Nature

VOLCANIC GROWTH FAULTS AND THE ORIGIN OF PACIFIC ABYSSAL HILLS

Ken C. Macdonald
Dept. Geology and Marine Science Institute, University of California, Santa Barbara, CA 93106

P.J. Fox
Ocean Drilling Program, Texas A & M University, College Station, TX 77845

Russ T. Alexander
Albert Einstein Medical School, New York, NY

Robert Pockalny
Graduate School of Oceanography, University Rhode Island, Narragansett, R.I. 02882

Pascal Gente
Universite de Bretagne Occidentale, 29287 Brest Cedex, France

The topographic features known as abyssal hills characterize >30% of the ocean floor, and yet their origin has been the source of vigorous debate for over 40 years. Submersible-based investigations show that Pacific abyssal hills are created on the flanks of the East Pacific Rise by horsts and grabens which lengthen with time. Hills are bounded by ridge-facing scarps produced by normal faulting, and on the other by more gentle slopes produced by volcanic growth faulting.

Although abyssal hills are unspectacular relative to terrains created by continental collisions, subduction-related processes or hotspot volcanism (e.g. Himalayas, Andes, Hawaiian Islands, respectively), abyssal hills are nevertheless the most abundant geomorphic structures on earth. The focus of this paper is the origin of the abyssal hills in the Pacific Ocean. They are typically 10-20 km long, 2-5 km wide and 50-300 m high, are oriented approximately perpendicular to the spreading direction, and cover virtually all of the ocean floor except where they are buried beneath sediment [1, 2, 3, 4]. Dietz [5] first suggested in 1961 that abyssal hills are created by a combination of "intrusion and extrusion and then placed under rupturing stresses as the sea floor moves outward." Ever since then, lively debate has ensued over whether the abyssal hills of the Pacific are created on the spreading axis or at some distance off-axis, and whether the hills owe their relief and shape primarily to volcanic constructional processes, faulting, or some combination of the two (Figure 1) [see, for example, references 3, 6-20].
In the past decade, acquisition of high-resolution acoustic data has focussed the debate over the origin of the Pacific abyssal hills. High-resolution profiles across abyssal hills were collected in a number of locations across the flanks of the East Pacific Rise (EPR) using a narrow-beam echo-sounder on the Deep Tow system. These profiles revealed that the hills are bounded by steep slopes facing both toward and away from the axis of the EPR [11, 13, 21]. These results suggest that inward and outward facing slopes of the abyssal hills are the product of normal faults which have created high standing horsts. The swath maps produced by multibeam sonar systems, have lower slope resolution than the Deep Tow profiles, but multibeam systems provide continuous coverage bathymetric maps and high fidelity images of seafloor features in plan view [22]. Multi-beam bathymetric maps reveal that the abyssal hills flanking the EPR have a distinctive asymmetric shape18. The side of the hill facing the axis of the EPR is usually a steep slope that is continuous and straight along the length of the hill. In contrast, the side of the hill facing away from the ridge axis is typically less steep, is often sinuous rather than straight along-strike, and is decorated with lobate landforms. This asymmetric shape may arise from the interplay of two processes: faulting and volcanic construction (Figure 1) [7, 9, 11, 13, 23]. To determine the relative importance and timing of these processes and to advance our understanding of abyssal hill formation, direct near bottom imaging of the inward and outward facing slopes of abyssal hills was required.
Here we report observations made with the submersible ALVIN, which lead to a model for the formation of Pacific abyssal hills. We find that they are created ~2-6 km away from the axis of the EPR, and are bounded by normal faults dipping toward the axis and "volcanic growth faults" dipping away from the axis. The repeated draping of fault scarps which face away from the spreading axis by syntectonic near-axis lava flows results in a volcanic growth fault whose surface exposure appears to be volcanic, but whose vertical relief is almost entirely faulted in origin. Activity on these outward-dipping volcanic growth faults diminishes ~6 km off-axis. In contrast, inward-dipping faults continue to be active at least 30 km off-axis and produce further growth in length and height of abyssal hills. For fast-spreading centers, the end product is an asymmetric abyssal hill.

Abyssal hill controversy

Independent of the model adopted to explain the relief and shape of the abyssal hills flanking the EPR, it seems clear that the hills are not formed along the axis of the EPR, because almost nowhere along the axis is significant relief created either by faulting or volcanic processes [24]. The axial high itself is not evident on the flanks and seems to disappear off-axis, suggesting that it is created primarily by the buoyancy of hot rock and magma beneath the rise [25]. Indeed, seismic layer 2A (interpreted to be the volcanic layer) [26, 27] is thinnest on the spreading axis and doubles in thickness only 1-4 km off-axis [27, 28]. The added 150-200m of lava is more than enough to inundate even the largest axial summit calderas imaged along the axis of the EPR, so the rare occurrences of significant relief on axis [29 ] are not preserved off axis [25].
The first development of terrain that has the relief and size of an abyssal hill is located several km off-axis on the flanks of the EPR axial high, where narrow troughs are observed on high-resolution maps of both the northern [22, 30] and southern EPR [31, 32]. These troughs are aligned discontinuously along the flanks of the EPR, range in length from a few km to greater than 30 km, and develop relief of tens of meters to over 200m (Figure 2, Figure 3a). Based on an analysis of multibeam maps, the origin of the troughs was ambiguous and they have remained unexplained. While the steep and straight axis-facing slopes of the troughs are suggestive of fault-controlled scarps, the outward facing slopes are not as steep or straight and are often associated with small (10-50 m high) conical structures suggestive of volcanism. Our observations indicate that it is the deepening, lengthening and linking of these troughs as a function of time which define the intervening abyssal hills, thus the origin and tectonics of these troughs are key to the question of abyssal hill origin.

ALVIN investigations of abyssal hills

To further investigate abyssal hills and troughs, we conducted a 17-dive program with the submersible ALVIN on the fast-spreading East Pacific Rise (109 mm/yr,33) between 9°-10°30'N. The relevant data sets collected from ALVIN include continuous video coverage from forward- and down-looking cameras, nearly continuous 35 mm still photographic coverage, and mesotech sonar 150 kHz, super-high-resolution swath bathymetry. Our field investigations spanned fully mature hills on crust ~0.7 my old to those in the initial stages of development on the flanks of the spreading axis in the zone of trough development. Dives on even older hills were not considered due to the obscuring effects of sediment cover.

Mature hills

We focussed our observations on features that can be used to distinguish between the different models proposed to explain abyssal hill formation (Figure 1). For the six hills studied using ALVIN, the sides of hills which face the spreading axis are found to be faulted escarpments, as evidenced by: exposures of sharply truncated pillow lavas, steepness of slopes (60°-90° based on mesotech profiles and direct observations), rectilinearity of scarps when traced along strike, and the abundance of fresh, well-sorted talus at the bases of scarps (in contrast to poorly-sorted whole pillow talus occurs at the bases of flow fronts) [13]. None of the inward facing faults exhibit the glassy cooling ledges of a subsiding lava lake which are so common along the walls of the axial summit caldera [34, 35]. This offers support for the idea that inward-dipping faults are not fossil axial summit caldera walls. We also observe that the tops of the hills are within a few degrees of horizontal, based on fine scale mesotech bathymetry, and constrained by observations of horizontal cooling ledges on lava pillars36 in a fossil lava lake at the top of a hill 6 km off-axis. Based on these observations, the back-tilted fault block and whole volcano hypotheses are eliminated (Figure 1). On the slopes of mature hills which face away from the axis, we observe intact lava flows (primarily elongate pillows)which are consistent with either the split volcano or horst/graben hypotheses.

Birth of hills

After establishing the asymmetric character of fully developed abyssal hills located on crust up to 35 km off-axis, we focused on a few selected troughs, located 2-6 km off-axis, to investigate the initiation of abyssal hill development (Figure 2, Figure 4a). As with the mature abyssal hills, the inward facing slopes of the troughs are fault-controlled scarps with relief of 10-60 m, while the outward facing slopes are dominated by elongate pillow lavas. On along-strike traverses of the outward facing slopes, we crossed well-sorted fault talus every few hundred meters. Maneuvering the submersible uphill from these talus ramps, we encountered faulted scarps 5-25 m high, dipping 60°-90° away from the spreading axis. At the tops of the scarps are fresh elongate pillow lavas in a frozen cascade over the near-vertical scarps of sharply truncated pillow lavas. Traveling north and south along strike (Figure 4a) we found fault scarps which include those that are entirely exposed; partly exposed with some sections covered by elongate pillows; or entirely draped by lavas flowing from the direction of the spreading axis (demonstrating that the source of the lava flow was either the axis itself or volcanic centers within 2-3 km of the axis). Many of the draping elongate pillows have pinched off and elongate pillow sections and tubes have fallen ~10m to the base of the scarps.
In a few cases, faults could be traced to their termini along strike. The fault relief decreases toward the ends of the faults. Where deformation does not relay to neighboring en echelon faults, the faults invariably terminate in fissures 1-3 m wide which narrow to closure within ~10 m along strike. The ends of exposed faults are not characterized by lava burial or by warping of overlying lava flows.
On 13 dives (including 4 from earlier work in the area [37]) we studied 3 near-axis troughs centered at 1.5 km, 2.5 km and 5 km away from the axis. The troughs appear to be successively more developed (i.e., exhibit greater relief and along-strike continuity) as a function of distance away from the spreading axis. The maximum depths of the troughs increase away from the axis, from 30 m to 100 m to 200 m respectively, while the lengths of the troughs increase from 3 km to 6 km to 11 km respectively. Eighteen km of zig-zag traverses along and up and down the outward facing slopes of these 3 near-axis troughs show the same pattern of exposed active fault scarps partially draped by elongate pillow lavas.

Volcanic growth faults

The outward facing slopes of the hills are neither simple outward dipping normal faults, as would be predicted by the horst/graben model, nor are they entirely volcanic-constructional, as would be predicted by the split volcano model. The near-axis troughs, which define the edges of embryonic abyssal hills, develop 2-6 km off-axis in a "flanking tectonic province," instead of developing directly on the spreading axis as the split volcano model (proposed for intermediate spreading ridges) would predict [9, 23].
We interpret the outward facing slopes as volcanic growth faults (Figure 4b). Small scarps produced by episodes of normal faulting are buried near the axis by syntectonic lava flows originating either at the axis or between the flanking tectonic province and the axis (i.e. within ~2-3 km of the axis, [e.g.37, 38]). Repeated episodes of dip-slip faulting and volcanic burial result in structures resembling growth faults, except that the faults are episodically buried by lava flows rather than being continuously buried by sediment deposition. An important difference between volcanic growth faults and traditional sedimentary growth faults is that occasionally enough dip-slip offset may accumulate (>10m) between episodic volcanic eruptions so that pillow lavas which flow over the scarp elongate to the point of necking (Figure 4a). This failure to drape the scarp completely and continuously leaves sections of the fault scarp exposed for some period of time. Based on these observations, the abyssal hills are horsts and the intervening troughs are grabens with the important modification to the horst/graben model that the outward facing slopes are created by volcanic growth faulting rather than traditional normal faulting. Thus, volcanism is a more important aspect of the morphology of abyssal hills than previously appreciated in the horst/graben model.
Multibeam bathymetric data and side-scan sonar records support the interpretation that outward dipping volcanic growth faults accumulate nearly all of their throw within ~6 km of the axis (within the flanking tectonic province, Figure 5). These faults are thus near enough to sources of volcanism, either the neovolcanic zone or near-axis sources, to be subject to burial throughout their development [39, 40]. In contrast, inward dipping faults act as tectonic dams to lava flows rather than being buried by them [39, 40]. They continue to be active >30 km from the spreading center in this area41, 42 resulting in continued growth (increase in length and height) and modification (e.g. minor outward tilting of 2º-3º) of some of the hills off-axis (Figure 4a and Figure 4b).

Lengthening of hills

To understand the temporal and spatial interplay of volcanism and tectonism in the development of abyssal hills along the axis of fast-spreading EPR, it is necessary to consider the process from a plan view perspective (Figure 3a). We observe that grabens which define the edges of abyssal hills lengthen as a function of distance off-axis (Figure 5, top). If the process is relatively steady-state, this means that they lengthen as a function of time. We propose that the near-axis grabens behave as an array of cracks, lengthening and linking via crack propagation along axis as the lithosphere is stretched within the plate boundary zone [43] (Figure 3b). In this way, abyssal hills begin as embryonic horsts and grabens that develop in height and length as a function of time and distance from the spreading axis. The integrated result of these processes is an asymmetric abyssal hill bounded by a steeply dipping normal fault on the side facing the spreading axis, and bounded by a volcanic growth fault on the opposing side.

Concluding Remarks

This process describes the origin of most abyssal hills at fast-spreading centers characterized by an axial high. At slow- to intermediate-rate spreading centers and adjacent to ridge axis discontinuities, other processes are likely to be important. Near ridge axis discontinuities on fast-spreading centers, the lithosphere appears to be sufficiently thick to support axial volcanic constructions [25]. Subsequent propagation and decapitation44 of the end of the spreading segment results in the rafting off of whole volcanoes as abyssal hills (Figure 1c). At intermediate-rate spreading centers, abyssal hill structure may vary with the local magmatic budget. Where the budget is starved and the axis is characterized by a rift valley, abyssal hills are generally back-tilted fault blocks12. Where the magmatic budget is robust and an axial high is present, the axial lithosphere is episodically thick enough to support a volcanic construction which may then be split in two along the spreading axis, resulting in split volcano abyssal hills [9, 25 ] (Figure 1d). At slow-spreading centers characterized by an axial rift valley, back-tilted fault blocks and half-grabens may be the dominant origin of abyssal hills (Figure 1a), although there is continued controversy over the role of high- versus low angle faults, listric faulting versus planar faulting, and the possible role of episodes of volcanism [17, 45, 46, 47].

We thank the U.S. Office of Naval Research for funding this program; diving scientists Daniel Scheirer, Corrine Fleutelot, Nellena Beedle for their expertise; M. Perfit and D. Fornari [37] for use of observations from a previous dive program; Jeff Karson, John Goff and Rachel Haymon for very sage reviews, Antoinette Padgett for graphics; and the ALVIN group, captain and crew of Atlantis II for their superb performance.

References

1. Dietz, R.S., Menard, H.W. & Hamilton, E.L. Deep-Sea Res. 1, 258-272 (1954).
2. Menard, H.W. Science 132, 1737-1746 (1960).
3. Searle, R. Tectonophysics 101, 319-344 (1984).
4. Goff, J.A. J. Geophys. Res. 96, 21,713-21,737 (1992).
5. Dietz, R.S. Nature 190, 854-857 (1961).
6. Larson, R.L. Geol. Soc. Amer. Bull. 82, 823-841 (1971).
7. Luyendyk, B.P. Geol. Soc. Am. Bull. 81, 2237-2260 (1970).
8. Anderson, R.N. & Noltimier, H.C. Geophys. J. Roy. Astr. Soc. 34, 137-147 (1973).
9. Kappel, E.S. & Ryan, W.B.F. J. Geophys. Res. 91, 13925-13940 (1986).
10. Rea, D.K. Geology 3, 77-80 (1975).
11. Lonsdale, P. Mar. Geophys. Res. 3, 251-293 (1977).
12. Klitgord, K.D. & Mudie, J.D. Geophys. J. Roy. Astr. Soc. 38, 563-586 (1974).
13. Macdonald, K.C. & Luyendyk, B.P. Mar. Geophys. Res. 7, 515-535 (1985).
14. Rona, P.A., Harbison, R.N. & Bush, S.A. Mar. Geol. 16, 275-292 (1974).
15. Goff, J.A., Malinverno, A., Fornari, D.J. & Cochran, J.R. J. Geophys. Res. 98, 13851-13862 (1993).
16. Smith, D.K. & Cann, J.R. J. Geophys. Res. 97, 1645-1658 (1992).
17. Gente, P., et al. Earth Planet. Sci. Lett. 129, 55-71 (1995).
18. Goff, J.A. & Jordan T.H., J. Geophys. Res 93, 13589-13608 (1988).
19. Carbotte, S.M. & Macdonald, K.C. J. Geophys. Res. 99, 13609-13631 (1994).
20. Edwards, M.E., Fornari, D.J., Malinverno, A., Ryan, W.B.F. & Madsen, J. J. Geophys. Res. 96, 7995-8017 (1991).
21. Bicknell, J.D., Fox, P.J., Sempere, J.-C. & Macdonald, K.C. Mar. Geophys. Res. 9, 25-45 (1987).
22. Wilcock, W.S.D., Toomey, D.R., Purdy, G.M. & Solomon, S.C. Mar. Geophys. Res. 15, 1-12 (1993).
23. Lewis, B.T.R. Geophys. Res. Lett. 6, 753-756 (1979).
24. Macdonald, K., Sempere, J.-C. & Fox, P.J. J. Geophys. Res. 89, 6049-6069 (1984).
25. Carbotte, S.M. & Macdonald, K.C. Earth Planet. Sci. Lett. 128, 85-97 (1994).
26. Anderson, R.N., et al. Nature 300, 589-594 (1982).
27. Christeson, G.L., G.M. Purdy, G.J. Fryer. J. Geophys. Res. 99, 17,957-17,973 (1994).
28. Vera, E.E. & Diebold, J.B. J. Geophys. Res. 99, 3031-3041 (1994).
29. Renard, V., Hekinian, R., Francheteau, J., Ballard, R.D. & Backer, H. Earth Planet. Sci. Lett. 88, 339-353 (1985).
30. Macdonald, K.C., et al. Mar. Geophys. Res. 14, 299-344 (1992).
31. Scheirer, D.S., et al. Mar. Geophys. Res. (in press).
32. Cochran, J.R., et al. Mar. Geophys. Res. 15, 65-75 (1993).
33. Carbotte, S. & Macdonald, K.C. J. Geophys. Res. 97, 6959-6982 (1992).
34. Haymon, R.M., et al. Earth Planet. Sci. Lett. 104, 513-534 (1991).
35. Haymon, R.M. & others, a.1. Earth Planet. Sci. Lett. 119, 85-101 (1993).
36. Francheteau, J., Juteau, T. & Rangan, C. Nature 281, 209-211 (1979).
37. Perfit, M.R., et al. Geology 22, 375-379 (1994).
38. Macdonald, K.C., Miller, S.P., Luyendyk, B.P., Atwater, T.M. & Shure, L. J. Geophys. Res. 88, 3403-3418 (1983).
39. Edwards, M.H. The morphotectonic fabric of the East Pacific Rise: Implications for fault generation and crustal accretion (Columbia, 1991).
40. Hurst, S.D & Karson, J.A., Tectonics 13, 789-802 (1994).
41. Alexander, R.T. & Macdonald, K.C. Mar. Geophys. Res. (in revision).
42. Lee, S.-M.a.S.C.S. Geophys. Res. Lett. 22, 3135-3138 (1995).
43. Macdonald, K.C. Ann. Rev. Earth Planet. Sci. 10, 155-190 (1982).
44. Macdonald, K.C., Sempere, J.-C., Fox, P.J. & Tyce, R. Geology 15, 993-997 (1987).
45. Mutter, J.C. & Karson, J.A. Science 257, 627-634 (1992).
46. Karson, J.A., et al. Nature 328, 681-685 (1987).
47. Pockalny, R.A., Detrick, R.S. & Fox, P.J. J. Geophys. Res. 93, 3179-3193 (1988).
48. Wilcock, W.S.D., Purdy, G.M., Solomon, S.C., DuBois, D.L. & Toomey, D.R., Geophys. Res. Lett. 19, 2131-2134 (1992).