下载掌阅APP，畅读海量书库
立即打开
Afar Mantle Plume: Rare Earth Evidence
J–G. Schilling
Editor’s Note
Iceland has an unusual geophysical setting, being located over two distinct sources of volcanic activity: a mid-ocean ridge, where hot material wells up from between two diverging tectonic plates, and a hotspot or mantle plume, where a column of magma rises from deep in the mantle. This paper by Jean-Guy Schilling was one of the first to provide confirmation of this picture. Since the two volcanic sources overlap at Iceland, they are hard to distinguish. But Schilling identifies a geochemical signature in the island’s basaltic (volcanic) rock that is distinct from that of mid-ocean ridge material and probably comes from a more “primordial” mantle plume. This helped to secure the very notion of deep-seated mantle plumes. 中文
Rare earth pattern zonation about the Afar Triangle, in tholeiitic basalts erupting along the Red Sea Trough and the Gulf of Aden, suggests the presence of a primordial mantle plume rising beneath the Afar and overflowing in a star-like fashion into the soft asthenosphere. 中文
THE origin of the Afar Triangle, the locus or near locus of the triple junction formed by the Red Sea–Gulf of Aden and East African Rift, has been debated in detail recently 1,2 . It is particularly important to determine whether the Afar Triangle has controlled the evolution of the three lines of plate divergence or vice versa , or what are the real relationships between these three different tectonic segments and the Afar Triangle. 中文
The first analyses in terms of oceanic plate tectonics were made by fitting the Red Sea conjugated coastlines. McKenzie et al . suggested 3 that the Afar Depression has been largely created from mantle materials during the separation of Arabia from Africa and, in support, referred to a preliminary report on a magnetic survey of the Afar Triangle 4 . In its latest form, this survey 5 reveals magnetic anomalies of three kinds: first, westward extension of the Gulf of Aden oceanic magnetic anomalies into the southern Afar; second, small amplitude and wavelength anomalies of continental sialic origin; and third, northwest–southeast anomaly trends fanning towards the south, north of about 12°. This suggests complex tectonic trends and geology of both continental and oceanic kind. Considering in greater detail the geology and tectonics of the Afar, Mohr 6,7 objected in part to McKenzie’s conclusions 3 and offered a refined model emphasizing the importance that the Danakil and Aisha Horsts may have played as continental remanent blocks. Independently Gass 8 proposed a three stage lithothermal model describing the embryonic development of a hot plume rising beneath the Afar. More recently and without further justification, Morgan 9-11 listed the Afar to be one of the 20 hot plumes which he believes may drive the lithospheric plates in a global way by viscous drag on their base. 中文
I now present rare earth geochemical evidence which gives strong support to the existence of a hot mantle plume rising beneath the Afar Triangle. 中文
I reinterpret rare earth abundance data from submarine tholeiitic basalts erupted along the Red Sea Trough, the Gulf of Aden, and tholeiites from Jebel-Teir Island, reported earlier, by first drawing analogy with similar but more extensive evidence obtained from the Iceland–Reykjanes Ridge-Plume System. 中文
La, K 2 O, TiO 2 , P 2 O 5 , radiogenic lead and perhaps strontium in tholeiitic basalts, as well as pyroxene phenocryst abundance relative to plagioclase, are now known to decrease regularly with distance from Iceland along the Reykjanes Ridge 14-18 . Over Iceland and south of 61° N along the Reykjanes Ridge, the abundances of these elements stay relatively constant but the level is different for both regions. The transition between the two end-member types of tholeiites, distinct chiefly in large lithophile element contents and radiogenic isotopes, occurs over some 400 km. It bridges island tholeiitic basalt type (for example, Iceland, Hawaii, Galapagos, Réunion, Jebel-Teir) to submarine tholeiite type erupting along most mid-ocean ridges away from any hotspot interference, respectively 13 . 中文
The progressive depletion of large ionic lithophile trace elements is a regular function of the ionic radius, particularly for the rare earth series. The smallest ions from Gd-Lu remain relatively constant in abundance and the light rare earths show a more pronounced variation along the Reykjanes Ridge Axis. Lanthanum, the largest rare earth ion, shows the greatest variation, and Sm an intermediate one. The La/Sm concentration ratio or related functions, such as the ratio of enrichment factors relative to chondrites, illustrate this point (Fig. 1). The [La/Sm] E.F. decreases markedly away from Iceland (possibly in a stepwise fashion). This ratio is a good indicator of fractionation of the light rare earth and can, with the heavy rare earths, set rigorous limits on the genesis of these basalts 13 . 中文
Fig. 1. La/Sm enrichment factors relative to chondritic meteorites 13 in tholeiitic basalts erupted along the post-glacial Reykjanes Ridge Axis and its extension over Iceland (see map insert).
A [La/Sm] E.F. >1 generally indicates light rare earth enriched fractionation patterns, such as for alkali basalts or island tholeiites (in the model I present here, tholeiites derived from primordial hot mantle plume [PHMP] including island tholeiites and related tholeiitic plateau basalts). 中文
[La/Sm] E.F. <1 generally indicates light rare earth depleted submarine mid-ocean ridge tholeiites (in my model, mid-ocean ridge tholeiites derived from the depleted low velocity layer far away from any hotspot interference [DLVL]). 中文
Fig. 2. La/Sm enrichment factors relative to chondritic meteorites 13 in tholeiitic basalts erupted along the Red Sea Trough and the Gulf of Aden, plotted as a function of radial distance from the centre of the Afar Triangle, irrespective of direction (see map insert).
Finally [La/Sm] E.F. ~1 refers to submarine mid-ocean ridge tholeiites generally occurring near Morgan’s proposed hot mantle plumes or triple junctions, and which could not be explained at the time 13 (in my model these hybrid tholeiites are produced by mixing [PHMP] and [DLVL] derived tholeiites). 中文
These results are not unique to the Reykjanes Ridge Iceland region. Similarly to Fig. 1, reconsideration of rare earth data for the Afar region 12,13 indicates that the [La/Sm] E.F. also decreases progressively with radial distance away from the proposed Afar hotspot centre 12,13 (Fig. 2). The rare earth patterns change from light RE enriched near the Afar hotspot (Jebel-Teir Island Volcano) to relatively flat RE patterns unfractionated but enriched relative chondrite abundances at intermediate radial distance (Gulf of Aden Ridge Axis), and finally to progressively more light RE depleted patterns along the Red Sea Trough further away from the Afar. All these basalts are tholeiitic in composition, again judging from the CIPW norm and Yoder and Tilley’s classification 19-22 . 中文
Thus a similar phenomenon seems to occur away from the Afar hotspot as in the case of the Iceland hotspot. The rate of progressive light RE depletion is smaller for the Afar. No data are available in tholeiitic basalt erupted from fissures along the Danakil Depression 23 , but these would be very interesting, because the results should serve as a test of the hot mantle plume mixing model I recently proposed 14-15 , and which is now applied to the Afar. 中文
The model requires two mantle sources distinct in large ionic lithophile trace elements and in radiogenic lead and strontium contents. Tholeiitic basalts are derived by partial melting at shallow depth from either of these two sources, interact and mix in various proportions along the zones of plate divergence, such as the Red Sea Trough and the Gulf of Aden (or Reykjanes Ridge). 中文
The first mantle source, primordial hot mantle plume [PHMP], upwells vertically and forcefully beneath the centre of the Afar Triangle (Iceland), and overflows into the surrounding upper asthenosphere either radially or preferentially along directions of lithosphere weakness, such as the Danakil Depression–Red Sea Trough, the Gulf of Aden, and perhaps the Ethiopian Rift (Reykjanes and Kolbeinsey Ridges). The PHMP is relatively richer in large ionic lithophile elements, and produces tholeiitic basalts with [La/Sm] E.F. >1. Further, the PHMP material has remained before rising long enough as a closed system deep in the mantle (greater than 250 km) to build up its radiogenic isotopes 24-26 . During its rise and decompression at shallow depth it will produce with sufficient extent of partial melting 13 , island tholeiite type basalt, such as for Iceland 27 , Hawaii 28 , Galapagos 29 and Réunion 30 , and by prediction tholeiites erupted over the Afar along the Danakil Depression 23 . Such tholeiitic basalts are here called PHMP-tholeiites. 中文
The second mantle source (DLVL), of more global extent, is the low velocity layer. This layer is characteristically depleted in large ionic lithophile elements (K, Rb, Cs, Ba, U, Th, light rare earths, and so on) 13,31 , and low in radiogenic Pb and Sr (refs. 32, 33). Usually, away from any hot plume interference, the DLVL feeds, passively and in response to plate divergence, mid-ocean ridge spreading with accreting lithospheric and crustal materials such as refractory solids, solid and melt mush, and melt. Most mid-ocean ridge basalts, depleted in the above elements with [La/Sm] E.F. <1, and also low in radiogenic Pb and Sr, are derived from the DLVL source. This is the case for tholeiites erupted along the Red Sea Trough, north of 22°N (south of 61°N on the Reykjanes Ridge). 中文
Near hotspots, between these two extremes there is a zone of transition, where interaction and mixing of the two principal tholeiitic melt types will occur (and perhaps of the two mantle sources as well). This is the case for submarine ridge basalts analysed from the Gulf of Aden (comparable to the Reykjanes Ridge around 63° N), and characterized by hybrid rare earth patterns intermediate between the two extreme component melt types. Thus flat rare earth patterns with [La/Sm] E. F. ~1 roughly represents an equal mixture, whereas in Jebel-Teir the PHMP-tholeiite is likely to dominate. 中文
So far the concept corresponds to Morgan’s original paper on hot plumes 9 , but differs over some minor points on which I have accumulated evidence and which can be easily tested. The model is also very similar to Gass’s elegant model 8 , but with some distinctions made on mantle source compositions. On the other hand, the model is quite distinct from Gass’s recent convective model 34 . 中文
In considering the evidence for the Iceland–Reykjanes system in detail 14-15 , I have proposed two mechanisms for the mixing of the two primary lava types to occur, both probably operating simultaneously. One mechanism calls for elongated magma chambers beneath the zones of rifting, and the other requires horizontal dike propagation over long distances as proposed and demonstrated for the Kilauea Rift Zone 35,36 . 中文
The mixing proportions along transitional zones depend on the PHMP flow pattern and intensity about the plume into the surrounding LVL. The overflow differs for each plume as well as with time (ref. 37; and J-G. S., in preparation). It depends on local conditions such as vertical PHMP flux, local geometry of the spreading axis and other tectonic elements, related lithosphere spreading rate(s) and prevailing stress field and thermal field. Further, within the upper 200-300 km depth both mantle sources are composed of partially molten rocks to variable degrees; and Deffeyes’s recent hot plume model 38 does not apply directly. I believe we are dealing with flow of porous media one into another, percolation of melts and penetrated convection. Such systems have been called lithothermal systems 39 , and the concept has also been used for the Afar 8 . 中文
In the case of Iceland, the PHMP overflow about Iceland seems to be chiefly bi-directional, southward along the Reykjanes Ridge and northward along the Kolbeinsey Ridge 14,15 . This is in contrast to Morgan’s proposed uniform radial asthenospheric flows about hot plumes 9-11 . Rather, I prefer more directional and perhaps more complex asthenospheric overflows about plumes as also considered by Vogt 37 . The choice is based on morphological, hydrodynamic, rheological and thermal grounds 14,15 . Such channelling of the overflow about the plume is controlled by existing zones of lithosphere weakness, especially spreading centres (actually lines). Once a steady state is approached or reached, zones of plate divergence in the upper asthenosphere should be characteristically hotter and mass deficient because of diapiric and magmatic injection, and therefore should favour mantle flows. Further, the overflow can be affected by damming effects caused by remanent continental blocks with colder roots such as the Danakil and Aisha Horsts and African blocks or fracture zones (Greenland, for the more complex flows along the Kolbeinsey Ridge relative to the Reykjanes Ridge). On this basis I suggest a three directional star-like flow for the PHMP about the Afar hotspot (see Fig. 3). The branching flows are along the Danakil Depression and Red Sea Trough, toward and along the Gulf of Aden, and also (but less vigorously) beneath the Ethiopian Rift. The spreading rate is very small and the thermal gradients less pronounced along this latter direction. The chemical gradients (Figs. 1 and 2) suggest a direct relationship between the [La/Sm] E.F. and radial distance irrespective of direction. This does not prove the flow to be uniformly radial but rather the rheological constraints prevail and the flow is directional. But the data of Fig. 2 do require, in conjunction with the mixing model proposed 14 , that the flow intensity be about the same along the two principal directions of flow (Fig. 3), and vary chiefly as a function of distance from the hot plume centre. I conclude that the PHMP flow about the Afar is not vigorous enough to feed along the entire length of plate divergence the Danakil Depression–Red Sea Axial Trough and Gulf of Aden (no data are available as yet along the Ethiopian Rift). Where PHMP deficiency occurs, accreting material rising at ridge axis is complemented by the DLVL source. This mantle source typically feeds mid-oceanic ridge segments far away from any hot plume interference. 中文
Fig. 3. Afar Triangle showing the three main zones of crustal attenuation also corresponding to preferred mantle plume flow direction about its centre in the upper asthenosphere. Length of arrows roughly proportional to intensity of flow. Estimate of the plume diameter cannot be made for lack of data on the Danakil Depression (see ref. 15 for Iceland). Hatched area 1 refers to Erta Ale Volcanic Range, and 2 to Alayta Volcanic Range 23 .
Along transitional segments of spreading ridges or diverging lithospheric blocks, both mantle sources will rise as diapirs, decompress and partially melt. The primary melts mix by coalescence, as well as in elongated magma chambers accumulating at the base of the crust beneath ridge crests or troughs. These magma chambers are ready to feed intermittently volcanic eruptions above 40 . Such shallow depth volcanism will give rise to tholeiitic volcanism of a transitional type isotopically and in terms of trace element composition, provided the plate divergence is sufficiently rapid to set up an oceanic dynamical and thermal regime of rapid diapiric injections 13 . This, of course, corresponds to Gass’s 8 third stage of volcanism with the distinction, however, that near the Afar such volcanism takes source from the PHMP material and not from the DLVL, just as beneath Iceland and for other hot plumes 14 . 中文
Geophysically, the proposed direction of the PHMP flow about the Afar hot plume should correspond to zones of crustal attenuations and oceanization, to a degree depending on the intensity of the flow and local spreading conditions. Further, the general area where the plume rises should be underlain by anomalous mantle to some considerable depth just as beneath Iceland 41 . Makris et al . 42 interpret gravity field survey over the Afar to suggest just that. According to their model, the Afar region is underlaid by 3.25 g cm –3 material, and along the Danakil Depression the crust is very strongly attenuated and partly oceanized; along the Wonji Fault Belt toward the Ethiopian Rift, the crust is slightly attenuated and most probably continental. This evidence is in excellent agreement with the prediction based on rare earth consideration. Unfortunately no gravity data are available from the hot plume centre toward the Gulf of Aden, but the available seismic data are also consistent with the rare earth inferences. Seismic profiles in the area 43 suggest that the part of the Gulf of Aden characterized by anomalously low mantle velocity extends toward the west. Near the Gulf of Tadjura the shallowing of the sea seems to be accompanied by a thickening of the crust and velocity-depth sections very similar to that derived by Palmason for Iceland 44 . Thus the Afar–Gulf of Aden plume-ridge system is analogous to the Iceland–Reykjanes system seismically 44-45 as well as chemically. 中文
Other geophysical and tectonic data also focus toward a tri-directional shallow asthenosphere flow pattern. These are the direction and types of magnetic anomaly lineaments 4,5 and the so-called megastructures 46 . 中文
How can the plume centre be located? The centre of reference from which radial distances were measured on Fig. 2 was taken arbitrarily at 11°15′N, 41°25′E, northwest of Lake Abbe, approximately at the centre of the Afar Triangle. This choice was made at the triple junction of the west–southwest Gulf of Aden trend, the south–southeast extension of the Danakil Depression (characterized by the Erta Ale and Alayta Ranges disposed en-echelon fissures of tholeiitic basalt outpouring) 23 , and finally the approximate extension of the Ethiopian African Rift along the Wonji Fault Belt (Fig. 3). I do not wish to imply that this point necessarily represents the centre of the hot plume. The exact position of these coordinates is not critical to the argument presented here. The plume centre could easily be displaced by some 100 km, without altering the conclusions reached from Fig. 3. Later, I found that this point coincides with the shaded area of Fig. 3 of Barberi et al . 23 , characterized by centres of thermal activity, and where evidence for earlier submarine volcanism exists 47-48 . It also corresponds to the area where the Wonji Fault Belt seems to die out and lose its identity 47 , and where direction of faulting and magnetic anomaly direction change from north–northwest to nearly west 4,5 . Finally, it corresponds to Mohr’s proposed quaternary triple junction near Lake Abbe (11°N, 41°30′E, ref. 49). It seems, then, that the synthesis of entirely independent structural, tectonic, magnetic and geochemical data all focus toward the presence of the hot plume centre in this area. 中文
Not only chemical but also mineralogical parameters seem to be correlative with distance from hot plume centres along mid-ocean ridges. The unusual presence of clinopyroxene phenocrysts in the Gulf of Aden pillow basalts 21 in contrast to mid-ocean ridge basalts which are usually rich in plagioclase phenocrysts, and the remarkable and regular increase of pyroxene phenocryst abundance over plagioclase toward Iceland along the Reykjanes Ridge 18 , are examples. A petrographic inspection of pillow basalts erupted along the Mid-Atlantic Ridge over the Azores Platform, another proposed hotspot 9-11 , also contains an unusual abundance of clinopyroxenes relative to plagioclase and olivine (unpublished). All these basalts are tholeiitic and have light rare earth enriched patterns similar to basalts derived from the Iceland and Afar hot mantle plumes. The presence of clinopyroxene in submarine ridge basalt erupting near hotspots thus seems to reflect the unusual PHMP mantle composition and thermodynamic melting conditions prevailing, which both need to be investigated further. 中文
It is important to carry out geochemical, petrological and geophysical characterization of both the PHMP and the DLVL mantles. The trace element and isotopic composition of the DLVL have been indirectly but extensively inferred from the so-called low-K 2 O mid-ocean ridge tholeiites 13,32,33,50,51 erupting sufficiently far away from any hotspot. In addition Sr and Pb isotopic ratios 25,33 suggest that the DLVL must have undergone such a depletion at some considerably earlier time in the Earth’s history, presumably by previous episodes of melting and magma extraction such as of alkali basalts 53 , and/or during continental growth 54 , and differentiation of the upper few hundred kilometers of the Earth’s mantle during geological time 14 . 中文
The PHMP source is less well defined and may vary from one hot plume to another. The PHMPs rising beneath the Afar, Iceland, Hawaii, the Galapagos and Réunion hotspots seem undepleted in large ionic lithophile elements, and derived from deeper than the DLVL (refs. 14, 15). It is, of course, too early to speculate whether the plumes composed of relatively more primordial mantle originate from a single deep earth layer of worldwide extent (homogeneous or inhomogeneous), or whether each plume is generated from deep but localized pockets of gravitationally unstable mantle. 中文
What seems more certain 13,24,25 is that such hot mantle plume source(s) have remained deep in the Earth as closed systems long enough to build up their radiogenic isotopes and heat content, before rising as plumes and transporting relatively more primordial material to the Earth’s upper zones 14 . Although Vogt 55 has recently called for global synchronism of mantle plume activity, until this suggestion is further substantiated I believe the balance of evidence suggests that the time of plume upwelling differs from one ocean to another, and from one hot plume to another, and that the intensity of activity is variable with time (J-G. S. and Noe-Nygaard, unpublished). This further complicates direct isotopic comparisons and inferences on the depth location of plume sources. 中文
I thank Mrs M. Osti and M. Zajac for help with typing and drawings. This work has been supported by the US Office of Naval Research and the National Science Foundation. 中文
( Nature Physical Science , 242 , 2-5; 1973)
J-G. Schilling
Graduate School of Oceanography, University of Rhode Island, Kingston, RI 02881
Received January 23, 1973.
References: