Submarine Canyons: Multiple Causes and

Long-Time Persistence'
FRANCIS P. SHEPARD^

ABSTRACT ty on the rock walls of incipient canyons has been

almost completely neglected in explanations, and
The actual investigation of submarine canyons as various types of currents such as those of the tides have
field work was begun about 50 years ago. A large been left largely out of the picture. Perhaps the most im-
amount of factual information has accumulated as a portant feature absent in these various hypotheses has
result of operations of deep diving vehicles, first in the been the realization that canyons may well be the result
Pacific Coast canyons and more recently in the of a long period of formation, much longer than the
remarkable dives of the Woods Hole minisubmarine short episodes of Pleistocene glacial sea-level lowering
Alvin into East Coast canyons. Taking the results of usually considered explanation enough of these giant
these recent dives and combining them with earlier in- features which commonly cut into hard crystalline rock.
vestigations, including much work done by the French New information is showing that the canyons may date
in the Mediterranean as well as our extensive studies off back to at least the Cretaceous.
California and Baja California, we can now say with
some confidence that these amazing deep excavations INTRODUCTION
into the sea floor off so many coastal areas can be ex-
plained. Having started my investigation of submarine can-
New methods such as side scanning have also given us yons some 50 years ago (Shepard, 1932) and having in-
a greater understanding of the exact character of sub- vestigated these great valleys of the sea floor in
marine canyons, particularly in the Bay of Biscay. The numerous places all over the world during the years that
development of multichannel sonar has greatly increas- followed, I have gradually been able to build up a mass
ed our knowledge of the nature of continental margins of factual information which coordinates remarkably
and hence their history. This has given us more insight with the results of the deep dives made recently into East
into the history of canyon development, particularly off Coast canyons off New England and in the Bahamas by
the east Coast where drilling for oil and gas has become geologists in the Woods Hole minisubmarine Alvin
so important. (Ryan et al, 1978; Hooke and Schlager, 1980). These
In the past we have seen a great variety of hypotheses results can now be used to explain the origin of sub-
for explaining submarine canyons. Unfortunately marine canyons. Any explanation which explains can-
almost all of these have been based on information from yons of the sea floor only on the basis of some par-
a small selection of the canyons, usually from one area. ticular area is far less likely to offer a thoroughly satis-
From the new information, it is evident that canyons are fying explanation.
of composite origin and that many of the hypotheses Thus, for example, Johnson's idea (Johnson, 1939,
suggested in the past were partly correct but did not ap- 1967) that the canyons were the result of underground
preciate that coordination of other processes was re- water circulation at the continental margins was based
quired. Thus there is growing evidence that, in the on his studies of East Coast canyons where the walls
history of many canyons, there was a period in which may have some soluble rocks, but paid no attention to
subaerial erosion was an important precursor, but that canyons which cut into crystalline rock. Johnson's con-
present features are predominantly the result of marine cept had many other shortcomings, such as the general
erosion. Those advocating turbidity currents as the uni- absence of sinkholes along the axes of canyons.
que cause of canyons failed to appreciate that debris Bucher's (1940) idea that tsunamis cut canyons along
flows down the incipient valleys, as well as other types the continental margins failed to show that canyons had
of landslides, could be an almost equally important fac- any relation in location to areas where tsunamis have
tor in marine erosion. The great effect of biologic activi- their chief effects, such as the South American coast of

© Copyright 1981. The American Association of Petroieum Geologists. the Geological Society of America left off, before World War II. Much
Ali rights reserved. help has come through the years from Scripps Institution of
'Manuscript received, September 15,1980; accepted, January 21, 1981. Oceanography and many other institutions, and such government agen-
'Scrlpps Institution of Oceanography, University of California, San cies as the Coast and Geodetic Survey and the National Oceanic and At-
Diego, La Jolla, California 92093. mospheric Administration have played an important part in providing
If would be impossible to acknowledge all the help which I have me the opportunity to Join their ship operations. Others are mentioned
received during the half-century of investigation of submarine canyons. in the text. I am particularly grateful to Robert Stevenson of the ONR
Accordingly, I can at least express appreciation for the many years of who encouraged me to undertake this review, and made excellent sug-
support by the Office of Naval Research and the National Science Foun- gestions relative to the manuscript. Suggestions by K. 0. Emery are
dation, which took up where earlier support from the Penrose Fund of much appreciated.

1062
 3
O

•a

FIG. 1—United States East Coast canyons showing clear independence of fault trends which, in general, should extend from northeast to southwest. Contours are in o
fathoms. w
 o

CO
c
o-
B
fi>

^^
3'
(D
O
a>
3
•<
O
3

FIG. 2—Gulf of St. Lawrence shows characteristic glacial trough topography due to ice excavation during Pleistocene.
 Francis P. Shepard 1065

Chile and Peru as well as the south-facing coast of of submarine canyons, despite most submarine canyons
Alaska. The mechanics of tsunamis are now known to showing a close relation to land rivers (Shepard and
have little effect in the deep water into which the can- Dill, 1966, appen., col. 6). Some submarine canyons,
yons extend. Beginning with Wegener (1924), submarine like those on the west side of Corsica (Fig. 3), extend
canyons and even what now are well known to be glacial right into the estuaries of land canyons and are clearly a
troughs on the continental shelves, such as those of the continuation of land canyons, even in general character.
Gulf of St. Lawrence and the Skagerrak strait, have In some places, fairly substantial evidence shows that
been explained by many as fault troughs. These the coasts have been submerged and drowned canyons
geologists seem regularly to have failed to note, formed when they stood higher than now, as clearly
however, that the canyons usually show little relation to demonstrated by the Hawaiian canyons (Fig. 4; Mober-
fault patterns (Fig. 1) and have a decided relation to ly and McCoy, 1966; Shepard and Dill, 1966, p. 211). It
land canyons or valleys inside their heads which are is now fairly well established that the Mediterranean
clearly not fault troughs. Troughs hke the St. Lawrence was largely desiccated during the Messinian of late
are all located off glaciated territory and have Miocene age (Hsu et al, 1973). It is not unreasonable to
characteristics typical of glaciated valleys (Fig. 2). believe, therefore, that subaerial erosion by rivers from
An early interpretation of submarine canyons as mountainous areas cut some ancestral canyons into
drowned river-cut canyons resulting from a former steep slopes that were laid bare at this time. Because
great emergence of the continents (Spencer, 1903) was there seems to be little doubt that turbidity currents can
developed before geologists had any appreciable keep such canyons open and even enlarge them after
knowledge of the Cenozoic history of the continental they become submerged, the possibility of late Miocene
margins and coastal area and was abandoned as this subaerial initiation of many Mediterranean canyons
history became more evident. The idea which I espoused should not be ignored.
in the early days (Shepard, 1933)—that glaciation may Many other canyons have stratigraphic relations in
have been much greater than generally believed and sea their wall rocks suggesting that the areas were at one
level lowered much more than Daly and others had im- time elevated sufficiently so that subaerial erosion could
agined—had to be abandoned also when we began to have allowed streams to cut canyons into their margins;
learn more about continental margins and about the they were later converted to submarine canyons by
canyons. submergence and reexcavation, probably by turbidity
Another hypothesis for origin of the canyons by currents and other marine processes. Thus the great
subaerial erosion was developed by the French school thickness of terrestrial deposits in the Cretaceous off the
(Bourcart, 1938). They believed that the continental northeast United States coast is well documented by oil
margins along the coast have been temporarily up- drilling and multichannel seismic profiles (Schlee et al,
warped, allowing streams to cut canyons, and then later 1976). The Alvin dives along the continental slope off
down warped so as to submerge them. This seems par- New England (Ryan et al, 1978) documented a great
ticularly applicable to the canyons along the Riviera and Cretaceous escarpment which was deeply eroded before
some supporting evidence was presented. However, the being largely covered by younger formations. Several
worldwide development of canyons off every continent stages suggest alternate erosion and deposition in the
and off coasts of entirely different histories from that of canyon area. This may help to explain the existence of
the Riviera makes the hypothesis difficult to maintain as so many deep canyons along the Atlantic margins.
a satisfactory explanation for submarine canyons in Another suggestion that parts of some canyons may
general. have undergone an early stage of subaerial erosion
The discovery of turbidity currents in the lakes form- comes from the discovery of salt domes and other in-
ed by dams was the first real breakthrough for an ex- dications of arid periods of deposition along the deep
planation of submarine canyons. Daly (1936) is credited margins of the Atlantic. Berger and Winterer (1974)
for first using this explanation which became for many gave evidence that a large basin was isolated during the
people the one and only explanation of all these Early Cretaceous in the South Atlantic and was prob-
features. When this idea goes so far as to say turbidity ably desiccated like the Mediterranean during the Messi-
currents produced all canyons during relatively short nian. This would have initiated submarine canyons
glacial episodes of low sea level in the Pleistocene, as along the margins, as in the Congo, although these are
became the general belief, some serious objections must now clearly related to marine processes (Leyden et al,
be raised. Canyons include many examples of excava- 1978).
tions thousands of feet deep into crystalUne rock, such The salt domes which were discovered in the walls of
as those of Sri Lanka (Ceylon) and the tip of Baja the Congo Canyon (Fig. 5; Shepard and Emery, 1973)
California. It seems unlikely that short glacial episodes may not be the right age to be explained by desiccation
could allow such enormous canyons to be cut into hard of this South Atlantic basin, but they do indicate that
rock, and geologists seemed to be somewhat at a loss in something in the nature of desiccation must have ocur-
trying to explain these huge underwater features by any red in the development of the Congo Canyon so that we
single cause. may have evidence of subaerial erosion in the canyon,
possibly after the early opening of the Atlantic rift.
SUBAERIAL EROSION AS CONTRIBUTING CAUSE
The importance of subaerial erosion in producing the
For years many geologists have been scornful of the shallow-water heads of many submarine canyons does
idea that subaerial erosion played any part in the origin not need to be defended because it is generally admitted
 o

(A
c
o-
3
ai
5'
(D
o
fi>
3
O
3

FIG. 4—Submarine canyons along north coast of Molokai, Hawaii, show clear con- show subaerial erosion in the walls of these canyons. Contours based on surveys by
nection to land canyons except where active young volcano has built out Kalaupaupa U.S. Navy and Scripps Institution of Oceanography's ship Spencer F. Baird.
(Kalawao) Peninsula. Recent deep diving operations have indicated what appears to
 Francis P. Shepard 1067

that sea level was lowered at least 100 m (by formation Pitman also indicated how complicated the effects of
of seawater into ice on the continents in the sea-level change may be on the continental margin.
Pleistocene). Perhaps the most surprising aspect of such With the reasonable assumption of a hinge line with
subaerial erosion is that there is almost no indication maximum transgression and regression at the outer edge
that the valleys stop at such levels, nor that there are of the margins, we might have large submergence and
deltas beyond. The Hudson shelf valley is almost unique emergence at this point, and land-cut canyons could be
in its discontinuity, and even there a great canyon ex- considerably submerged during transgressions, produc-
tends well beyond the shelf valley. ing submarine canyons. During emergence, subaerial
A somewhat greater rise of sea level, perhaps as much canyons could be cut well out on the margin and pro-
as 300 m (Pitman, 1978) may have resulted from the for- duce deltas; common development of submarine can-
mation of ridges in relation to sea-floor spreading. yons beyond the early stages of cutting may well be the
Midocean prominences, if they rose synchronously, combined result of sea-level lowering and the hinge-line
would cause extensive submergence of valleys. Thus rotation. This sequence seems to be well illustrated in
they may contribute to some extent to the formation of the marginal area off Georges Bank. Thus we return to
submarine canyons, but it does not seem likely that this a previously discarded hypothesis for some canyons, ex-
was of major significance. cept that the sea-level changes are due to ridges and

FIG. 3—Typical submarine canyons along west coast of Corsica deeply indent similar land canyons. Contours are in meters.
 Xd:^
FIG. 5—Diagrams of seismic profiles of Congo Submarine Canyon indicate tliat parts of canyon are cut into salt domes
(shaded). From Shepard and Emery (1973).
 Francis P. Shepard 1069

rates of spreading, augmented by hinge-line dicate a difference from most land fault valleys. The
movements. The complete story is still speculative, but difference could be explained by turbidity currents act-
serves to show how complicated the origin of many can- ing on what was originally a fault valley of trough-
yons may be. shaped cross section; it is in the process of becoming a
submarine canyon. Some straight submarine canyons
SUBMAMNE FAULT VALLEYS AND THEIR RELATION which are attributed to erosion may be primarily fault
TO SUBMARINE CANYONS valley in origin. Seismic profiles obtained by Gary
Greene (personal commun., 1979) suggest that La JoUa
Some valleys on the sea floor appear to be primarily Canyon follows a fault and the offset where Scripps
fault valleys of the graben type. One example includes Canyon enters La JoUa Canyon suggests that there may
the great trough or trench which trends south down the have been right-lateral movement so that Scripps Can-
center of Sagami Bay off the Tokyo area of Japan (Fig. yon may be offset and continue to the northwest where
6) and curves gradually to the east. The north-south there is a right-angled bend in La Jolla Canyon. The
valley that is a seaward continuation of the great fault straightness of the narrow Scripps Canyon (Fig. 9) is
near the southwestern end of Baja California (Fig. 7) is also suggestive of rifting action.
another. A third is on the south side of the Bering Seimsic profiles of the canyons extending from the tip
Peninsula and islands of Alaska (Fig. 8) where a series of Baja California (Normark and Curray, 1968) have
of short straight valleys extends seaward beyond faults several indications that the valleys follow faults,
on the islands. although erosion along the faults is possible.
Theoretically, fault valleys should be recognized on One tributary which enters Carmel Canyon outside
the sea floor from the presence of undrained depres- Carmel Bay appears to represent a fault valley, and part
sions, trough-shaped floors, and straight walls. There of the Monterey Canyon clearly follows a fault, placing
may also be an absence of tributary valleys on the sides. granite on one side and sedimentary rocks on the other.
Actually the characteristics are not too well developed
on the sea-floor examples, except perhaps in the Aleu- IMPORTANCE OF LANDSLIDES AND DEBRIS FLOWS
tians, so that I am forced to admit that my two examples IN CANYON DEVELOPMENT
are based largely on their being continuations of fault
valleys on land. At the head of Sagami Bay, there were Landsliding on the sea floor in the development of
extensive movements during the great 1923 earthquake. canyons is difficult to evaluate because the slides are an
The Japanese first reported extensive depth changes in important cause of turbidity currents, and the results
Sagami Bay, but later expressed the opinion that ap- may be just initiated by slides which put enough sedi-
parent changes had been the result of inaccurate ment into the water to cause major turbidity currents.
surveys. It does seem strange, however, that a large area Discoveries of considerable deepening at the heads of
became deeper and another somewhat smaller area had submarine canyons in southern California (Shepard,
all shallower soundings. One does not see how inac- 1951) through the repetition of sounding profiles at fre-
curate surveys could produce such results. The Sagami quent intervals and confirmation by observations of
Trough, in cross section, is largely a V-shaped valley not scuba divers, notably Dill (1964), leave no doubt that
much different from a submarine canyon, except at its large changes do appear and new valley heads are occa-
trough-shaped head. It has typical dendritic submarine sionally found. Perhaps the best example was the for-
canyons entering from the east and, farther out where it mation of a large tributary valley to Scripps Canyon
bends east, they enter from the north without any par- (Fig. 9) in 1975. Much earlier than these discoveries I
ticular indication of a hanging-valley relation as might had reported that new valleys were developing in the
be expected if this were a fauk valley. On the west and foreset slope off some of the main passes of the
south sides there are no tributaries, but there are indica- Mississippi delta (Shepard, 1960). That report has been
tions that volcanism is important and may have filled confirmed by Coleman (1964) and his associates while
such tributaries. studying the instability of the slopes off this great delta
The supposed fault valley off southwestern Baja (see also Garrison, 1974; Prior and Coleman, 1978). Of
California (Fig. 7) trends straight to the south in a line course, sliding is well known from the difficulties the
of continuation of the land fault valley. It is somewhat petroleum companies have had with their structures on
trough-shaped at its head, but gradually becomes these gentle, deltaic slopes.
V-shaped in its lower part. It has only two tributaries; The discovery that wherever rivers have built deltas
the one that enters on the east side seems to have a con- across the continental shelf, the foreset slope beyond is
tinuation on the west side that represents a former con- creased by valleys, some even of canyon dimension
tinuous valley that was offset by strike-slip faulting. The (Shepard et al, 1979), makes one wonder if landslides
western continuation deepens to the west to indicate may not be common in areas of forward-building sedi-
such an origin. The other submarine canyons off Baja ment slopes. Again, the effect of turbidity currents is
California are twisting and rather typical erosion valleys not easy to evaluate. The dives in the Alvin off New
contrasting markedly with the straight feature which I England showed the importance of debris flows which
am attributing largely to faulting. moved large rocks great distances down the axes of the
Even some of the Alaskan fault valleys have V-shaped canyons investigated. On the outer slope, they even
cross sections (Fig. 8) suggesting that such are found masses of rock (larger than the Alvin) which must
characteristic of marine fault valleys. This seems to in- have been moved down by these flows (Ryan et al,
 o
o

CI
'• >

c
3
8.««4* 3

o
o>
o
3

FIG. 6—Sagami Trough and sea-floor topogaphy off Tokyo area of Japan. Earth- side, and probably by volcanism on west side. Contours based on surveys taken by
quake of 1923 centered at head of Sagami Bay. Bay seems to be primarily a fault Japanese Navy and Scripps' Spencer F. Baird. Subsea contours are in fathoms.
trough, but has been considerably modified by submarine erosion, especially on east
 Francis P. Shepard 1071

^:^f

NAUT
CONTOUR

FIG. 7—Presumed submarine fault valley (Tinaja Trough) extends along southwest end of Baja California as continuation of
fault valley on land farther north (arrows). Only tributary valley which enters from east apparently continues seaward after off-
set to north suggesting strike-slip faulting. Based on surveys by writer, on Scripps Institution ships. More recent soundings of
deeper valley suggest straighter course than shown here.
PACIFIC OCEAN

FIG. 8—Presumed fault valleys south of eastern Aleutian Islands. General trough suggest faulting, and valleys are in line of continuation of faults on islands. From
shape of valleys, lack of tributaries, depression contours, and straight steep sides all Gates and Gibson (1956). Bathymetry is in fathoms, topography in feet.
 Francis P. Shepard 1073

1978). These debris flows must be considered as an im- on !he canyon walls in producing erosion was partly
portant cause of erosion. Sand flows in the Baja understood from our dives into the California and Baja
California canyon at Cape San Lucas (Shepard, 1964) California canyons, and became even more clearly
were clearly producing important erosion from the walls demonstrated by the dives in the Alvin in East Coast
of this canyon, as shown by the abundance of large and canyons, and by those of biologists at shallower depths
small rock fragments which were broken off and found in the same area. In many places the rock walls are com-
on the floor of the canyon below, on top of the ripple pletely undermined by organisms, and large cavities are
marks. produced where animals have dug into the walls, leaving
Possible indication of the importance of slumping in their debris at the base in small cones. Soft rocks are
producing tributary valleys is the abundance of these very much eroded by this activity so that a rock face
tributaries all along the Bay of Biscay found by side- may appear somewhat like the cavernous sandstone
scan operations (Belderson and Kenyon, 1976; Kenyon walls in land canyons.
et al, 1978). Many of these valleys seem to have no rela-
tion to the supply of sediments from land and could be ALTERNATING LONG PERIODS OF MARGINAL
explained as the result of general slumping on the steep BUILDUP AND CANYON-FLOOR EROSION
slopes, as suggested to me by K. O. Emery. As study of canyons of the sea floor has proceeded, it
BIOLOGIC EROSION has become more and more evident that these mon-
strous valleys are not necessarily the product of one
The importance of boring and digging animals living generation of excavation. I first suggested that the can-

flSfniM hi iiMfvcs

FIG. 9—Apparently newly opened landslide valley on southwest side of Seripps Canyon. Contours that cross others were obtain-
ed from soundings by Neil Marshall in 1975 (Marshall, 1978).
1074 Submarine Canyons

W L : . -.• ' '^

LAND AREAS • . • ; . * ;

0-50 FATHOMS

50 — 500 '.•.••-ililT •

500 — 1000
1000-1500
-• in iJ* NAUTtCAL
ID

MILES
20

1500-2000
/A I.
>2000

FIG. 11—Great Bahama Canyon: Tongue of the Ocean on south, and Northeast and Northwest Providence Channels. Samples
and borings in canyons show relatively deep-water formations, whereas borings into Bahama Banks indicate shallow-water
origin suggesting that canyons existed as deep water for long period whereas banks grew up as reefs as entire mass sank. From
contours by Shepard in Andrews et al (1970).
 Francis P. Shepard 1075

FIG. 10—Seismic profile (left) and interpretation sketch (right) of Banyuls Canyon off southwestern France indicating earlier
deeper canyon, later filled, then partly reexcavated (Dangeard et al, 1968).

yons off New England might have been excavated dur- canyons which rely largely on a single factor, such as
ing some time of emergence and then filled by deposi- erosion by turbidity currents or tectonic downfaulting,
tion, subsequently to be reopened during a second to explain the deeps or the building up of the banks on
period, so that the walls had greater height than would the margins. They favored the idea, however, which we
have been possible during just one period of erosion expressed (Andrews et al, 1970)—a combination of ero-
(Shepard, 1936). Although this idea lay dormant for sion in the deep areas and upbuilding on the shallow
several decades, it seems to have gained new recognition banks. As they showed (Fig. 11), some 3,000 m of
as a result of studies of the Mediterranean canyons, deposition of shallow-water calcareous sediment has ac-
notably off southwestern France (Dangeard et al, 1968). cumulated on the banks as noted from borings by oil
Seismic profiles there have shown that in the Banyuls companies. The deep-water formations were first
Canyon a fill in the canyon floor has been partly reex- discovered by Gibson and Schlee (1967) from samples
cavated to form a new canyon, not so deep as the old obtained by the Alvin. The importance of debris flows
one (Fig. 10). Other seismic profiles suggest the same was also shown by the discovery of a block of shallow-
process in various parts of the world. water reef material up to several feet in diameter found
Probably the most complete evidence of this com- in the canyon and evidently transported by the debris
bination of deposition on the canyon margins with alter- flows along the canyon floor. Photographs in the can-
nating periods of erosion and fill within the canyons has yons at or near the floor also show ripple marks and
come from the East Coast operations of the Alvin. Us- rounded cobbles, suggesting strong current activity
ing extensive notes made by the late Bruce Heezen on his (Rusnak and Nesteroff, 1964; Andrews et al, 1970).
dives into the New England canyons in the Navy The submarine canyons in the Bahamas (Fig. 11) may
minisubmarine, the group at Lament Geological Obser- include the deepest canyon in the world (Shepard and
vatory, along with others, made dives into the same can- Dill, 1966, p. 194). Thus it seems possible that many of
yons (Ryan et al, 1978) where at least three periods of the great canyons may be features formed over long
erosion, dating back as far as the Cretaceous, have been periods and kept open, or reopened from time to time,
recognized. The remarkable interfingering of the by turbidity currents and debris flows.
Eocene into the Cretaceous platform provides an abun-
dance of evidence of this ancient erosion. A Cretaceous TURBIDITY CURRENTS
escarpment, considerably eroded but later buried by
younger material, gives additional evidence. Thus there Having at one time been considered as one who did
is evidence that the New England canyons were initiated not beUeve in turbidity currents, and admittedly one
shortly after the Atlantic rift developed, separating the who did not give them their due in earlier publications, I
Americas from Europe and Africa, and it appears that would like to close this discussion with an acknowledge-
the canyons have persisted since then. Another impor- ment that there is now a considerable body of evidence
tant example comes from the Bahama Islands (Fig. 11). suggesting that these currents are amazing in their
This was summarized by Hooke and Schlager (1980) in a capacity to erode the sea floor. Perhaps the most con-
study of the observations obtained from the Alvin in the vincing evidence is that deep valleys can be traced across
Tongue of the Ocean (Toto Canyon) and of cores from thousands of miles of deep-sea floor and actually have
drilling in the other branch of the Great Bahama Can- coarse sediments, even gravel beds, at unimaginable
yon in Northwest Providence Channel. Hooke and distances from any continental source. These valleys are
Schlager referred to the various explanations for the found in every ocean. Some of them even cut through
1076 Submarine Canyons

ocean ridges off the East Coast of the United States. and upgrowth in the intervalley areas, while the canyons
Our descents in deep-diving vehicles seem to indicate maintained their relative deepness.
that these powerful currents occur much more frequent- 7. The significance of plate tectonics in the early
ly than every few hundred years, as had previously been development of many of the Atlantic canyons may be
suggested; canyons are altered radically in less than a considerable, although it is perhaps a little too soon to
year—the time between deep dives. We observed a pile evaluate this with any certainty. We can at least
of boulders in Scripps Canyon with a steep descent speculate on the possible origin of some canyons when
below on one side, and some months later they had early bodies of water formed by separation of the
disappeared. Surely a powerful flow had intervened. various parts of Pangea became isolated, and desicca-
The auto body placed by Dill shortly prior to 1964 in the tion caused exposure of the steep sides of the new con-
same canyon, and anchored to the walls by heavy tinental margins, allowing the cutting of canyons. We
weights, could not be found a few years later when can also speculate on the importance of eustatic sea-
repeated dives were made down this canyon. Slumping level changes, augmented by marginal warping in rela-
could have been the cause, but it seems more probable tion to the newly developed mid-ocean ridges, and the
by all other evidence that it was carried away by turbidi- adjustment of the margins to weighting of sediment
ty currents and debris flows. from the continents. This could have been very impor-
Of course, we should not overlook the possibility of tant in starting some canyons by subaerial erosion dur-
cable breaks which may be due to slumping, but surely ing the stages of regression which must have occurred.
currents which can transport coarse sediment hundreds Thus it seems unwise for authors to continue to offer
or even thousands of miles along deep-sea channels simple, unified explanation for these huge valleys of the
must be considered as a likely cause of the breaks. sea floor.
Cables placed in the Congo Canyon at a variety of
distances from the canyon head promptly broke during
REFERENCES CITED
times of high water in the river—a seemingly confirmed
case of strong currents. Andrews, J. E., F. P. Shepard, and R, J. Hurley, 1970, Great Bahama
That turbidity currents of relatively small velocity can Canyon: Geol. Soc. America Bull. v. 81, p. 1061-1078.
be frequent seems to be established through our ex- Belderson, R. H., and N. H. Kenyon, 1976, Long-range sonar views
perience of putting current meters down in valleys off of submarine canyons: Marine Geology, v. 22, p. M69-M74.
Berger, W. H., and E. L. Winterer, 1974, Plate stratigraphy and the
submarine deltas for a few days and encountering tur- fluctuative carbonate Une: Internat. Assoc. Sedimentol. Spec. Pub.
bidity currents in several places (Shepard et al, 1979, p. 1, p. 11-48.
29). Also, we encountered such currents each time we Bourcart, J., 1938, Essai sur les regressions et transgressions marines:
put out a current meter during a period when an on- Soc. Geol. France Bull., v. 5, p. 393-474.
shore storm occurred at La JoUa, California. Bucher, W. H., 1940, Submarine valleys and related geologic prob-
lems of the North Atlantic: Geol. Soc. America Bull., v. 51, p.
489-512.
CONCLUSIONS Coleman, J. M., 1964, Minor sedimentary structures in a prograding
distributary; Marine Geology, v. 1, p. 240-258.
1. Submarine canyons are no longer a complete Daly, R. A., 1936, Origin of submarine canyons: Am. Jour. Sci., 5th
ser., V. 31, p. 401-420.
mystery, but they still have puzzling aspects. Dangeard, L., et al, 1968, Resultats de la plongee en Soucoupe no. 421
2. Probably the most important factor in the dans la vallee sous-marine de Planier, au large de Marseille: Inst.
development of the canyons as they now exist is erosion Oceanog., Monaco, Bull., v. 67, no. 1384, 21 p.
by turbidity currents, but this is by no means the entire Dill, R. F., 1964, Contemporary submarine erosion in Scripps Sub-
marine Canyon: PhD thesis, Univ. California, Scripps Inst.
explanation. Oceanog., 269 p.
3. There is growing evidence that many canyons were Emery, K. O., 1972, Eastern Atlantic continental margin: some results
initiated by subaerial erosion and, after later of the 1972 cruise of the R.V. Atlantis II: Science, v. 178, p.
submergence, have been preserved and greatly enlarged 298-301.
by marine processes such as turbidity currents. Garrison, L. E., 1974, The instability of surface sediments on parts of
the Mississippi delta front: U.S. Geol. Survey Open File Rept.,
4. Many submarine canyons may have started as sub- Corpus Christi, Texas, 18 p.
marine fault valleys which later were greatly modified Gates, O., and W. Gibson, 1956, Interpretation of the configuration
by marine erosion. of the Aleutian Ridge: Geol. Soc. America Bull., v. 67, p. 127-146.
5. Submarine landslides and debris flows, as well as Gibson, T. G., and J. Schlee, 1967, Sediments and fossiliferous rocks
various types of submarine slumping, have played an from the eastern side of the Tongue of the Ocean, Bahamas: Deep-
Sea Research, v. 14, p. 691-702.
important part in the development of submarine can- Hooke, R. LeB.,and W. Schlager, 1980, Geomorphic evolution of the
yons, particularly in the formation of tributaries which Tongue of the Ocean and the Providence Channels, Bahamas:
have proven to be far more numerous than was suppos- Marine Geology, v. 35, p. 343.
ed prior to the advent of side scanning. Hsu, K. J., W. B. F. Ryan, and B. B. Cita, 1973, Late Miocene desic-
6. Many large submarine canyons, particularly those cation of the Mediterranean: Nature, v. 242, p. 240-244.
Johnson, D. W., 1939, The origin of submarine canyons, a critical
cut into crystalline rocks, have probably been the result review of hypotheses: New York, Columbia Univ. Press, 126 p.
of several stages of erosion, sometimes with intervening 1967, The New England-Acadian shoreline: New York,
periods of deposition. This accounts for the great height Hafner Pub. Co., 608 p.; reprint, originally published 1925.
of some canyon walls. The canyons off the New Kenyon, N, H., R. H. Belderson, and A. H. Stride, 1978,
Channels, canyons and slump folds on the continental slope be-
England Coast and those off the Bahamas seem clearly tween south-west Ireland and Spain: Oceanologica Acta, v. 1, p.
to have had this origin during a long period of sinking 369-380.
 Francis P. Shepard 1077

Leyden, R., et al, 1978, Salt diapirs on Sao Paulo plateau, M-39,
southeastern Brazilian continental margin:AAPG Bull., v. 62, p. 1936, The underlying causes of submarine canyons: Natl.
657-666. Acad. Sci.-Natl. Research Council Pub., v. 22, p. 496-502.
Marshall, N. F., 1978, Large storm-induced sediment slump reopens 1951, Mass movements in submarine canyon heads: Am.
an unknown Scripps Submarine Canyon tributary, in D. J. Stanley Geophys. Union Trans., v. 32, p. 405-418.
and G. Kelling, eds.. Sedimentation in submarine canyons, fans, 1955, Delta-front valleys bordering the Mississippi
and trenches: Stroudsburg, Pa., Dowden, Hutchinson & Ross, p. distributaries: Geol. Soc. America Bull., v. 66, p. 1489-1498.
73-84. 1960, Mississippi delta—marginal environments, sediments,
Moberly, R., Jr., and F. W. McCoy, Jr., 1966, The sea floor north of and growth, in F. P. Shepard et al, eds., Recent sediments, nor-
the eastern Hawaiian Islands: Marine Geology, v. 4, p. 21-48. thwest Gulf of Mexico, 1951-1958: AAPG, p. 56-81.
Normark, W. R., and J. R. Curray, 1968, Geology and structure of 1964, Sea-floor valleys of Gulf of California, in Marine
the tip of Baja California, Mexico: Geol. Soc. America Bull., v, 79, geology of Gulf of California: AAPG Mem. 3, p. 157-192.
p. 1589-1600. and R. F. Dill, 1966, Submarine canyons and other sea valleys:
Pitman, W. C , Jr., 1978, Relationship between eustacy and Chicago, Rand McNally & Co., 381 p.
stratigraphic sequences of passive margins: Geol. Soc. America and K. O. Emery, 1973, Congo Submarine Canyon and Fan
Bull., V. 89, p. 1389-1403. Valley: AAPG Bull., v. 57, p. 1679-1691.
Prior, D. B., and J. M. Coleman, 1978, Submarine landslides on the et al, 1979, Currents in submarine canyons and other
Mississippi River delta-front slope, in Man and environment in the seavalleys: AAPG Studies in Geology, no. 8. 173 p.
lower Mississippi Valley: Geosci. Man, no. 19, p. 41-53. Spencer, J. W., 1903, Submarine valleys off the American coasts and
Rusnak, G. A., and W. D. Nesteroff, 1964, Modern turbidites—ter- in the North Atlantic: Geol. Soc. America Bull., v. 14, p. 207-226.
rigenous abyssal plain versus bioclastic basin, in R. L. Miller, ed.,
Papers in marine geology—Shepard commemorative volume: New Vail, P. R., R. M. Mitchum, Jr., and S. Thompson, III, 1977, Seismic
York, Macmillan Co., p. 488-507. stratigraphy and global changes of sea level, pt. 4, Global cycles of
Ryan, W. B. F., et al, 1978, Bedrock geology in New England sub- relative changes of sea level, in Seismic stratigraphy; applications
marine canyons: Oceanologica Acta, v. 1, p. 233-254. to hydrocarbon exploration: AAPG Mem. 26, p. 83-97.
Schlee, J., et al, 1976, Regional geological framework off north- Veatch, A. C , and P. A. Smith, 1939, Atlantic submarine valleys of
eastern United States: AAPG Bull., v. 60, p. 926-951. the United States and Congo Submarine Valley: Geol. Soc.
Shepard, F. P., 1932, Landslide modifications of sub marine valleys; America Spec. Paper no. 7, 101 p.
Am. Geophys. Union Trans., 13th Ann. Mtg., p. 226-230. Wegener, A., 1924, The origin of continents and oceans: New York,
1933, Canyons beneath the seas: Scientific Monthly, v. 37, p. Dutton, 212 p.