma and subsequent release of magmatic volatiles plays a critical role in strain localization and rift evolution processes in the Eastern rift branch of East Africa (Muirhead et al., 2016).

2. Quantifying the flux of magmatic CO₂ across rift-wide fault systems

To test if rift-wide faults transport magmatic volatiles to the surface, diffuse CO₂ flux was measured in the Natron and Magadi basins with an accumulation chamber in fault zones (Lee et al., 2016) (Figs. 1 and 2). Gas samples were collected in pre-evacuated glass vials and analyzed for δ¹³C-CO₂ (‰) and CO₂ concentrations to constrain the source of the observed CO₂ (i.e., biogenic or magmatic).

Carbon isotope compositions (δ¹³C-CO₂ values range from -3.8 to -11.7‰) and elevated CO₂ concentrations (mean = 929 ± 386 ppm) indicate a strong magmatic contribution to the observed CO₂ (Fig. 2). High CO₂ flux data in fault zones (mean values in Magadi of 36.6±11.0 g m^-2 d^-1) suggests that faults act as permeable pathways that facilitate the ascent of this CO₂. CO₂ flux data (Fig. 2) support annual outputs of 4.05±1.9 Mt yr^-1 in the ~180 km-long Natron-Magadi basin (Lee et al., 2016). If this flux were applied to the entire Eastern rift branch, it would equate to an annual flux of 71 ± 33 Mt yr^-1 of CO₂, which could potentially increase the global flux from natural systems by 11% to 708 Mt yr^-1 (estimates from Burton et al. (2013)). These results suggest the EARS is an important contributor to the deep carbon cycle, and reveal that significant volumes of magmatic volatiles are released through fault systems during early-stage rifting.

3. Spring systems

Thermal spring samples (T = 36.8–83.5 °C) in the Natron-Magadi basin were analyzed to investigate the sources of magmas and fluids in the rift, as well as their potential interactions with the surrounding crust (Lee et al., 2017) (Fig. 3). Helium isotope data support both mantle (10-66%) and crustal contributions to the spring fluids. ³He/⁴He ratios (R/Ra, Ra = 1.4 × 10^-6), reaching as high as 4.0 Ra, are close to those observed for subcontinental lithospheric mantle-derived volatiles (6.1 ± 0.9 Ra), instead of MORB-type (8.0 Ra), which is consistent with the model suggested by Mana et al. (2015).

Production rates of ⁴He in the Natron and Magadi basins range from 3.74 × 10¹¹ to 3.34 × 10¹⁴ atoms m⁻² s⁻¹, and are up to four orders of magnitude greater than mean production rates (4.18 × 10¹⁰ atoms m⁻² s⁻¹) from continental settings on Earth. These high ⁴He fluxes therefore cannot be attributed solely to alpha decay of U and Th. It is instead likely that the elevated production of crust-derived ⁴He is driven by heating and fracturing of country rock in response to magma intrusion at depth (Lee et al., 2017).

4. Crustal structure of an early-stage rift basin

The 39-station CRAFTI seismic array located 3427 events in the study region with magnitudes ranging from 0.4 to 4.4 (Weinstein et al., in review) (Fig. 1). Moho depths, determined from receiver functions, transition from 41 km beneath the rift flank to 27 km below the rift valley (Plasman et al., 2017). Joint inversion of ambient noise, arrival time, and gravity data was performed by Roecker et al. (2017) to image Vp, Vs, and Vp/Vs variations within the crust, with resolution to about 25 km. Receiver function models indicate high Vp/Vs at the sometimes highly reflective crust-mantle boundary, suggesting sill intrusions, or magmatic underplate. Beneath the central Natron and Manyara basins and the Crater Highlands to the west, low S-wave speeds are observed at mid to lower crustal depths, and attributed to the presence of partial melt (Roecker et al., 2017). Seismicity occurs in these areas of magmatic underplate (20-30 km), suggesting that the intrusion process is ongoing (Weinstein et al., in review). Seismicity also occurs in pancake-shaped swarms in the mid-crust and along the boundaries of the large volume low-velocity zone beneath Oldoinyo Lengai and Gelai volcanoes (Weinstein et al., in review). Unusually low Vp/Vs ratios characterize the footwall of the Natron basin border fault, which is actively degassing large volumes of mantle-derived CO₂ (Lee et al., 2016). The low Vp/Vs could be caused by CO₂-filled rocks, and may be a diagnostic feature of active degassing (Roecker et al., 2017).

5. Magmatic fluid release and strain localization

Fault systems in the Eastern rift branch are classically sub-divided into large (throws >1 km) border faults, which bound the half-graben basins, and smaller intra-rift faults, which occur along the basin floor in the rift center. As rift basins evolve to sea-floor spreading ridges, the locus of strain accommodation gradually migrates, or jumps, from border faults into intra-rift fault systems, which are underlain by zones of magmatism where dike intrusion accommodates a large proportion of the plate-boundary strain (Ebinger and Casey, 2001; Keir et al., 2006; Corti et al., 2009; Muirhead et al., 2016).

We investigated the processes controlling this transition from border-fault controlled rifting (asymmetric rift) to strain localization to the rift center (intra-rift faulting with accompanying dike intrusion at depth) through an analysis of the Natron and Magadi basins (Muirhead et al., 2016) (Fig. 4). Remote-sensing analyses (0.5 m aerial photography, and 30 m SRTM data) of intra-rift fault systems were employed to quantify the distribution of fault strain. ⁴⁰Ar-³⁹Ar dating of faulted lavas constrained the ages of the faults, and were then used to estimate time-averaged extension rates in the fault populations (Muirhead et al., 2016). Results revealed that intra-rift faulting in the Magadi basin accommodate 1.53-1.73 mm yr^-1 of extension, whereas faults in the Natron basin accommodate 0.33-0.61 mm yr^-1 of extension (Fig. 4). Accounting for extensional strains related to basin-scale flexure of border fault hanging walls (described in detail in Muirhead et al. (2016)), it is estimated that intra-rift faults in Magadi accommodate 1.34-1.60 mm yr^-1 of the total 2 mmyr^-1 of regional extension across the basin (estimates from Saria et al., 2014) (Fig. 5). In contrast, intra-rift faults in the Natron basin account for only 0.02-0.56 mm yr^-1 of the 1.8 mmyr^-1 of regional extension (Fig. 5). These strain analyses, however, cannot constrain what proportion of this extension is accommodated via dike intrusion at depth. For example, the 2007 dike intrusion below Gelai volcano at the southern end of the Natron basin is shown to have accommodated the majority of extensional strain during this discrete rifting event (Calais et al., 2008), with normal faulting accommodating extension in only the upper few kilometers of the crust. As static stress changes above the dike tips are shown to initiate faulting and graben formation in rifting environments (Rubin and Pollard, 1988), extensional strains measured at the surface in the Natron and Magadi basins are expected to mirror that accommodated at depth regardless of what proportion is accommodated through either dike intrusion or normal faulting (Muirhead et al., 2016).

In all, our fault system analyses suggest that strain in the 7 Ma Magadi basin has focused into the intra-rift fault system, which is underlain by a magma intrusion zone above the mid-crustal magma bodies imaged in seismic studies (Roecker et al., 2017; Weinstein et al., in review). In the 3 Ma Natron basin, time-averaged strain is primarily concentrated along the border fault (1.24-1.78 mm yr^-1 extension rate) (Fig. 5). Also observed in the region is a clear link between areas of high strain and hydrothermal fluid release (Figs. 3 and 4). In the Natron basin, hot springs are observed along the active rift borders, whereas the focusing of strain into a system of intra-rift faults in Magadi has occurred with an accompanying transition to fluid release in the rift center. Geochemical analyses of spring fluids and diffusely degassing CO₂ along these fault systems support significant magmatic contributions to these fluids (Figs. 2 and 3). Similar patterns are seen in active deformation patterns from discrete magmatic rifting episodes (e.g., Calais et al., 2008), and the inter-magmatic period (Weinstein et al., in review). Therefore, we concluded that the flow of magmatic fluids plays an important role in focusing upper crustal fault strain to the rift center in early-stage continental rift basins, thereby assisting in the strain localization processes that drive rift basin evolution and along-axis segmentation (i.e., Ebinger, 2005).

6. LIP-style continental rifting: continental breakup, crustal accretion and volatile exchange

Results from the CRAFTI project reveal the importance of magmatism and magmatic volatile release during continental rifting (Fig. 6). Indeed, LIP-style continental rifting, and its associated magmatism, is critical for the growth of continental lithosphere, volatile exchange between the lithosphere and atmosphere, and initiating continental breakup.

The earliest stages of rifting in the Eastern rift branch are associated with volumetrically significant magmatic underplating beneath the faulted rift valleys and their uplifted flanks. Early in the basin’s formation, discrete magma chambers develop in the lower to middle crust, and they feed smaller dike-sill complex formation at upper crustal levels (<12 km depth; Calais et al., 2008; Muirhead et al., 2015; Weinstein et al., in review) (Fig. 1). In the Natron-Magadi basin, high-resolution tomographic models indicate that a single ~30 km-wide magma chamber feeds the active carbonatitic volcano (Oldoinyo Lengai) as well as stacked sills beneath a monogenetic cone complex (Roecker et al., 2017). The volume of the melt is largely unconstrained, as is the flux rate into the lower crust to sustain this persistent partial melt zone. Extruded volumes for the Kenya Rift are roughly estimated at 310,000 km³ (Guth, 2015). Given that ratios between the volumes of intruded and extruded products are shown to be greater than 3:1 for LIPs and volcanic rifted margins (Coffin and Eldholm, 1994), intruded volumes in the ~600 km-long Kenya Rift could easily exceed 300,000 km³_. Assuming an average age of 15 Ma for basins along this part of the rift, these crude volume estimates suggest that magma flux rates could be on the order of 10s of km³/km/Myr, which is comparable to some volcanic arc systems (23-200 km³/km/Myr estimated for intraoceanic arcs; Jicha and Jagoutz, 2015).

Volatile production rates in the East African Rift are also high. Active volcanoes in EARS represents some Earth’s largest gas emitters, with Oldoinyo Lengai and Nyiragongo emitting 2.4 and 3.4 Mtyr^-1 of CO₂, respectively (Brantley and Koepenick, 1995; Sawyer et al., 2008). These persistently active volcanoes represent a commonly investigated source of volatile transfer to the atmosphere. A more hidden mechanism of volatile exchange occurs via CO₂ ascent along deeply penetrating extensional fault systems, which facilitate degassing of volumetrically significant partial melt detected at middle to lower crustal depths (Lee et al., 2016; Roecker et al., 2017) (Fig. 1). This tectonic degassing represents a poorly constrained source of deeply-derived CO₂ release in extensional rift systems globally. However, in the EARS it is clearly responsible for CO₂ release on the order of 10s of Mtyr^-1_. These observations suggest that magma-rich continental rifts represent an important source of naturally occurring and potentially climate modulating CO₂. Neglecting the short and rapid environmental changes that accompany plume-related magmatism during rift initiation (Burgess and Bowring, 2015), the massive and prolonged CO₂ emissions associated with LIP-style continental rifting could contribute to prolonged greenhouse conditions like those of the Cretaceous (Kidder and Worsley, 2010; Ernst, and Youbi, 2017).

The forces driving rift initiation (e.g., gravitational potential energy, basal shear from mantle convection) are theoretically too small to rupture the thick, slow-moving continental lithosphere beneath East Africa (e.g., Bott et al.,1991; Buck, 2004; Stamps et al., 2014). Our work suggests that additional processes facilitate the earliest stages of continental rifting in East Africa: (1) the widespread percolation of mantle-derived fluids that metasomatize and weaken the mantle lithosphere (Lee et al., 2016; Muirhead et al., 2016; Weinstein et al., in review); (2) heat transfer from progressively shallow magma intrusions (Lee et al., 2017; Roecker et al., 2017); and (3) the buoyancy forces of an upwelling low-density mantle plume (e.g., Ebinger and Sleep, 1998; Stamps et al., 2014). The CRAFTI project also represents the first cross-disciplinary study constraining the importance of the above processes for localizing strain and driving rift basin evolution. Specifically, the high Vp/Vs and seismicity in the lower crust indicate that intrusion is ongoing, consistent with the ongoing and volumetrically large CO₂ degassing (Lee et al., 2016). The Natron basin appears to be at a transition from border-fault controlled rifting to intrusion and shallow faulting above a zone of magma intrusion (Roecker et al., 2017; Weinstein et al., in press). Lastly, areas of long-term extensional fault strain coincide with zones of magmatic volatile release at the surface, suggesting a feedback between fluid flow and strain localization in the upper crust of early-stage continental rift basins (Muirhead et al., 2016).