You are on page 1of 27

Downloaded from http://jgs.lyellcollection.

org/ at North Carolina State University on December 10, 2012

Journal of the Geological Society

The Karroo Volcanic Cycle


KEITH GORDON COX

Journal of the Geological Society 1972, v.128; p311-336.


doi: 10.1144/gsjgs.128.4.0311

Email alerting click here to receive free e-mail alerts when new
service articles cite this article
Permission click here to seek permission to re-use all or part of
request this article

Subscribe click here to subscribe to Journal of the Geological


Society or the Lyell Collection

Notes

© The Geological Society of London 2012


Downloaded from http://jgs.lyellcollection.org/ at North Carolina State University on December 10, 2012

The Karroo Volcanic Cycle


KEITH GORDON COX

CONTENTS
1 Introduction . . • 312
2 The tectonic model . • • 312
3 Geochemical provinces and the tectonic model 3z4
4 General aspects of variation in the Mesozoic province 3z4
(A) M g O content . . . . 315
(B) Saturation with respect to silica . 3x6
(c) K~O/Na~O ratio . . 3z6
5 Absolute content of K~O and the concept of mantle processing 3x6
6 The distinction between rocks of mantle and crustal origins 319
7 Variation of M g O with time in the mantle-derived rocks 320
(A) The culmination . 320
(B) The 'steady state' period . . . 322
(c) Tectonic setting and the 'steady state' 323
(D) Time relations of the cycle. . . 323
(E) The waning stage of the cycle . . . 323
(F) The termination of the cycle and the problem of kimberlites . 324
(o) Applicability to other provinces . . 324
8 T h e cycle in rocks of crustal derivation 325
9 K , O ] N a ~ O ratios and the diversity of parental magmas• 325
(A) Relations between the parental magmas 326
I o Silica saturation and the problem of depths of magma generation 327
11 Conclusions . . . . 329
12 References . . . . . . . . . 331

SUM MARY
T h e petrogenesis of all the Mesozoic igneous emitted remained relatively uniform and
rocks of south-eastern Africa is considered in moderately fractionated over a long period of
relation to a single unifying thermotectonic time is referred to as the 'steady state' phase.
event postulated as initiating the disruption A waning stage m a y be represented by more
of Gondwanaland. The cycle began with the salic rocks such as the phonolites of Lupata
rise of a body of potassium-rich picrific magma (Mozambique) and the associated plutons of
from a depth of at least 50o km. This is the the Chilwa province (Malawi). The Cretaceous
predominating source material for most of the kimberlites of Southern Africa m a y represent
rocks in the northern part of the province the terminal igneous activity of the cycle.
(e.g. Rhodesia). Overlying and peripheral to Superimposed on the cycle of mantle-derived
the main magma body a zone of normal (sodic) rocks is a minor cycle of anatexis in the crust
magma was generated and gave rise to some of which gave rise to most of the acid rocks of the
the northern rocks and most of the southern province, particularly the rhyolites of the
rocks (e.g. Karroo do!erites, basalts of Lesotho Lebombo monocline. I t is suggested that other
a n d Swaziland). Initial eruptive products were provinces such as the Deccan Traps and the
locally well fractionated but gave way rapidly Tertiary of Greenland may show patterns of
to highly primitive picritic lavas representing igneous activity which are similar to the
what is termed the culminatory stage of the Karroo cycle in important respects.
cycle. A subsequent stage in which the basalts

Jl geol. Soc. Lond. vol. x28~ x972 , pp. 311-336 , 6 figs. Printed in Northern Ireland.
1
Downloaded from http://jgs.lyellcollection.org/ at North Carolina State University on December 10, 2012

312 K. G. Cox

I. I n t r o d u c t i o n
THE KARROO SYSTEM of Southern Africa consists of a series of largely-
continental sedimentary rocks, ranging in age from Carboniferous to late
Triassic, overlain by an extensive series of basalts mainly of late Triassic or
early Jurassic age. The basalts were accompanied by the intrusion of a great
swarm of dolerite sills and dykes, described in classic papers by du Toit (I92o)
and Walker & Poldervaart (i 949). The igneous activity was on a very large scale
and there remain today about I4o ooo km 2 of lava outcrop, though formerly the
area of southern Africa covered by the lavas or penetrated by minor intrusions
probably exceeded this by a factor of at least i o.
The term 'Karroo Volcanic Cycle' was used by Cox, Monkman, Johnson,
Stillman, Vail & Wood (I965) to describe a concept in which not only the rocks
of Karroo age in the strict sense but all the Mesozoic igneous rocks of southern
Africa were related to a single broad thermo-tectonic event. The cycle appears to
be connected with the establishment of the continental margin of south-eastern
Africa during the fragmentation of Gondwanaland (Cox I97O ). Igneous rocks of
the cycle range in age from late Triassic (c. 200 m.y.) to early Cretaceous (c. x IO
m.y.). The waning stages of this cycle overlap in time with the main period of
activity in South West Africa (Siedner & Miller i968 , Manton & Siedner I967)
which is dated at I i 4 - i 3 6 m.y. and is presumed to be associated with the de-
velopment of the South Atlantic continental margin.
The Mesozoic igneous province of south-eastern Africa is extensive (see Fig. I)
and contains a great variety of rock types. The reader is therefore referred to a
number of review articles by du Toit (I954), Haughton (I963) and Cox (in press)
which give general geological details, Manton (i 968) for age determinations and
other isotopic work, Fitch and Miller (in press) for age determinations, Cox,
Macdonald and Hornung (I967) and Woolley and Garson (x97o) for chemical
features, and Vail (I967) , Flores (I97o) and Cox (I97O) for tectonics. Papers
dealing with the broader aspects of petrogenesis have been few, and include
those by du Toit (i929) , Walker and Poldervaart (I949) as well as Cox et al.
(i965) and Woolley and Garson (op. tit.).

2. The tectonic model


The tectonic model which will be assumed as part of the basis of the following
discussion is derived from a review by Cox (I97O).
It was postulated that a convective uprise of mantle material, approximately
equidimensional in plan, took place beneath the central part of the Gondwanaland
continent and, by a component of outward flow at relatively high levels, was
responsible for the observed surface volcanic and tectonic effects. Central to the
hypothesis is the idea that the surface effects were localised by concentrations of
strain along existing lines of crustal weakness. Therefore the study of surface effects
does not allow a precise reconstruction of the pattern of flow beneath the litho-
sphere. Although the hypothesis is a typical example of plate tectonics it differs
from the earlier sea floor spreading models of Hess and Dietz in that it does not
Downloaded from http://jgs.lyellcollection.org/ at North Carolina State University on December 10, 2012

The Karroo Volcanic Cycle 3 x3

1 , I
E~ + + + + + + +30OE + + + + + + 434°E +
+ + + + + + + + + + 4- + + A _
+ + + ++ ~ +: ~ + + + + + + + + +
+ 4- + + ,
~ _++4.++
+ + + +
+ 4- +
+ + + 4-
+ ::" .. +
+ + + + + 4- + + + + + + Lupata,
+ +
i: + 4- +
• -'" • + + + + + + + + + + -I
::: + + + + + + + + + +
+ + + + + + + + + +

..:~ + + + + + +:~
+ ~1~ + ÷ 4.
~ ~
~+ +
++ + \ +
+ + ++ ÷+ + + ~
:~':: ~+ + + ¢0 + + + +
~~'i= + + + A" + + + +.~'-,~^:.
:ira
~ ~ ' + ÷ + + , X ) + + + S a b i + ~ : : : r ~ ' 4 v . ~ "~''
~,~+~Bulawayo . x k . .s + +. ~ ' ~
+ + + + + --X"
4. + + + +%~- + + + + ^. ,o ~ ~i i ~' ~. ', ~ r r ~~
ii=...'~ + + ~.xo\"~ "- - , ~ . ' ~
;::::.~ + + ~ s . ~ ~ V ~ l ~ l l ~ - - .~ _______q
4.

+ + + + + ~ + +
+ + + + + + + ÷ +,,.¢waz;.
~ 4. + + + + ~ 0 100 200 300 4o0
+ ~ ~ + +5O+U T +H ++ + ~ i i ~ '
+ + ÷ 4- + +
4.
+
-lan~
+
°
Kilometres
. . . . ~'~';';~',';'.'~+ .+ /_._'4

~Post Karroo
~] Plutonic'~ Karroo
Centres~ Volcanic
~ Lavas J Cycle

~ Sediments

~Pre Karroo
30OE

F xo. x. Locality m a p for south-eastern Africa s h o w i n g distribution of K a r r o o


sediments, a n d lavas a n d m a i n p l u t o n i c centres associated w i t h t h e K a r r o o
volcanic cycle.
Downloaded from http://jgs.lyellcollection.org/ at North Carolina State University on December 10, 2012

314 K. G. Cox

require convective systems to be spatially related to surface tectonic features.


The evidence on which it is based is not derived from seismic, palaeomagnetic
or crustal geometrical studies (of. Oliver & Isacks 1967; Le Pichon 1968; Morgan
I968 ) but on the conclusion that southern Africa was intersected by numerous
linear zones, forming a polygonal pattern, across all of which simultaneous
crustal extension took place during the Mesozoic.

3. Geochemical provinces and the tectonic model


The Karroo basic rocks (basalts and dolerites) were divided into two provinces
by Cox, Macdonald & Hornung (i967) , one in the northern part of the area
(Rhodesia, Northern Transvaal) characterised by unusually potassie tholeiites,
and the other to the south (Swaziland, Lesotho, Karroo dolerites of South Africa)
of normal tholeiites. It was not possible to define the limits of the provinces
precisely because of lack of data and the authors did not discuss the petrogenetic
implications of the existence of the two provinces.
It had been anticipated when the sampling programme was carried out, that a
relationship might be found between geochemistry and tectonic environments,
for example, in the contrast between the rocks of the highly disturbed zones such
as the Limpopo and the Lebombo, and the undeformed platform areas of Lesotho
and western Rhodesia. However, no such simple relationship emerged. In fact
the Lebombo zone was found to change character along its length, with normal
basalts in the south (Swaziland) and highly potassic basalts in the north. Con-
versely, the Rhodesian basalts all appear to be more or less potassic, whether
they occur in the deformed zone along the south and east (Tuli, Nuanetsi) or in
what may be a platform area in the west and north-west (Bulawayo, Livingstone).
Thus, on the present evidence it seems that the geochemical provinces delimited
on the basis of potassium contents are not related to the obvious tectonic divisions
of the region.
In the light of the tectonic model it now seems that it may be the very lack of
correlation which is the significant point. For reasons to be discussed in a later
section the potassic province may coincide with the postulated area of sub-
lithospheric up-welling, which, it is argued, is spatially unrelated to the disposition
of surface tectonic units.

4. General aspects of variation in the Mesozoic province


Details of variation in the province are given in the review papers already referred
to. Briefly the rocks range in age from Triassic-Lower Jurassic (the main
period of Karroo volcanism) to Cretaceous. The Karroo rocks are mainly tholeiitic
basalts and dolerites, with rhyolites in the Lebombo zone and early alkalic rocks
in south-eastern Rhodesia and north-eastern Transvaal. Later alkalic rocks
(largely Cretaceous) are found principally at Lupata (Mozambique) and in
southern Malawi (Chilwa series). The most important specific lava types within
the province are olivine-poor or olivine-free tholeiitic basalts, olivine-rich
(picritic) tholeiites (locally termed limburgites), olivine-poor alkali basalts,
Downloaded from http://jgs.lyellcollection.org/ at North Carolina State University on December 10, 2012

The Karroo Volcanic Cycle 315

phonolites, and nephelinites. Bearing in mind what has previously been said
about the potassium-rich and potassium-poor provinces it is thus possible to
select three important and largely independent parameters to describe specific
groups of rocks in a way which will be useful for purposes of genetic speculation.
These are the content of MgO, the degree of saturation with respect to silica,
and the ratio K 2 0 / N a , O . The relative independence of these parameters is
shown in Table i where a considerable number of the possible permutations are
represented by rock types present in the province. Discussion of the parameters is
given below.
(A) M g O CONTENT
Variations in MgO are extremely high, ranging from over 20 % in the picrites of
the Nuanetsi area to almost nil in salic rocks such as rhyolite, granite, syenite
and phonolite. For extrusive rocks, which are the main material of the present

TABLE I : The three main variables

High MgO Moderate MgO Low MgO


*

olivine monzonite
(Nuanetsi)
shoshonites
"0
(Tuli)
K,O/Na20 D
HIGH
basalts
limburgites, olivine- (Nuanetsi and
basalts, picrites other Rhodesian
(Nuanetsi) localities)
absarokites (Tuli) olivine monzonite
(Nuanetsi)

alkali basalts nepheline


(Lebombo) syenites
nephelinites (various localities)
~9 (Nuanetsi-N. phonolites
"ZJ
Lebombo) (Lupata)
K~O/Na~O
NORMAL quartz
syenites
limburgites, olivine- basalts (widespread) (various localities)
basalts, picrites dolerites (S. Africa) rhyolitcs and
~ (Nuanetsi-N. gabbros granites
Lebombo) (Nuanetsi) (Nuanctsi-Lcbombo,
©
Lupata)

* Undersaturated = with normative nepheline


Oversaturated = with normative hypersthene or quartz
Downloaded from http://jgs.lyellcollection.org/ at North Carolina State University on December 10, 2012

316 K. G. Cox

study, the value of this parameter can be taken as a rough indication of temperature
at the time of eruption. This is justified in cases where the volcanic rocks are not
notably enriched in cumulus ferromagnesian phases. This appears to be a con-
dition met by most of the Karroo lavas, for although the more picritic basalts
are rich in olivine phenocrysts there is abundant petrographic and chemical
evidence (e.g. Cox et al. I965; Jamieson I966) that this is very largely of quench
origin. A study of the variation of temperature of erupted products with time,
forms an important part of the discussion of the volcanic cycle which follows.

(B) SATURATION WITH RESPECT TO SILICA


Study of this variable, again particularly with respect to time, is also an important
aspect of the problem. Used in conjunction with the data of experimental
petrology it is possible to place some limits on the depths '(pressures) at which
various magmas developed during the cycle.

(c) K , O / N a . O RATIO
Variation in the value of KzO/Na, O imposes considerable constraints on the
postulation of liquid lines of descent, for the ratio must remain substantially
constant in any series of liquids produced by the closed-system fractionation of
minerals such as olivine, alkali-poor pyroxenes, magnetite and ilmenite. Where
there is evidence that the only fractionation affecting a particular group of rocks
is of this type, it follows that any accompanying variation in K.O/Na20 is likely
to be a reflection of variation in the immediately parental material, whether
magmatic or solid, and the rocks of the group do not therefore belong to a single
liquid line of descent. In the present study, considerations of this sort have lead to
the postulation of the existence of a large number of different parental magmas,
and this forms one of the principal features of the petrogenetic model presented.

5. Absolute content of K20 and the concept of mantle processing


One of the most important assumptions of the present work is that the high content
of potassium and other incompatible elements in the northern province Karroo
basalts is a geochemical feature originating in the mantle rather than as a result
of crustal contamination. Jamieson & Clarke (197o , pp. 196-199 ) have discussed
the general problem of the levels of these elements in tholeiites and have concluded
that crustal contamination is of little importance. For the specific case in question
the most notable feature of the contrast between the two Karroo basalt provinces
is that differences are almost entirely confined to potassium and related elements
and are not reflected to any great extent in the major elements. As will be seen
in Table 2 typical olivine-poor northern province rocks are enriched in potassium
and other trace elements such as strontium and barium by a factor of about 2
compared with similar rocks from the southern province. The average analysis
of the olivine-rich basalts from Nuanetsi illustrates a more extreme case. Clearly
wholesale contamination by granitic material cannot account for the difference
between these groups.
Downloaded from http://jgs.lyellcollection.org/ at North Carolina State University on December 10, 2012

The Karroo Volcanic Cycle 317

Alternatively, several plausible hypotheses exist to account for wide variations


in the contents of potassium and related elements in magmas produced entirely
within the mantle.
These include zone refining (Harris I957) , which may be regarded as a special
type of wall-rock reaction, a general process favoured by Green & Ringwood (i 967).
These processes treat the generation of m a g m a as an open system and contrast
with closed system hypotheses such as those involving eclogitefractionation ( O ' H a r a
& Yoder 1967) and the degree of partial melting (Gast 1968; O ' H a r a 1968 ). A
critical review of these hypotheses is given by Jamieson & Clarke (op. cir.) but
at the present time it does seem extremely difficult to devise criteria which will
distinguish adequately between them. T h e argument which follows depends on
the assumption that variation in the incompatible elements in the erupted lavas is
inherited from processes affecting the liquids in the mantle rather than in the
crust. It does not however depend on any specific mantle process, and thus the
problem of distinguishing criteria is avoided.
I n a region of widespread flood vulcanism it is plausible to assume that surface
lava sequences have been derived from the mantle directly underlying them.
I f a composition for the mantle is also assumed, and an enrichment factor for
the incompatible elements in a lava sequence of known thickness is determined,
relative to the concentration of these elements in the mantle, it becomes possible

TABLE 2" Average analyses of basalts


Wt. % A B C

SiO~ 51.8 52"I 49"7


TiO9 I .I 2"6 2"7
AI~.O3 14-8 14. i 8.9
FegOs 3"9 3"7 2.0
FeO 7"3 8"4 lO'5
MnO 0.2 0.2 0.2
MgO 7"I 5 "5 15"3
CaO Io.6 8"7 7"7
Na~O 2"4 2.6 I-6
K~O 0"7 I-7 1.6
P~Os o-i 0"4 0"4

p.p.m.
Ba 256 635 795
Sr I9o 614 859
Zr 85 192 3oo

A. Average of 21 Lesotho basalts (Cox ¢t al. 1967, Table 3, column


F).
B. Average of x4 Nuanetsi basalts with MgO below 8%. (Cox et al.
1967, Table 3 columns Dx and Dg: 5 analyses from Jamieson,
I969).
C. Average ot"47 Nuanetsi olivine-rlch basalts (Jamieson i969).
Downloaded from http://jgs.lyellcollection.org/ at North Carolina State University on December 10, 2012

3x8 K. G. Cox

to calculate the minimum vertical thickness of mantle which has to be processed


to provide the incompatible elements in the lava sequence.
The content of incompatible elements assigned to the mantle model constitutes
an uncertainty which is difficult to assess. For purposes of the present calculation
the author adopts a value of o.oI 5 % for K20. This is consistent with estimates
made by Gurney and Berg (in press) for mica-free peridotite inclusions in South
African kimberlites and also consistent with the average content of 80 p.p.m.
K estimated by Hurley (I968) for the core and mantle together. Jamieson
(I969, unpublished) recalculated Hurley's figure on the assumption that the
core was free of potassium and adopted a value ofo.oI 4 % for K20 in the mantle.
Estimates at these levels are low compared with most of those based on the analyses
of peridotite inclusions in kimberlite (e.g. the average of Ito & Kennedy I967)
and much lower than the K,O content of 'pyrolite' (R.ingwood I966). Uncer-
tainty about the potassium content of the mantle is inevitable and the following
calculation therefore seeks only to explore some of the petrogenetic consequences
of accepting a low value. Apart from difficulties concerning the potassium content
of the source rocks a further uncertainty is provided by the unknown amount of
the stratigraphic sequences which have been lost by erosion.
With these limitations in mind the calculation can be carried out for two areas,
where sufficient information about chemical composition and thickness is available,
that is Nuanetsi (thickness of basalts c. 5 km, average K,O content 1.5 per cent)
and Lesotho (thickness I. 5 km, average K~O content 0.74 per cent). With
enrichment factors for K,O of Ioo and 5° respectively this gives answers of
500 km and 75 km for the two areas.
The figure of 500 km for the Nuanetsi area is of course critically dependent on
the value of K~O assigned to the mantle model. Since however other uncertainties
attached to the calculation tend to minimise the answer there seems no great
objection to accepting that the Nuanetsi magmas may ultimately have been
derived from very considerable depths within the mantle. An alternative which
assumes that the mantle in this area was unnaturally rich in potassium before
the volcanic cycle occurred can not of course be excluded but in a sense begs
the question.
It is because of this result that the author tends to think of the potassic province,
as exemplified by the Nuanetsi-N. Lebombo area as representing the site of the
mantle upwelling postulated in the tectonic model.
To summarise, the petrogenetic model presented so far involves the postulation
of a zone of upweUing mantle material, partly magmatic, derived from a depth
of at least 500 km. From arguments presented elsewhere (Cox I97O, p. 214) it
is envisaged that movement of this material started at least 80 m.y. before the
first manifestations of surface volcanic activity. The depth of origin and the long
time involved ensured optimum conditions for the extraction of incompatible
elements so that the liquid phase was considerably enriched in potassium relative
to most basalts. The large scale of the model, both in terms of time and space,
is not inconsistent with a phenomenon postulated as responsible for the initiation
of the disruption of Gondwanaland.
Downloaded from http://jgs.lyellcollection.org/ at North Carolina State University on December 10, 2012

The Karroo Volcanic Cycle 3 I9

6. The distinction between rocks of mantle and crustal origins


The discussion up to this point has been mainly concerned with basic rocks,
which are assumed to be derived from the mantle. However, locally within the
province there are large volumes of rhyolite, rhyodacite, granite and granophyre.
These are mainly confined to the Lebombo Monocline and the Nuanetsi and
Lupata area, and are separated by a large compositional gap from the associated
basic rocks. A histogram showing silica values for analysed rocks of the Nuanetsi
and Lebombo area is given in Fig. 2. The number of different workers involved in
collecting the analysed samples is some guarantee against sampling bias.

FIG. o

30-
Histogram showing distribution of silica
i
by I per cent intervals in 2o 5 analysed
0
c
lavas and intrusive rocks from the Lebombo
,,_ 20
monocline and the Nuanetsi-Sabi area,
0 Rhodesia. Analyses are from Assunc,ao
e~ et al. (I962), Cox et al. (1965 , I967),
E Henderson (I 9o9), Hunter and Urie (i 958),
Z 1o, Jamieson (I969), Lightfoot (1938), Lom-
baard (1952), Prior (191o), Rogers (1925),
Swift et al. (I953), van Eeden et al. (I955),
Walker and Poldervaart (1949), and Young
! D
40 50 6O 7O
(192o).
% Si02
Faced with a substantial 'silica-gap', affecting the volcanics over such a wide
area the present writer suggests that the most reasonable postulate is that there
are two distinct source materials involved, one in the mantle, the other in the
crust. Manton (I 968) has argued from strontium isotope studies that the rhyolitic
rocks of the southern Lebombo (Swaziland and Zululand) are derived from the
mantle because of their relative isotopic homogeneity and low initial ratios
(0.7042 + o.0005). The very variable initial ratios of Nuanetsi basic rocks he
attributes to crustal contamination of mantle derived material. Neither of these
conclusions is at present acceptable to the writer. Crustal contamination may have
affected the strontium of the Nuanetsi basic rocks, but it can not be responsible for
their other geochemical peculiarities (see discussion by Jamieson & Clarke 1970).
However, there is an interesting conflict here between the isotopic and other
lines of evidence. It is to be hoped that further isotopic studies will be carried
out so that this very complex province becomes adequately sampled.
For the moment the writer assumes that the majority of the acid rocks are de-
rived from the crust. There are however several occurrences of salic rock types
which might be considered as derivatives of basaltic magma. These include the
syenitic, quartz syenitic, and foyaitic intrusions of the Chilwa province (Malawi),
the nordmarkites of the Nuanetsi complexes, the nepheline syenite of Marangudzi,
the phonolites of Lupata, and the peralkaline rhyolites of the southern Lebombo.
Although locally these rock types are voluminous (e.g. the Mlanje intrusion of
Downloaded from http://jgs.lyellcollection.org/ at North Carolina State University on December 10, 2012

3~o K. G. Cox

Malawi) within the context of the province as a whole they are of relatively
minor importance.
In the discussion which follows, the volcanic cycle is divided into two elements
which overlap in time, a major cycle involving the mantle-derived rocks and a
superimposed minor cycle involving the rocks of crustal origin.

7. Variation of MgO with time in the mantle-derived rocks


As stated earlier the value of MgO will be used as a gauge of the temperature of
the erupted rocks.
The main period of basaltic vulcanism in the Karroo province is represented
best in the Tuli-Nuanetsi-North Lebombo area. The sequence here consists of a
few basal flows of nephelinite, occurring only in half a dozen localities, overlain
by three groups of basalts, termed the Olivine-rich Basalts, the Upper Basalts,
and the Interbedded Basalts. The latter group locally forms the top of the suc-
cession and consists of those basalts which are interbedded with rhyolites. For
the present study it is essential to consider specimens of which the stratigraphic
position is reasonably well known. Figs. 3(a) and (b) show the results of plotting
MgO against stratigraphic height for two areas, the north limb of the Nuanetsi
syncline and the south limb of the Tuli syncline. Data are from Cox et al. (1965),
Jamieson (1969, unpublished) and Vail, Hornung & Cox (1969). Stratigraphic
heights for the Nuanetsi specimens were determined from the structural map on
which Fig. 17 of Cox et al. (op. cir.) was based. Heights for the Tuli specimens
were determined on the assumption that the average dip of the southern limb is
2 °, giving a maximum thickness of I km of flows.
With reference to the figures it should also be noted that t h e U p p e r Basalt and
Interbedded Basalt sequences of Nuanetsi have not been extensively sampled.
However petrographic evidence suggests that virtually all the basaltic rocks from
the 2 km level upwards are normal basalts and probably have MgO contents of
less than 8 %. The solitary high MgO basalt at 2.2 km on the Nuanetsi graph is a
rare type, probably enriched in cumulus clinopyroxene.
Considering only MgO content, a similar and relatively simple pattern is seen
in both graphs. MgO starts relatively low, rises rapidly to a maximum near the
base of the sequence, and then falls and becomes relatively steady for the rest of
the sequence. An explanation of this phenomenon is offered in the following
generalised terms and is followed by a discussion of the applicability of the model
to other provinces.
(A) THE CULMINATION
In a major and widespread volcanic cycle there is a reasonable chance that
the rocks exposed at the surface will be representative of the magmas available
in the near-surface environment. This proviso is an important one, for the general-
isations which follow cannot be expected to hold for small areas where irrational
"quirks" of local tectonics, erosion level etc. may prevent the appearance of
certain members of the cycle. However given that the scale of study is sufficiently
large a cycle may be expected to start with extrusive rocks with relatively low
Downloaded from http://jgs.lyellcollection.org/ at North Carolina State University on December 10, 2012

The Karroo Volcanic Cycle 32 ~


•. 2 ~ -

o
c9
0 too%
<~ <I ~ 0 ~
Iolo 0
<--

ol o
0

o
0

0
0 0 .~e~ ~0
"~, ,~
0 9
0
0
0 ~- ~ =-~0
.- 0 0
0 e- 0
OlD .e-
0
,I, o Z., - ~ ' ~
0
~. o
= ~ .
•- "~ ~ "a
.9

0
•o • 0
o
o. %'I," *0 0
000~0@O • •
0
• o• .--.~
O•
O
O .- %
o
~ -o
o.~>
o e
.-, J~ A Jo 4
e4 6~ ~- ~ 0 0
A

0 ~-.- 0

o 0

i
0 n,**a o o %
0
0 0

0
0
0
o~-.~ o ~ h"
o _9
0
. _u,,l, 0 o .0 ~ .~ ~ 0~
0 ~ ._~.~ ~ - ~
¢- 8 c-

0 O
• O0 0 0
o, ~., ~- o~ o.~ O., ¢,)

•- 0 ~ u.~ ~ u
~ . ~ ~,
9

..- ..?- " 0


0
0

..,,,~ ~ ~ ~ ~ ~"
"~ ¢~.,,~. ~ ~ ~
% e~
Oo
~-,,, 0
@ 0
0 . ~ o o o
-o I:~0~ ~ 0 - , . ~, o""

II u!
~- o- & 4 a ,;,
A .-j ~ .,.,q . ~ .,,.4
o
Downloaded from http://jgs.lyellcollection.org/ at North Carolina State University on December 10, 2012

3~2 K. G. Cox

eruption temperatures. These are the magmas which first succeed in overcoming
the effects of cold wall-rocks and possibly a relatively unfractured crust. Slow
ascent through cold crust will ensure optimum conditions for fractionation
en route, and the resulting surface products will have relatively low MgO. Following
these initial products the temperature is likely to rise, so that there is a thermal
culmination during the cycle, and then will fall as heat losses bring the cycle to a
close.
In the present case the culmination occurs very close to the beginning of the
cycle, a feature which suggests the possibility that magma generation was sub-
stantially augmented by pressure relief at the beginning of the cycle. Current
plate tectonic theory, with its emphasis on the rigidity of the lithosphere, suggests
that serious consideration should be given to such a hypothesis. The effect of
pressure relief along the junction between two rigid blocks which were being
pulled apart from each other would presumably be reduced once fracturing was
well advanced and the blocks were effectively detached from each other. This
might explain the eruption of hot liquids early in the cycle (high degree of partial
melting) followed by cooler liquids.

(B) THE CSTEADY STATE' PERIOD


Figure 4 illustrates an idealised form of the volcanic cycle suggested by Figs. 3 (a)
and (b). After the culmination, the eruptions settle down to a steady emission of
basaltic lavas with relatively low and uniform MgO contents. Lava sequences
such as these have been taken to be evidence of the primary nature of the magma,
that is to say that it is produced by partial melting and subsequently erupted
without further significant change of composition. However, the polybaric
models of magmatic evolution presented by many workers including O ' H a r a
(1965, 1968 ) and Green & Ringwood (1967) have suggested that such magmas
may in fact be well evolved (see particularly O ' H a r a 1965). If we are to accept
the general validity of these hypotheses, and specifically that primary magmas are
unlikely to reach the surface unmodified by fractionation, it becomes logically
necessary to postulate a "steady state" period in any dynamic model. This
condition can only be realised when the factors affecting the temperature of the
eruptive product, that is to say rate of heat supply at source, rate of heat extraction
en route to the surface, and the rate of magmatic ascent, become balanced in some
way over a substantial period of time.
In practice it seems most likely that constancy in the erupted product is
achieved if the rates all become approximately constant. The fact that on a
world-wide scale basalts with MgO contents of about 5-8 % are by far the most
abundant may be due simply to the operation of two contrasting effects. On the
one hand the eruption of a very hot liquid is inhibited because of its high temper-
ature. A hot liquid will tend to fractionate and become a cooler liquid impoverished
in MgO. On the other hand as fractionation continues the volume of the liquid is
reduced, therefore low-MgO liquids will not be erupted in large quantities.
Somewhere in between there may lie an optimum condition for the eruption of
liquids in large volumes.
Downloaded from http://jgs.lyellcollection.org/ at North Carolina State University on December 10, 2012

The Karroo Volcanic Cycle 393

(C) TECTONIC S E T T I N G AND T H E 'STEADY STATE'


Reference to Figs. 3(a), (b) and (c) shows that the degree of perfection of the
'steady state' is quite variable. In the Nuanetsi area the Upper Basalts and
Interbedded basalts have relatively widely scattered MgO values mainly in the
range 3"5-7 %. In Tuli the range is much smaller (3"5-5"5 %) while in the Lesotho
samples there is a steady drift from about 7"5 % down to 6. 5 % with a variation of
no more than i 1% (weight) over any small part of the succession. The Lesotho
area is wholly cratonic, and Tuli, although lying in the Limpopo zone, is never-
theless well within the main southern African craton, as opposed to Nuanetsi
which lies on the strongly deformed cratonic margin. It appears, therefore, that
the steady state condition is better achieved in the more tectonically stable areas.

(D) TIME R E L A T I O N S OF T H E CYCLE


A comparison of the Tuli and Nuanetsi successions indicates that a similar pattern
in the cycle is associated with greatly differing thicknesses of erupted products.
The initial and culminatory stages give rise to I-4 km of lavas at Nuanetsi but
only o.2 km at Tuli. Although there is no indication that the lavas were erupted
at the same rate in the two areas it does raise the possibility that the cycle might
develop at quite different rates in two different areas. Thus, particular basalt types,
e.g. picritic basalts or normal basalts, need not be thought of as synchronous over
wide areas; they could be, as it were, diachronous facies related to a widespread
volcanic cycle developing in each area at its own rate.

(E) T H E W A N I N G STAGE OF T H E CYCLE


The sequences in the areas so far discussed are terminated upwards by erosion
and the youngest rocks still represent the 'steady state' condition. However a
series of phonolites is seen above the Karroo basalts at Lupata on the Zambezi
(see Woolley and Garson i97o), and since MgO contents lie below ~% it can
clearly qualify as representing the waning stage of the cycle, as so far expressed in
terms of liquid temperatures. For plutonic rocks of similarly low MgO content

1400I-~ ~.--x~ culmination


l 1300| / ~. _'steady state" waning
1200-t / MA N T L.E C Y C L E stage
T°C 1100 // "-.~
1000~/
USTALCYCLE
9001
8001 /
/ x
\

time
FIo. 4. Idealised diagram showing temperature of lavas on eruption plotted
against time. The form of the curve for the mantle cycle is based on Figs.
3 (a) and 3 (b). The form of the crustal cycle curve is conjectural.
Downloaded from http://jgs.lyellcollection.org/ at North Carolina State University on December 10, 2012

324 K. G. Cox

the age relations are more difficult to determine with the exception of the Chilwa
Series plutons of southern Malawi which are approximately contemporaneous
with the Lupata lavas. The waning stage of the cycle is not so adequately repre-
sented as the preceding 'steady state' stage. One might expect however that
waning stages of such a cycle would in general be only sporadically represented by
surface rocks.

F) THE TERMINATION OF THE CYCLE AND


THE PROBLEM OF KIMBERLITES
A major volcanic cycle must ultimately be terminated by the dissipation of heat
upwards to the atmosphere and laterally into surrounding regions. Not all
magma produced will be erupted, and that which remains below will probably
solidify from the top downwards. Ultimately, long after surface activity has
apparently ceased, isolated pockets of magma may exist at depth. It is in such
pockets that the large number of Cretaceous kimberlites of southern Africa may
have their origin (du Toit I92O, Verschure 1966). Advanced crystallisation of
eclogite from picritic magma has been postulated as capable of producing
their geochemical peculiarities (e.g. O'Hara and Yoder 1967). It seems possible
that the build-up of volatiles accompanying such crystallisation may be the
driving force behind the creation of diatremes. It is further postulated that the
restriction of the kimberlites to the cratonic areas is due to the fact that only
under the most stable tectonic conditions is it possible for the residual magma
pockets to remain sufficiently sealed to build up the necessary volatile pressure.
The hypothesis of Harris and Middlemost (1969) for the origin of kimberlites,
which is based on the fact that cratonic areas have low heat-flow, is considered
unlikely since heat flow in the southern African craton must have been very
high at times during the Mesozoic.

(G) APPLICABILITY TO OTHER PROVINCES


The hypothesis of the volcanic cycle presented in this paper can be tested, up
to a point, by comparison with other areas. The phenomena described from
southern Africa may be essentially random, and the patterns discerned by the
author may be fortuitous. On the other hand if the phenomena are related to a
single unifying event then elements of the same patterns should be detectable in
other broadly comparable provinces. If this proves to be so, the existence of
long-term thermo-tectonic events which give rise to complex series of igneous
rocks related to them, becomes a probability. This would not in itself form a test
of the mechanisms involved in the cycle but should be regarded as an essential
preliminary to further investigations.
Comparable provinces include the Deccan Traps (India), the Parafa Basin,
Baffin Island and West Greenland, East Greenland, the Columbia R.-Snake
R. province and the Siberian Platform. From existing published information it
is not easy to say whether the different stages of the volcanic cycle can be identified
in these provinces. However the Greenland and Deccan provinces are tectonically
closely analogous to the southern African province in their association with
Downloaded from http://jgs.lyellcollection.org/ at North Carolina State University on December 10, 2012

The Karroo Volcanic Cycle 325

continental margins and monoclinal flexuring, and studies in these areas might
be particularly rewarding.

8. The cycle in rocks of crustal derivation


As stated previously the present author regards the majority of the rhyolitic and
granitic rocks of the Nuanetsi-Lebombo zone as being crustal remelts. In a
previous discussion of the Karroo Volcanic Cycle (Cox et al. I965 p. 2o9) it was
assumed that the rhyolites represented the culmination of the whole cycle. With
the present assumptions, however, it becomes necessary to consider two cycles,
one concerned with rocks of mantle origin, and a second, distinct and superim-
posed, crustal cycle. The transference of heat upwards with the passage of time,
presumably largely by the movement of magmas, must inevitably mean that the
initiation of the crustal cycle lags relative to the mantle cycle. In the whole of
the Nuanetsi-Lebombo region the mantle cycle had entered the steady state
period well before the first acid eruptives appeared. The fact that during the
eruption of the rhyolites the basalts continued to appear and there were no
intermediate rocks, can be interpreted as due to there being insufficient time
available for them to become mixed. In the southern Lebombo the succession
shows that rhyolitic activity ceased but that "steady state" basalts continued to
be erupted for some time. This may also be true at Lupata but details of the
succession are not available.
The relationship between the two cycles is shown diagrammatically in Fig. 4.
At present insufficient petrographic and chemical data for rhyolites of known
stratigraphic position are available, and therefore the form shown for the crustal
cycle curve is entirely conjectural.

9" K20/Na20 ratios and the diversity of parental magmas


Variations with time of the K~O/NazO ratio are shown for Nuanetsi, Tuli and
Lesotho in Fig. 3. The similarity of the curves to those already given for MgO
is rather strong. Jamieson (I966 and I969, unpublished) showed that the major
element variation (excluding potassium) of the Nuanetsi olivine-rich rocks could
be adequately accounted for by the fractionation or accumulation of olivine and
orthopyroxene at a pressure of about 8 kb. To interpret the variation of KzO in
terms of a similar fractionation model he considered the possibility that phlogopite
might also have been a fractionating phase (Jamieson I969). However, it is
necessary to postulate the fractionation of very large amounts of phlogopite to
account for the observed K / M g relations. This must clearly remain a possibility
but the fact that no phlogopite has been observed in these rocks makes this
explanation appear unlikely. Thus it will be assumed that there must have been
a considerable range of parental magmas available to produce the olivine-rich
basalts, since they have K~O/Na~O ratios which vary by a factor of at least io.
The initial nephelinites of the Nuanetsi-North Lebombo region could however
have been derived from a single parent magma because their K,O/Na~O ratios
are substantially constant (0"2 to 0.4).
Downloaded from http://jgs.lyellcollection.org/ at North Carolina State University on December 10, 2012

326 K. G. Cox

The Upper Basalts and Interbedded Basalts could not apparently have evolved
by any normal fractionation process from a parental magma like the average
olivine-rich basalt, though on the basis of KoO/Na20 ratios they could have
evolved from magmas represented by some of the less potassium-rich flows.
The Upper Basalts, Interbedded Basalts, and gabbros of the Nuanetsi area
show considerably more range in K , O / N a , O ratio (o.i to I.O) than do the
equivalent rocks at Tuli (0"2 to o'5). The Lesotho rock show even less, so that
there is correlation between variability in this ratio and variability in MgO.
Clearly, in the ideal 'steady state' period of the volcanic cycle, as exemplified by
Lesotho, it is permissible to postulate a single relatively uniform parent magma.
Conversely, if the steady state is less perfectly developed a wider range of parental
magmas is involved. Ultimately, in the culminatory period in areas such as Tuli
and Nuanetsi it is necessary to postulate the existence of a very wide range of
parental magmas, at least in so far as their contents of incompatible elements are
concerned.
From the above considerations it might be inferred that the highly-potassic
magnesium-rich magmas of the Nuanetsi and Tuli areas did not give rise to any
more salic differentiation products. This is almost true but there are in fact a few
occurrences of rocks which correspond closely with the expected differentiation
products. One of these is the olivine monzonite of the Chilembeni intrusion (Cox
et al. 1965 p. 138 ) which forms the upper part of a differentiated picrite body.
Other examples are afforded by the shoshonites and the solitary absarokite
described by Vail et al. (1969) from Tuff. The fact that it is possible to recognise
these very distinctive fractionation products reinforces the argument that the
olivine-poor basalts are not the fractionation products of the magmas represented
by the Olivine-rich Group.

(A) RELATIONS BETWEEN THE PARENTAL MAGMAS


It was postulated in an earlier section that the volcanic cycle arose in response
to the ascent of a body of potassium-rich magma from a considerable depth within
the mantle. If ascent were slow the magma body would inevitably be overlain and
surrounded by a zone of newly melted material. Such magma, while it remained
in situ, would have normal geochemical characteristics and would be essentially
sodic rather than potassic, although the extreme peripheral zone might be
potassium-enriched as a result of minimal degrees of partial melting (e.g. see
O ' H a r a 1968 , Fig. 8; Gast 1968 ). This, however, would be quantitatively
insignificant.
The model now proposed for magma generation in the Nuanetsi-North Le-
bombo zone has three stages as follows:
I. Initial magmas (nephelinites) are shed from the upper part of the in situ
partial melt envelope.
2. Culminatory magmas are derived from both the envelope (magmas
with low K , O / N a 2 0 ratios) and the main magma body (high K , O /
Na20 ratios). There may or may not be some hybridisation between
the two types. In any event it is likely that gradations exist because the
Downloaded from http://jgs.lyellcollection.org/ at North Carolina State University on December 10, 2012

The Karroo Volcanic Cycle 327

main magma body is likely to become zoned as newly produced sodic


magma is incorporated in the upper part by upward migration of the
body.
3. Subsequently the build-up of in situ magma in the upper parts of the
system inhibits the eruption of potassic magma from deeper levels.
The eruption settles down with the emission of relatively low-potassium
lavas, but K~O/Na~O ratios are still moderately variable.
For the Tuli area the sequence of events appears to be similar though there is
no evidence of the initial sodic phase. However, bearing in mind the rather
sparse sampling of this very critical lower part of the sequence this is perhaps not
surprising.
The situation in Lesotho is clearly very different from that postulated above.
It is presumed that magma generation in areas peripheral to the main zone of
upwelling of potassic magma took place in response to a general increase of thermal
gradient which accompanied the more restricted rise of magma in the central
area. The zone of magma generation may then be visualised as a wide lateral
extension of the in situ partial melt zone of areas such as Nuanetsi. The relation-
ship is illustrated diagrammatically in Fig. 5. The absence of any potassic magma
accounts for the constant K~O/Na~O ratio of the Lesotho basalts.

I o. Silica saturation and the problem of depths


of magma generation
The earliest known rocks in terms of their relation to the volcanic cycle of the
southern African Mesozoic province are the highly undersaturated nephelinites
of the Nuanetsi-North Lebombo area. These are followed by great thicknesses
of tholeiitic basalts, which form the only rock type present in any significant
quantities in most of the cratonic regions e.g. Lesotho, Springbok Flats (N.
Transvaal), Kalahari Basin (Botswana). In the southern part of the Lebombo,
Assun,cao et al. (1962) have recorded the presence of nepheline-normative basalts
in the upper part of the sequence, while at Lupata the normal Karroo basalts and
rhyolites are overlain by phonolites. Thus the later part of the cycle is marked by
a distinct drift towards undersaturated compositions, either basaltic or of more
fractionated type. Closely associated in time with the late undersaturated rocks
are acid rocks such as the peralkaline rhyolites of the southern Lebombo and the
quartz syenitic and granitic members of the Chilwa Alkaline plutonic province.
Such rocks are perhaps differentiates of near-saturated basaltic magmas which
have compositions lying close to the plane of critical silica undersaturation in the
Yoder & Tilley (1962) normative classification (see for example Coombs 1963).
The fractionation of closely similar but slightly more undersaturated basaltic
magmas could give rise to the associated phonolific and nepheline-syenitic rocks.
Abundant experimental data (e.g. Green & Ringwood 1967; O ' H a r a 1968;
Kushiro 1968 ) suggest that undersaturated magmas become involved in frac-
tionation, or must equilibrate with mantle wall rocks (possibly, but not necessarily
Downloaded from http://jgs.lyellcollection.org/ at North Carolina State University on December 10, 2012

328 K. G. Cox

NUANETSI SWAZILAND LESOTHO

surface volcanics
+ + + + + ~- -r ÷ + + + 4- + t .__ T nr • , -

+ +
÷ t +
÷ T +
÷ T +
÷ 4- 4-~4- + . + . +. . + . +. + + AIII~
~ + + + t + 'c r u s't '"÷ 4- + + 4- + + ~ * /' ++~l+

I~ \"I-- - \ - -,\ - -~x , / ~.X,- - x / x .- . I - ' - --


\ /
\ \ /~- - / ~ N , ..,a r . a~.I " melt /i-/
\ =~
, ~ / / \.,~

~//~/~/zOner/////....//./..~/
\ -- ~

_~..-.. o I
7//// ~ ,,./-/ i

,n. s.,tu/..par2,/a.l.,
~ . / / > f /././

me'l(.Y/////~~9//////..//////////~
/./..- ..

__,_..:. i~/- Z " L\ ,', 'l , I - ' _ " / ( "_x _ ~t ? y - L ' ~


I n n r f i n l m~JfilmIF ~ X • . .:-\"~A . . . . J " / " " , " \ ",4
• O -
~ i ~ > ~ : - : : _
.'::-.--"-/
\ unaffected mantle ", - Ikx/-x
--
200kml

L " . . . . . -.'. • " -'-~"


- / - ',
,
\ 1 "," , - - /\' , ", ;-,
- I
\ / - ,\ , i'(-
',' /
" --
";I/ I

F IO. 5-Diagrammatic section showing postulated relationship of zones of


partial melting. The line of section runs from the Rhodesian shield (left)
southwards across the Nuanetsi syncline, southwards across the Transvaal
shield (west of the Lebombo), takes a brief excursion eastwards into Swaziland
and then runs south-west to Lesotho (see Fig. I). T h e apparent synclinal
structure in the surface volcanics in Swaziland is due to the bend in the line of
section. Horizontal and vertical scales are equal.

during the process of partial fusion) at an intermediate depth within the mantle.
Depths suggested are approximately 25-65 km (O'Hara I968 ) and 35-7 ° km
(Green & R.ingwood 1967). These estimates give some idea of the highest levels
at which the zones of magma generation lay during the initial and final phases of
the cycle. The magmas concerned may have originated at greater depths than
these but there seems to be no direct means of determining whether this was so.
However it is worth recalling that virtually all the undersaturated rocks have
normal K~O]Na~O ratios and there is therefore no particular reason to suppose
that they had a substantial early history of eclogite fractionation, zone refining,
or reaction with wall rocks.
The depth of origin of the tholeiitic rocks is more problematical. It has been
argued in an earlier section that the potassium-rich tholeiites of areas such as
Nuanetsi may have originated at great depths within the mantle. Jamieson's
study of phase relations (I966), however, showed that the only available direct
evidence is a record of fractionation history at levels which are probably near the
top of the Upper Mantle. Jamieson (1969, unpublished) suggests that the observed
trend of olivine and orthopyroxene fractionation seen in the olivine-rich basalts
was produced at a depth of 22-3 ° km. Above that level the olivine-rich magmas
were transported to the surface so fast that they did not undergo further frac-
tionation. Again, as with the undersaturated rocks, there is no direct means of
deciphering any details of their history at earlier stages. However some speculation
is possible.
Downloaded from http://jgs.lyellcollection.org/ at North Carolina State University on December 10, 2012

The Karroo Volcanic Cycle 329

I f it is assumed that the nephelinitic initial eruptive products originated as


partial melts in the envelope of the main potassic magma body at a depth of
some 5° km or more, the main body may have lain at somewhat greater depths
and been considerably hotter at this stage. At depths of greater than about 65 km,
even initial melts of mantle peridotite would be hypersthene normative (O'Hara
I968), and at slightly lower pressures only moderate amounts of additional
melting would be required to achieve this condition. Hence, on the grounds of
both depth and temperature, it seems likely that the main magma body was
tholeiitic at the time of the initial eruptions. The body must subsequently have
given rise to magmas which reached the 22-3 ° km level without losing their
tholeiitic character. This could be achieved by fairly rapid ascent so that they
only fractionated olivine (O'Hara ot3. cit.) or by a slow ascent provided that the
liquid remained sufficiently hot to avoid fractionation of harzburgite. There is
some evidence that olivine fractionation may in fact have been involved for
Jamieson (op. cit.) has shown some of the picritic intrusives of the Nuanetsi area to
be versions of the potassium-rich lavas enriched in cumulus olivine. On the other
hand it is possible that not all the magmas derived from the main body succeeded
in reaching the 22-30 km level without losing their tholeiitic character. The two
analysed shoshonites from Tuli, for example, contain small quantities of nepheline
in the norm. Other rocks such as the Chilembeni olivine-monzonites are almost
exactly critically undersaturated, while a small intrusion of a similar rock type
discovered by Jamieson (op. cir.) contains significant amounts of normative nephe-
line.
The olivine-poor and olivine-free basalts erupted so copiously during the
'steady state' period of the volcanic cycle present the greatest problems when it
comes to unravelling their history. Their advanced condition of low-pressure
fractionation of phases including plagioclase and ore minerals precludes any
detailed investigation of high-pressure events. However the moderately potassic
character of many of the lavas belonging to this stage in the Nuanetsi-North
Lebombo region suggests that here at least there was still some contribution of
material from the main magma body. The tholeiitic character suggests that for
much of this period magmatic ascent was still sufficiently rapid to be controlled
largely by olivine fractionation. However, ultimately, undersaturated rocks began
to appear (southern Lebombo) with the implication that the rate of ascent began
to decline or that the rate of heat loss to wall rocks began to increase.
In regions such as Lesotho the erupted rocks are so uniform that little can be
said of them as regards depth of origin or last equilibration with mantle rocks.
The evidence of the amount of mantle processed (at least 5 ° km.) presented in an
earlier section, suggests that they may have originated over a moderate depth-
range.

I I, C o n c l u s i o n s

A simplified chart illustrating the postulated origins of the various rock types is
given in Fig. 6. References to the 'rate of ascent' imply mainly the rate of ascent
Downloaded from http://jgs.lyellcollection.org/ at North Carolina State University on December 10, 2012

33 ° K. G. Cox

THOLEIITIC K-RICH PICRITIC _ _ moderate low-pressure ~ ABSAROKITE


very rapid ascent BASALT AND PICRITE ffaationation ~ OLIVINE MONZONITE

POTASSIUM-RICH ~.____ __ U NDERSATURATED SHOSHONITE


PARENTAL MAGMA ~ -- slow ascent OLIVINE MO NZONITE

HLY U NDERSATURATED ? ~ KIMBERLITE


advanced hlgh-pressure
fractlonation followed
by very rapid ascent

THOLEIITIC moderate low-pressure OLIVINE-POOR, K-POOR


very rapid ascent ~ K-POOR
BASALT PICRITIC
AND PICRITE fmctio~ion BASALT AND GABBRO

POTASSIUM-POOR ~ INTERMEDIATE , BASALT NEAR PLANE advanced Iow-~ssure l O PERALKALINERHYOLITE


PARENTAL MAGMA ~ slower ascent ~ OF CRITICAL ~ fractionatim - ~ QUARTZ SYENITE
UNDERSATURATION ALKALI GRANITE

\\ UNDERSATURATED advanced low-~essure PHONOLITE


\~ .I . . . . . . . t ~ ALKALI BASALT fractlanation -- ? "~-~" NEPHELINE SYENITE

\\
\ HIGHLY UNDERSATURATED NEPHELINITE moderate low-pressure ~ ' ~ NEPHELINITE
very slaw ascent ~ ?MELANEPHELINITE - - fractionatlon

\advanced hlgh-pressure
frectionatlon fallowed ? ~ KIMBERLITE
by very rapid ascent

PARENTAL ACID CALC-ALKALI NE


MAGMA ~ RHYODACITE, RHYOLITE
GRANITE, GRANOPHYRE

F I o. 6. Schematic diagram showing postulated genetic relationships of main rock-


types.

relative to the rate of heat extraction through the critical depth range of about
25-65 km where fractionation and partial melting, and equilibration with wall
rocks are likely to give rise to undersaturated products.
Other main postulates of this paper are summarised as follows:

I. All the Mesozoic igneous rocks of Southern Africa are considered to


belong to a single long cycle of igneous activity.
2. The cycle was initiated by the rise of a potassium-rich magma body
approximately in the Rahodesian area. This may have taken 80 m.y. to
evolve and may have originated at a depth in excess of 500 km.
3. A zone peripheral to the Rahodesian area was characterised by more
normal magmatism derived from only moderate depths in the mantle
and caused only by increased heat-flow.
4. A cyclic developement of mantle-derived magmas shows early well-
fracdonated products (initial phase) followed by the most primitive
rocks (culminatory phase). A subsequent 'steady state' period is char-
acterised by relatively uniform lavas. A waning stage is represented
by a return to more fractionated products, sporadically developed.
5. A second cycle, of rocks produced by anatexis in the crust, is superimposed
on the cycle of mande-derived rocks. Acid and basic rocks belonging to
the respective cycles show almost no tendency to hybridise.
Downloaded from http://jgs.lyellcollection.org/ at North Carolina State University on December 10, 2012

The Karroo Volcanic Cycle 331

6. I n terms of silica s a t u r a t i o n in the m a n t l e - d e r i v e d rocks the d e v e l o p m e n t


is also cyclic. T h e m a i n p e r i o d of v u l c a n i s m is tholeiitic b u t b o t h
initial a n d final phase are c h a r a c t e r i s e d b y u n d e r s a t u r a t e d m a g m a s .
This is i n t e r p r e t e d as being largely controlled b y the effectiveness or
otherwise of f r a c t i o n a t i o n a n d / o r e q u i l i b r a t i o n w i t h wall rocks in the
d e p t h r a n g e 2 5 - 6 5 km.
7. E r u p t e d p r o d u c t v a r y from h i g h l y potassic types derived from the m a i n
b o d y of potassic m a g m a , to p o t a s s i u m - p o o r types d e r i v e d from m a g m a s
p r o d u c e d as p a r t i a l melts m a r g i n a l to the m a i n body.
8. K i m b e r l i t e s are r e g a r d e d as o r i g i n a t i n g from pockets of residual m a g m a
t r a p p e d at d e p t h after the t e r m i n a t i o n of the m a i n cycle.

ACKNOWLEDGEMENTS, The writer is particularly indebted to Dr. B. G. Jamieson who generously


allowed him to draw on much unpublished information on the Nuanetsi lavas.

I2. References

ASSUNgAO, C. F. T. DE, CO~LHO, A. F. T. P. & ROCHA, A. T. I962. Petrologia das lavas dos
Libombos (Mo,eambique). Estudos, Emaios Docum dta Invest. Ultramar, 99, 74 PP.
COOMBS,D. S. 1963. Trends and affinities of basaltic magmas and pyroxenes as illustrated on the
diopside-olivine-silica diagram. Miner. Soc. Am. (Spec. Pap. 1), 227-5o.
Cox, K. G. 197o. Tectonics and vulcanism of the Karroo period and their bearing on the postu-
lated fragmentation of Gondwanaland. In Clifford, T. N. & Gass, I. G. (Eds.) African
magmatism and tectonics. Oliver and Boyd, Edinburgh, 211-35.
~ ( i n press). Karroo lavas and associated igneous rocks of southern Africa. Bull. volcan.
& HomnuNG, G. 1966. The petrology of the Karroo basalts of Basutoland. Am. Miner.,
5 x, 1414-32.
, JoHnsoN, R. L., MONKMAI%L. J., STILLMAN, C. J., VAIL, J. R. & WOOD, D. N. I965.
The geology of the Nuanetsi igneous province. Phil. Tram. R. Soc., Ser. A. 257, 71-218.
, MACDONALD, R. & HoistinG, G. I967. Geochemical and petrographic provinces in the
Karroo basalts of southern Africa. Am. Miner., 52, 1451-74 •
vu TOlT, A. L. 192o. The Karroo dolerites of South Africa: a study in hypabyssal injection.
Trans. geol. Sot. S. Afr., 23, 1-42.
, I929. The volcanic belt of the Lebombo---a region of tension. Trans. R . Soc. S. Aft. x8,
189-217.
, 1954. The geology of South Africa. 3rd edition. 0liver & Boyd, Edinburgh.
FrrcrI, F. J. & MILLER, J. A. (in press). Potassium-argon radioages of Karroo volcanic rocks
from Lesotho. Bull volcan.
FLORES, G. I97O. Suggested origin of the Mozambique channel. Trans. geol. Soc. S. Aft., 73, I-I6.
GAST, P. W. 1968. Trace element fracfionation and the origin of tholeiitic and alkaline magma
types. Geochim. cosmochim. Acta, 32, lO57-86.
Gm~l% D. H. & RINGWOOD,A. E. 1967. The genesis of basalt magmas. Contr. Mineralogy Petrology,
x5, IO3-9 o.
GURN~Y, J. J. & B~Ro, G. W. (in press) Potassium, rubidium and cesium in South African
kimberlites arid their peridotite xenoliths. Geological Society of S. Africa: Upper Mantle Sym-
posium, July 1969.
HAm~lS, P. G. 1957. Zone refining and the origin of potassic basalts. Geochim. cosmochim. Aaa,
12, i95-2o8.
~& MmDLPMOST,E. A. K. I969. The evolution of kimberlite. Lithos, 3, 77-88.
Downloaded from http://jgs.lyellcollection.org/ at North Carolina State University on December 10, 2012

332 K. G. Cox

HAUGHTON,S. H. 1963 . Stratigraphie history of Africa south of the Sahara. Oliver and Boyd, Edinburgh.
365 pp.
HF~DERSOI% J. MCC. 19o9. Notes on some rocks in the volcanic series of the Karroo System in
the Lebombo mountains. Trans. geol. Soc. S. Afr., 12, 24-31 .
HUNTER, D. R. & URIE, J. R. 1958. Some recent investigations in Stormberg lavas. A. Rep.
geol. Surv. Swaziland, 42-5.
HURLEY, P. M. 1968. Correction to: Absolute abundance and distribution of Rb, K, and Sr in
the Earth. Geochim. cosmochim. Acta, 329 lO25-3 o.
ITO, K. & Kzm~DY, G. C. 1967, Melting and phase relations in a natural peridotite to 40
kilobars. Am. J. Sci., 265, 519-38.
JAmeSON, B. G. 1966. Evidence on the evolution of basaltic magma at elevated pressures. Nature
Lond., 2129 243-6.
1969. The petrology of olivlne-rich basaltic rocks, Nuanetsi, Rhodesia. Unpublished Ph.D.
thesis. University of Edinburgh.
& CI.ARI~, D. B. 197o. Potassium and related elements in tholeiitic basalts. J. Petrology,
xx, I83-0o 4.
KusHmo, I. I968. Compositions of magmas formed by partial zone melting of the earth's upper
mantle. J. geophys. Res., 739 619-34.
Lr. PmHON, X. 1968. Sea-floor spreading and continental drift. J. geophys. Res., 73, 3661-97.
LIOHTFOOT, B. 1938. Notes on the south-eastern part of Southern Rhodesia. Trans. geol. Soc. S. Afr.,
419 193-8.
LormAARD, B. V. I950. Karroo dolerites and lavas. Tram.geol. Soc. S. Afr., 55, 175--98.
MA2croN, W. I. I968. The origin of associated basic and acid rocks in the Lebombo-Nuanetsi
igneous province, southern Africa, as implied by strontium isotopes. J. Petrology, 9, 23-39.
& SmDNER, G. 1967. Age of the Paresis complex, South West Africa. Nature Lond., =16,
1197-8.
MORGAN, W . J . i968. Rises, trenches, great faults, and crustal blocks. J. geophys. Res., 739 I959-80-
O'HAa~, M . J . i965. Primary magmas and the origin of basalts. Scott. J. Geol., x~ i9-4 o.
~ 1 9 6 8 . The bearing of phase equilibria studies in synthetic and natural systems on the origin
and evolution of basic and ultrabasic rocks. Earth-Sci. Rev., 49 69-133.
& YOD~R JR., H. S. I967. Formation and fractlonatlon of basic magmas at high pressures.
Scott. J. Geol., 3, 67-I 17.
OLrvxR, J. & ISACKS,B. 1967 " Deep earthquake zones, anomalous structures in the upper mantle,
and the lithosphere. J. geophys. Res., 7=, 4059-75.
PRIOR, G. T. I9IO. Petrographic note on the dolerites and rhyolites of Natal and Zululand.
Ann. Natal Mus., ~,9 141-57.
RmGWOOD, A. E. I966. The chemical composition and origin of the Earth. In: Hurley, P. M.
(ed.), Advances in earth sciences. M.L T. Press, Boston, Mass., 087-356.
Roo~Rs, A. W. 19o5. Notes on the north-eastern part of the Zoutpansberg district. Trans. geol.
Soc. S. Aft., =89 33-53.
SI~DI~r.R, G. & MILLER, J. A. I968. K-Ar age determinations on basaltic rocks from South West
Africa and their bearing on continental drift. Earth Planetary Sci. Let., 49 451-8.
SWIFT, W. H., WHirr., W. C., WILr.S, J. W. & WORST, B. G. 1953. The geology of the Lower
Sabi Coalfield. Bull. geol. Surv. Sth. Rhod. 409 96 PP.
VAIL, J. R. I967. The southern extension of the East African rift system and related igneous
activity. Geol. Rdsch., 579 6o1-14.
, HomburG, G. & Cox, K. G. 1969. Karroo basalts of the Tuli syncline, Rhodesia. Bull.
volcan., 33, 398-418-
van EEDr.N, O. R., VlSSr.R, H. N., VAN Zyr, J. s., COr.RTZr., F. J. & WESSr.LS,J. T. I955. The
geology of the eastern Soutpansberg and the lowveld to the north. Geol. Surv. S. Afr. Ex-
planation of sheet 40 (Soutpansberg).
Vr.RSCHURr., R. H. I966. Possible relationships between continental and oceanic basalts and
kimberlites. Nature Lond., axx, 1387.
WALK~*R, F. & POLI)r.RVAART, A. 1949. Karroo dolerites of the Union of South Africa. Bull.
geol. Soc. Am., 6o, 591-7o6.
Downloaded from http://jgs.lyellcollection.org/ at North Carolina State University on December 10, 2012

The Karroo Volcanic Cycle 333

WOOLLEY, A. R. &GARSON, M. S. 197o. Petrochemical and tectonic relationship of the Malawi


carbonafite-alkaline province and the Lupata-Lebombo volcanics. In Clifford, T. N. &
Gass, I. G. (Eds.) African magmatism and tectonics. 0liver and Boyd, Edinburgh., 237-62.
YOUNO, R. B. I92o. The rocks of a portion of Portuguese East Africa. Trans. geol. Soc. S. Afr., **3,
98--x I3 .

Revised manuscript received 13 July x97I ; read 24 November I97I


K. G. Cox, Grant Institute of Geology, University of Edinburgh, West
Mains Road, Edinburgh.

DISCUSSION
Mr. F. J. FITCH asked the author if he could relate the volume of lava being
erupted at any particular time and place to the stage in the volcanic cycle attained
at that locality. He confirmed that the great bulk of Karroo Volcanism was
confined to a short period of about 20 m.y. in duration at the beginning of
Mesozoic time, but agreed that the total time span of Karroo and apparently
related magmatic activity might be IOO m.y. or more. In collaboration with
Dr. J. A. Miller he had recently suggested that the time sequence of Karroo
activity viewed on the K-Ar time scale is as follows:
Stages of cycle
Waning phases Lupata alkali-basalts ( i o 5 - I i 5 4- IO m.y.)
Subsidiary maxima of 'Karroo' (e-~ 155 m.y.; 161 m.y.)
dolerite magmatism and meta-
somatism
Lupata rhyolites (I66 4- IO m.y.)
Granites of Mateka Hills (I67 4- 7 m.y.)
Subsidiary maxima of 'Karroo' (,-~ i72 m.y.)
dolerite magmatism and
metasomatism
Culminating phases 'Karroo' dolerites, Drakensberg (,~ I87 4- 7 m.y.)
basalts of Lesotho and wide-
spread flood basalts of Central
Southern Africa
Nuanetsi and Labombo rhyolites (I9O-I94 4- I2 m.y.)
Waxing phases Lower alkaline and tholeiitic (e-~a190-I 95 m.y.)
basalts of Nuanetsi and
Lebombo districts
Marangudzi igneous complex (,-~, 195 m.y.)
Shawa igneous complex (,'~ 197 m.y.)

The AtrrHog thanked Mr. Fitch for his contribution to the time relations
of Karroo igneous rocks. When referring to stages in the volcanic cycle it is
necessary to be clear about the definitions employed. In terms of the thermal
cycle discussed in the present paper there is little or no relation between the
stage of the cycle and the volume of lava erupted. In most of the area, for example,
Downloaded from http://jgs.lyellcollection.org/ at North Carolina State University on December 10, 2012

334 K. G. Cox

the initial and culminatory stages are not represented by any surface material
at all. Mr. Fitch's stages are based mainly on volume relationships and as such
the culminatory period must of course refer to the widespread olivine-poor
tholeiites and the Lebombo rhyolites. It is also worth noting that the radiometric
ages suggest the thermal cycle may be developed with strong diachroneity,
since the 'steady state' period may have been reached in the Lebombo-Nuanetsi
region rather earlier than elsewhere.

Professor I. G. CrAss said that in one of his diagrams, Dr. Cox depicted mantle
and crustal thermal events by plotting temperature against time. No indication
was given of the temperature or time scales although the former could be deduced
from experimental data. Could Dr. Cox indicate from the East African evidence,
the time interval between his proposed mantle and crustal thermal peaks ?

The AUTHOR replied that the time interval between the proposed thermal
peaks in the mantle and crustal cycles was likely to be less than IO m.y., based on
currently available radiometric ages.

Dr. J. B. AUDEN asked if a dyke swarm was associated with the Lebombo
monocline in the same manner as in Greenland. The Panvel flexure which
affects the Deccan lavas along the west coast of India between latitudes 18° and
21 ° displays a network of dykes, the longest, 20-30 km in length, running north-
south parallel to the monocline axis. The major dyke swarm is not, however,
related to the flexure, but is orientated ENE-WSW in the Narmada-Tapi region
at the west end of the great lineament extending IO° of longitude across the
peninsula. The dykes are up to ioo m. in width and some exceed IOO km in
length (Geol. Mag., 9 I, pp. 95, 98, 1954).
It is only in the area west of longitude 74 ° that striking differentiates are present
connected with the plutonic centres of Saurashtra, Broach and Baroda, including
nepheline syenite, olivine gabbro, limburgite, monchiquite, lamprophyre
(Rec. Geol. Surv. Ind., 58, p. 382, 1925: Journ. Geol. Soc. Ind., IO, p. 177 , 1969).
The author relates the Karroo volcanic cycle to the initiation of disruption of
Gondwanaland. The greater part of the Karroo volcanics, being Triassic to
Jurassic in age, are much older than the Deccan lavas. Although almost 5oo,ooo
km 2 of the Deccan lavas lack a cover of rocks to act as a stratigraphic control,
the lavas of the Cambay graben, Kutch and westwards to the Kirthar range near
Karachi are known from surface evidence and from exploratory wells to range in
age from topmost Cretaceous to Palaeocene, or roughly 6o-7o m.y., which is
supported by K-Ar determinations by McElhinney (Nature Lond. 227, p. 595,
197o ). It follows that the Panvel flexure, which affects a column of 3ooo m. of
lavas, must be later than Palaeocene, and presumably developed subsequent to
the initiation of the Lebombo monocline. The plateau basalts of Ethiopia are
even younger in age, and it is evident that volcanic phenomena migrated across
the dismembered Gondwanaland throughout a long period of time, during which
the fragmented units were moving apart.
Downloaded from http://jgs.lyellcollection.org/ at North Carolina State University on December 10, 2012

The Karroo Volcanic Cycle 335

The Lebombo flexure has a strong gradient of Bouguer gravity anomalies over
a width of2o km, from --5 ° mgals on the west side to + 5 o mgals along longitude
3 2°. A somewhat similar condition obtains on the west coast of India, with the
difference, however, that there is a pronounced gravity high 3 ° km north of
Bombay city, west of which the anomalies become negative, with a gradient of
2.2 5 mgals/km over a distance of 7° km. This gravity high is attributed by M.
Takin to the presence of a mass of dense rock under Salsette island. (Geophys.
Journ. R.A.S., : i , p. 531, 1966 ). Some aspects of the Deccan volcanic problem
were discussed by the writer 22 years ago (Trans. Nat. Inst. Sci. Ind. 3, PP- 123-157,
1949).

In reply the AUTHOR said that the Lebombo monocline was associated with a
dyke swarm similar to that of East Greenland but an analogy to the Narmada-
Tapti dykes is also found, in the dyke swarms associated with the Limpopo
lineament. Considering the difference in age the similarity between the Karroo
and Deccan provinces is remarkable. The author had postulated that the Karroo
phenomena might be associated with the separation of Madagascar from Southern
Africa, while the Deccan phenomena were associated with the later separation
of India from Madagascar (in African magmatism and tectonics pp. 211-35 , T. N.
Clifford & I. G. Gass, eds., 197o ).

Mr A. M. Q UENNELL suggested that if Dr. Cox is correct in relating the igneous


activity described to the disruption of Gondwanaland, then a comparison with
the neogene vulcanism of East Africa may be fruitful. Tectonic setting, that is,
disruption of a continental mass, is similar, and igneous history could be expected
to follow similar courses.

The AUTHOR agreed that a comparison with East African vulcanism would be
interesting and he had given some thought to the problem in general terms. The
relative rareness of basaltic and more marie rocks coupled with the prominence
of salic rock types such as trachyte and phonolite suggests that the East African
cycle, if it exists, was characterised by generally lower temperatures of erupted
products. Thus it is interesting to speculate what a cycle of the postulated Karroo-
type might give rise to if magma ascent were accompanied by greater amounts
of heat loss. Firstly, if heat loss were considerable at mantle depths, a crustal
anatectic stage might not develop. Secondly, a middle period of silica-oversat-
urated vulcanism might not be found, but instead a period during which the
silica-undersaturation of the surface products was simply reduced. However a
culminatory period of relatively high temperature products would still be present,
though one would not expect them to be so picritic as in the Karroo province.
Syntheses of existing data on East Africa should be made in order to test these
predictions.

Professor R. M. SHACELETON said that Dr. Cox interprets the Lebombo


rhyolites and other Karroo acid magmatic rocks as melts from the crust rather
Downloaded from http://jgs.lyellcollection.org/ at North Carolina State University on December 10, 2012

336 K. G. Cox

than derivatives of the mantle, because plotting abundance against silica per-
centage yields a bimodal curve. Yet the low initial SrS~/Sr8e ratios of the Lebombo
rhyolites suggest their mantle origin. To the speaker it appeared that the bimodal
distribution must in some cases be accepted as a normal result of processes in the
mantle, not only because it is found in places such as Iceland and the Ert Ale
range in Afar, where continental crust appears to be absent and where the acid
rocks have low initial SrS~[Sr8e ratios, but primarily because the initial separation
from the mantle of the continental crust, with its abundance of granites, is itself
an expression of the same bimodal distribution. Dr. Cox replied briefly to Prof.
Shackleton.

You might also like