Introduction

Sediment burial in submarine depocenters drives biogeochemical processes that sequester organic carbon over geological timescales, but also convert part of it into mobile hydrocarbons, principally methane (Dickens et al. 2004). In the deep sea, methane that migrates toward the seabed may be stored in gas hydrates and/or authigenic carbonates (Arning et al. 2013), or released to the oceans by seafloor venting (Römer et al. 2014). Recent studies of continental slope settings suggest seafloor gas venting to be a more widespread phenomenon than previously recognized (Skarke et al. 2014). While most vented gas may dissolve in the water column (McGinnis et al. 2006), a portion may reach the atmosphere (Solomon et al. 2009). The latter possibility is of interest in relation to carbon stable isotopic evidence that methane emissions are 60–110% greater than current global estimates (Schwietzke et al. 2016). An improved understanding of the occurrence and mechanisms of deep-sea gas venting is thus of importance for our understanding of global carbon cycling in the context of ongoing climate-driven changes (Biastoch et al. 2011), of mass transfers between sediments and the ocean (Lamaitre et al. 2014), and of marine environmental and ecosystem dynamics (Hovland et al. 2012, and references therein).

Seafloor gas venting has been investigated in a variety of deep-sea settings worldwide, all associated with known or inferred occurrences of gas hydrates (Hovland et al. 2012). Gas plumes have been observed to align along the edge of the regional gas hydrate stability zone (GHSZ) in several areas, suggesting the climate-driven dissociation of gas hydrates (Westbrook et al. 2009; Skarke et al. 2014; Hautala et al. 201; Johnson et al. 2015). However, gas venting also takes place at greater water depths within the GHSZ (Olu-Le Roy et al. 2007; Praeg et al. 2014; Johnson et al. 2015; Andreassen et al. 2017). Gas venting from the GHSZ has been explained in terms of mechanisms to elevate or modify the phase boundary (see Liu and Flemings 2007; Haacke et al. 2009), or alternatively by the creation of fluid migration pathways along faults or fractures (e.g., Hornbach et al. 2007; Riedel et al. 2010), or by seafloor erosion (Skarke et al. 2014). Deep structural pathways are recognized to be important for fluid migration through the GHSZ on convergent margins, and there is evidence that they also play a role on passive margins, notably where Cenozoic depocenters experience gravitational collapse above detachment surfaces (Rowan et al. 2004).

The Amazon River has one of the world’s highest sediment discharges (Damuth and Kumar 1975) and is the largest supplier of terrestrial organic carbon to the oceans (31–55 × 1012 g carbon per year; Richey et al. 1980). It culminates in one of the world’s largest deep-sea fans, a Neogene depocenter within which loading drives ongoing gravitational collapse (Reis et al. 2010, 2016). The presence of gas hydrates in the Amazon deep-sea fan has been inferred from observations of a discontinuous bottom simulating reflector (BSR; Manley and Flood 1988; Tanaka et al. 2003; Berryman et al. 2015) and from pore water freshening in ODP 155 leg cores taken between 2400–4100 m depth below seafloor (Fig. 1; Piper et al. 1997). Here, we present the first evidence of gas release to the water column from the Amazon fan, using multibeam echosounder (MBES) imagery acquired across an area of 18,000 km2, in water depths of 250–3500 m (Fig. 1), which also allows us to map the distribution of gas flares in relation to seafloor vents and structures. We also present geochemical analyses of the first gas hydrate samples recovered from the NE continental margin of South America. Our findings provide new insights into the interaction of climate-driven changes and/or seasonal changes in water temperature in gas hydrate stability with gravitational tectonics on large deep-sea fans and their contribution to the global carbon cycle, and will help to guide future research investigations on global climate change, geohazards, and energy resources.

Fig. 1
figure 1

(a) Location map showing the upper Amazon fan study area (black polygon) and the multibeam echo sounder survey area in inset. Black hexagons show the location of (one or more) plumes; stars represent sites where gas hydrates were recovered in cores. White dots show the location of water column (XBT or CTD) measurements. Labels 2A, 2B, 2C, 2E, and 2F are viewpoints for Fig. 2(A, B, C, E, and F, respectively). Line D–D’ indicates the position of the seismic line in Fig. 2D. Red dots indicate sites drilled during ODP Leg 155 (Piper et al. 1997). The isobath line of – 600 m (in red) represents the approximate depth of the edge of the methane hydrate stability zone in the area. The two scale bars refer to the location map (lower bar) and multibeam echo sounder survey area in inset (upper bar). (b) Graphic representation of water column temperatures from XBT and CTD relative to the equilibrium field for pure methane hydrate in seawater (dashed line, from Dickens and Quinby-Hunt 1994)

Full size image

Geological setting

The Amazon fan extends 700 km seawards from the continental shelf to water depths of over 4000 m in the Demerara Abyssal Plain, occupying an area of 330,000 km2 (Fig. 1). The fan is a depocenter up to 10 km thick that has built out rapidly since the Late Miocene in response to sedimentary supply linked to Andean uplift (Figueiredo et al. 2009). Loading is driving the gravitational collapse of the fan above deep décollements, expressed at seabed in paired extensional–compressional belts on the shelf and upper slope above water depths of 2000 m, including thrust-folds up to 500 m in seafloor relief (Silva et al. 2009; Reis et al. 2010, 2016). The upper slope is also the source area of giant mass-transport deposits (MTDs) that characterize the Neogene stratigraphic succession of the Amazon fan (Reis et al. 2010, 2016; Silva et al. 2016). The triggering of giant MTDs has been linked to climate-driven variations in sedimentation rate and/or gas hydrate stability during sea level changes, based on drilling results from four examples on the middle fan emplaced within the last 45 ka (Piper et al. 1997; Maslin et al. 1998, 2005). More recently, the emplacement of MTDs throughout the Neogene has been linked to tectonic activity within the upper slope compressional belt (Reis et al. 2016).

Data and methods

Seafloor features and water column gas flares were identified using Kongsberg EM122 12-kHz multibeam echosounder data acquired in 2012–2013 (horizontal grid size 30 × 30 m), visualized using QPS Fledermaus software, including the Midwater module. Subsurface structures were examined using two sets of multichannel seismic data, acquired for hydrocarbon exploration (made available by the Brazilian National Petroleum Agency - ANP) and interpreted using IHS Kingdom Suite software: (i) ca. 20,000 km of 2D seismic profiles acquired during the 1980s and 1990s by different companies using different air-gun acquisition arrays, with record lengths of up to 10 s, frequency range between ~ 8–60 Hz (dominant frequency at ~ 40 kHz), and vertical resolution of ~ 10m; (ii) 3D blocks covering a total area of 3800 km2, record length of 8 s; frequency range between 12 and 100 Hz (dominant frequency at 80 kHz); vertical resolution of ~ 5 m).

Water column temperatures were obtained during the multibeam echo sounder survey for determination of the sound velocity using 10 XBT (expendable bathythermograph) measurements and one CTD (conductivity, temperature, depth) measurement (Fig. 1). The temperature vs. depth profiles generated by the XBT and CTD were also used to calculate the limits of the methane hydrate stability zone (MHSZ) in the area, applying the Dickens and Quinby-Hunt (1994) equilibrium equation for pure methane hydrates in seawater. Different gas compositions (e.g., a mixture of carbon dioxide and methane, or heavier hydrocarbons) will dislocate the calculated limit of GHSZ to shallower depths (upslope). The XBT and CTD data showed a significant variation in temperatures for the same depth (ca. 1.7 °C) and, therefore, we estimated the edge of the GHSZ for a depth interval. Our time-limited temperature measurements may not cover the full range of natural variability of water temperatures in the area and, therefore, the estimated depth range is approximated.

Piston cores up to 6 m long were obtained from selected sites in 2015 using the R/V Teknik Perdana. Transparent core liners were used to allow visual identification of gas hydrates on deck. Gas hydrates observed in three cores were immediately sampled using a spatula and placed in IsoJar containers at room temperature. Headspace gas composition within Isojars were determined using a GC-2014 Shimadzu gas chromatograph equipped with a VP-Plot Alumina/KCl capillary column. Stable isotopes of carbon were analyzed using a Thermo Fisher Scientific gas chromatograph coupled to a DELTA-V Plus mass spectrometer via an IsoLink and ConFlo IV interface. Isotopic data is reported using the delta notation (δ13C) in parts per thousand (‰) with ratios relative to the international standard Vienna Pee Dee Belemnite (V-PDB).

Results

Seafloor seeps and water column gas plumes

Analyses of MBES data reveal 53 hydroacoustic anomalies within the study area, interpreted as gas plumes that rise sub-vertically from the seafloor to 85–1720 m in the water column (Figs. 1 and 2). Observations from seepage sites worldwide show that such anomalies may be formed of several individual bubble streams, located meters to tens of meters apart (Römer et al. 2012, 2016). In the Amazon study area, the gas plumes are associated with different seafloor signatures that allow us to distinguish four types of seeps (Fig. 2). Type I seeps are observed in water depths of 650–715 m (Fig. 2(A)), where a high density of plumes (23 in total) is observed in three main clusters, each associated with a seafloor mound 10–20 m high, similar to plumes of types II and III described below.

Fig. 2
figure 2

Multibeam echo sounder imagery of sea floor seeps associated with water column gas plumes: (a) cluster near the upper limit of the methane hydrate stability zone (the dashed line corresponds approximately to the calculated limit of the stability zone at 620 m water depth—type I seeps); (b) plume associated to the fold and thrust belt outcropping on the seafloor (type II seep); (c) aligned along a geological structure on the sea floor (dashed line corresponds to the trace of the seismic line in 2D—type II seeps); (d) seismic profile crossing the fault (arrows) shown in 3C (dashed line), and a detail of the profile showing the bottom simulating reflector (yellow line) and the structure as a deep fault (red line); (e) type III seeps associated with sea floor mounds; (f) type IV seeps within canyons north of the Amazon fan. See Fig. 1 for locations of the images

Full size image

Type II seeps show a clear spatial relationship with seafloor lineaments observed on MBES imagery (Fig. 2(B and C)), which seismic profiles show correspond to structures related to the compressional thrust-fold belt of the upper Amazon fan (Fig. 2(D)). A total of 17 seeps are observed in water depths of 1000–1800 m and are associated with seafloor mounds 10–20 m high located on or a few hundred meters from the lineaments (Fig. 2(C)). Type III seeps occurring at seven sites in the same area as Type II are also associated with seafloor mounds 10–20 m high (Fig. 2(E)) but are not clearly associated with seafloor lineaments. Type IV seeps correspond to six sites observed north of the Amazon fan proper, within slope-confined canyons in water depths between 1000 and 2600 m (Fig. 2(F)).

Considering all observed seeps, there is a clear correlation between water depth and the height of gas plumes (Fig. 3). Interestingly, we observe that the top of the gas plumes does not necessarily coincide with the top of the MHSZ (Fig. 3). Type I plumes (< 715 m water depth) are less than 350 m high and rise a few tens to hundreds of meters above the top of the MHSZ at 550–620 m water depth (Fig. 1). Type II–IV plumes (water depths > 1000 m) are not observed to rise above the MHSZ (Fig. 3).

Fig. 3
figure 3

Plots showing a linear correlation between: (a) water depth and the height of gas plumes, and (b) depth of the top of plumes versus seafloor depth (the gray area represents the depth range of the methane hydrate stability zone for pure methane hydrate in equilibrium with seawater in the study area, cf. Dickens et al. 1994; see Fig. 1 inset)

Full size image

Gas hydrates

The composition of venting gases was examined using samples of hydrates recovered from three sites. Two samples (GH2 and GH3) are from the Amazon fan proper; the third (GH1) is from a site on the canyonised continental slope some 50 km to the west (Fig. 1). The two samples from the Amazon fan proper both consist of small hydrate nodules (1–10 mm diameter) dispersed in a muddy matrix (Fig. 4(A)), and can be described as “nodular type” (Malone 1985). The gas hydrate nodules occur in association with authigenic carbonate nodules (millimeter to centimeter size), as well as fragments of a possible chemosynthesis-based community (shells, tube worms; Fig. 4(B)). The gas contained in the hydrates is predominantly methane (85–98%), with variable amounts of CO2 (2–15%), and trace amounts of ethane (Table 1). Methane δ13C values range from − 81.1 to − 77.3‰ (Table 1).

Fig. 4
figure 4

Photographs of sediments recovered in piston cores from the Amazon deep-sea fan, showing (a) nodule of gas hydrate (arrow) in sediments (sample GH3; see Fig. 1 for location), and (b) carbonate concretion with shells (black arrows) and tubeworm (white arrow) fragments found together with GH3 gas hydrate sample

Full size image
Table 1 Chemical and isotopic composition of gases trapped in 3 gas hydrate samples recovered in piston cores of three sites (GH1, GH2, and GH3) in the Amazon fan. See Fig. 1 for site location
Full size table

In contrast, the gas hydrates recovered west of the Amazon fan from a site on the canyonised continental slope consist of pieces up to 10 cm in length that can be described as “massive type” (Malone 1985). Gas analyses show methane to dominate (95%), followed by CO2 (4%) and ethane (1%; Table 1). The methane has a δ13C value of − 59.2‰ (Table 1).

Discussions

The results presented above allow us to identify two mechanisms that promote ongoing gas seepage from the Amazon fan over Plio-Quaternary timescales. One is the dissociation of gas hydrates along the edge of their stability zone (Fig. 5), inferred to be the cause of 40% of the seeps mapped in our study. Such a process has been recognized on other continental margins and may be driven by warming of ocean bottom waters over post-glacial or anthropogenic timescales (Skarke et al. 2014; Hautala et al. 2014; Johnson et al. 2015) and/or by seasonal variations in bottom water temperatures (Berndt et al. 2014). A second mechanism is methane migration through the MHSZ to the seafloor along structural pathways, inferred to account for 60% of the seeps that we map on the upper Amazon fan (Fig. 5). We infer this process to have operated throughout gravitational collapse of the fan, which has been underway at least since the Pliocene (Reis et al. 2016).

Fig. 5
figure 5

Schematic diagram showing the Amazon discharge of sediment and organic matter from the continent to the deep sea. Part of the organic matter deposited in the deep-sea fan will be altered to methane by microbes, which will, in turn, migrate upward to be trapped as free gas, gas hydrates and authigenic carbonates. However, another part will seep from the seafloor, either along the edge of the gas hydrate stability zone in response to, for instance, bottom water warming, or along structural pathways including extensive and compressive fault systems formed by gravitational collapse of deep-sea fans. For reference, the total mass of carbon stored in sediments of the Amazon fan from Late Miocene to Recent is equivalent to several times the quantity stored at present-day in the Amazon Forest (carbon budgets presented in this figure are from Richey et al. 1980; Schlünz et al. 1999; Brienen et al. 2015)

Full size image

The type I seeps lie in water depths of 650–715 m, near the upper limit or “feather edge” of the methane hydrate stability zone (MHSZ), which our measurements of water column temperatures suggest to lie in water depths of 560–620 m, assuming pure methane hydrate in equilibrium with seawater (Fig. 1). A wider depth range (e.g., 520–680 m) results if water temperature measurements from the World Ocean Data Base (WODB; Boyer et al. 2013) are used to calculate the limits of the methane hydrate stability zone. This suggests that the distribution of type I seeps may be related to the dissociation of gas hydrates near the edge of their stability field (Hautala et al. 2014). Gas hydrates in such a setting are sensitive to changes in pressure and bottom water temperatures, even if the latter are small (< 1 °C) and of short duration (< 1 year; Berndt et al. 2014). Warming of bottom waters since the last glacial maximum, including in response to present-day climate change, has been proposed as the main cause of gas hydrate dissociation at the upper edge of the stability zone in several basins around the world (Westbrook et al. 2009; Skarke et al. 2014; Hautala et al. 2014). During last glacial maximum (LGM), colder bottom water temperatures in these depths imply the stability zone to have been shallower/thicker, so that the type 1 seeps lie in an area in which thinning may have liberated free gas accumulated within gas hydrates. Venting implies that free gas is migrating along pathways through the thin GHSZ, which might have been formed by the overpressures created during dissociation of gas hydrates (Stranne et al. 2017). The liberation of free gas in sediments owing to slow dissociation of gas hydrates since the LGM has been modeled for other areas (Foucher et al. 2009; Stranne et al. 2016). Alternatively, the dissociation of gas hydrates at the edge of the gas hydrate stability zone could be caused by seasonal variations in bottom water temperatures (not necessarily related to climate change), promoting the periodic formation and dissociation (with methane venting) of near seafloor gas hydrates (Berndt et al. 2014).

The depth range of type II and III seeps and associated bottom water temperatures (3.5–5 °C) lies well within the stability zone of methane hydrates, indicating that neither are related to phase boundary changes driven by ocean warming. The spatial relationship between type II seeps and gravity-driven structures at the seafloor, which in depth correspond to transfer and thrust faults (Fig. 2(D)), indicates that the gravitational collapse of the Amazon fan creates active pathways for deep gas migration through the GHSZ until the seafloor. A more detailed study is underway to investigate whether type III seeps may be connected with gas migration through buried structures. The origin of gas in the area of lower sedimentation rate adjacent to the Amazon fan (type IV seeps) is unclear, but could either be deep-sourced and brought to the surface via faults, or shallow-sourced and leaking in response to erosional processes such as canyon/gully incisions, and/or mass wasting, which can physically cut gas hydrate deposits or accumulations of free gas in subjacent sediments (Skarke et al. 2014).

Our observation that there is a linear, positive correlation between water depth and the height of gas plumes, and that bubbles become totally dissolved and/or unresolvable by echosounding techniques before reaching the top of the MHSZ (seep types II, III, and IV), is indicative that coating of bubbles by a protective hydrate skin exerts influence (Rehder et al. 2002), but it is not the single dominant control of the permanence of gas flares in the water. The variability of plume heights in the same water depth, both within and above the MHSZ, may suggest that other controls like pressure (depth), temperature, original bubble size, and/or gas composition may exert important controls too (McGinnis et al. 2006; Waite et al. 2017).

The molecular and stable carbon isotope compositions of methane trapped in hydrates in the two sites in the Amazon fan (sites GH2 and GH3; Fig. 1 and Table 1) indicate a dominant biogenic origin. This is consistent with results from disseminated and nodular gas hydrates sampled in other basins worldwide, particularly in areas of high sedimentation rates and large organic matter accumulations, such as other deep-sea fan structures (Miller et al. 2015). Gas hydrates in the adjacent continental slope (site GH1; Fig. 1 and Table 1), however, show different molecular and stable carbon isotopic compositions. Although a biogenic origin is still indicated, the presence of heavier hydrocarbon (ethane) and depletion in the 12C isotope in the methane gas suggests a possible thermogenic contribution, distinct from the biogenic character of the two samples from the Amazon fan proper. This result supports the idea that deep-sea fans such as the Amazon are loci for the accumulation of large quantities of organic matter (Schlünz et al. 1999) that, in turn, provide the source for biogenic methane production (Arning et al. 2013), which may mask thermogenic signatures in gas seeps at the seafloor.

The Amazon and other large deep-sea fans have been suggested to act as net carbon sinks, in which rapid deposition and shallow methanogenesis favor carbon storage within methane hydrates, and as authigenic carbonates (Arning et al. 2013). Our results show that part of the organic carbon stored within the Amazon fan is being released by methane venting through the seafloor to the ocean (Fig. 5). Modeling of biogenic methane formation in Amazon fan sediments at ODP sites at water depths between 3000–4000 m (see Fig. 1 for location) indicates that methanogenesis occurs at burial depths greater than 10 m down to hundreds of meters below the sea floor (Arning et al. 2013), possibly mobilizing carbon stored thousands to hundreds of thousands of years ago. This ongoing process of methanogenesis could be the source for the venting sites described here or, alternatively, venting gas could be sourced by deeper (and older) methane reservoirs.

More generally, our work demonstrates that large deep-sea fans like the Amazon are complex systems that play an important role in the accumulation of organic carbon. On one hand, they sequester significant amounts of organic carbon, which is, in the case of the Amazon fan (e.g., 62 × 1016 g carbon deposited only during the last 20 k.y.; Schlünz et al. 1999), equivalent to several times the quantity stored at present day in the Amazon Forest (150–200 × 1015 g carbon; Brienen et al. 2015). On the other hand, they are characterized by long-term geological dynamics (e.g., gravitational collapse) and shorter-term sensitivity to climate changes and/or seasonal variations in bottom water temperatures (e.g., gas hydrate instability) that favor the release of part of this carbon to the oceans. Gravitational collapse of depocenters above shale or salt detachment surfaces is increasingly recognized in passive margin settings as a long-term (millions of years) component of their evolution (Rowan et al. 2004; Boyd et al. 2011; Tassy et al. 2015). It has been argued that most methane emitted from seafloor seeps will not normally reach the atmosphere, owing to dissolution and oxidation in the water column (McGinnis et al. 2006; Mau et al. 2007), but it has been proposed that catastrophic slope failure may allow massive methane release from gas hydrates (Maslin et al. 2004; Maslin et al. 2005). The dynamics of carbon cycling in deep-sea fans are still poorly understood, so the nature and impact of methane emissions from these systems remain speculative and may be significantly underestimated (Römer et al. 2012; Skarke et al. 2014).

Conclusions

Our study documented for the first time the occurrence of widespread gas seeps, both as seafloor structures and acoustic anomalies in the water column, in the Amazon deep-sea fan and adjacent continental slope area. The observed gas flares lie in two main settings: (1) near the limit of the gas hydrate stability zone (40% of the observed features), possibly related to the dissociation of gas hydrates owing to warming of ocean bottom waters, over post-glacial or anthropogenic timescales, and/or seasonal variations in bottom water temperatures; and (2) along the trace of seafloor faults (60% of the observed features) marking deeper structures related to the gravitational collapse of the fan. The latter shows that gas can migrate through the gas hydrate stability zone to reach the deep oceans, and suggests that this process is facilitated by the gravitational collapse of passive margin depocentres. Gas in the continental slope area adjacent to the Amazon fan possibly reaches the seafloor via faults, or is shallow-sourced and leaking in response to erosional processes such as canyon/gully incisions, and/or mass wasting, which can physically cut gas hydrate deposits or accumulations of free gas in subjacent sediments.

Our results also include analyses of the first gas hydrate samples recovered from the NE continental margin of South America. The molecular and isotopic compositions of gas trapped in hydrates in sediments of the Amazon fan proper are consistent with a biogenic origin (dominantly methane with 2–15% of CO2; δ13C from − 81.1 to − 77.3‰), whereas those on the adjacent continental slope to a possible thermogenic contribution to the gas mix (methane 95%, CO2 4%, and ethane 1%; δ13C – 59.2‰). These results are consistent with the generation of large amounts of mobile biogenic methane within the Amazon fan, which may be diluting contributions from deeper thermogenic sources as observed on the adjacent slope.

This paper demonstrated the potential of deep-sea fans for studies of gas hydrate systems as an interaction between externally driven climate changes and the internal dynamics of passive margin depocenters. We suggest a systematic study of large deep-sea fans on the world’s continental margins to refine the global methane emission and gas hydrate inventories, and improve our understanding of the natural versus anthropogenic drivers of climate change.