School of Natural Resource Sciences Queensland University of Technology MAGMATIC EVOLUTION OF THE SHIRA VOLCANICS, MT KILIMANJARO, TANZANIA By Stephen John Hayes B.App.Sc. (QUT) 2004 Supervisor: Associate Professor David A. Gust Queensland University of Technology A Thesis submitted for the degree of Master of Applied Science (Queensland University of Technology)
117
Embed
MAGMATIC EVOLUTION OF THE SHIRA VOLCANICS, MT KILIMANJARO ...eprints.qut.edu.au/15963/1/Stephen_Hayes_Thesis.pdf · MAGMATIC EVOLUTION OF THE SHIRA VOLCANICS, MT KILIMANJARO, TANZANIA
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
School of Natural Resource Sciences
Queensland University of Technology
MAGMATIC EVOLUTION OF THE
SHIRA VOLCANICS, MT
KILIMANJARO, TANZANIA
By
Stephen John Hayes
B.App.Sc. (QUT)
2004
Supervisor:
Associate Professor David A. Gust
Queensland University of Technology
A Thesis submitted for the degree of Master of Applied Science
(Queensland University of Technology)
KEYWORDS
Kilimanjaro, East African Rift, alkalic magmatism, petrogenesis, magma
evolution, fractional crystallisation
I
ABSTRACT
Mt Kilimanjaro, Africa’s highest mountain (5895m), is a large, young (<1.6Ma)
stratovolcano at the southern end of the East African Rift, in northern Tanzania.
Consisting of three distinct volcanic centres, Shira, Mawenzi and Kibo, Shira
contains the highest proportion of mafic rocks. Shira samples are strongly silica
under-saturated rocks, ranging from picro-basalt, to nephelinite and hawaiite
(Mg numbers (Mg #) ranging from 77.2–35.5). Phenocrysts constitute up to
55% of some samples, and include aluminous augite (often containing
Figure 1. East African Rift and location of Mt Kilimanjaro 3 Figure 2. Kilimanjaro regional geology, lava correlation and Shira cross section 4 Figure 3. Active versus passive rifting models 7 Figure 4. Hypothetical East African Rift model 7 Figure 5. Principle igneous centres of the East Africa Rift 8 Figure 6. Kilimanjaro geology, topography and sample locations 10 Figure 7. Shira geology, topography and sample locations 12 Figure 8. Sr and Ba comparisons for ICP-AES and LA-ICP-MS 16 Figure 9. Olivine microprobe results 20 Figure 10. Clinopyroxene microprobe results 22 Figure 11. Feldspar microprobe results 23 Figure 12. Spinel microprobe results 24 Figure 13. Feldspathoid microprobe results 25 Figure 14. Olivine LA-ICP-MS results 27 Figure 15. Clinopyroxene LA-ICP-MS results 28 Figure 16. Feldspar LA-ICP-MS results 30 Figure 17. Spinel LA-ICP-MS results 31 Figure 18. Major element analysis results 38 Figure 19. Trace element analysis results 39 Figure 20. Chondrite normalised REE and primitive mantle normalised
multi-element spider diagrams 40 Figure 21. Total alkalis silica and silica saturation diagrams 41 Figure 22. Mg number versus CaO/Al2O3 and K2O versus P2O5 42 Figure 23. Normative plots distinguishing groups 42 Figure 24. Fractional crystallisation paths of Shira samples 44 Figure 25. Fractionation vectors produced from the removal of olivine,
clinopyroxene, spinel and plagioclase 45 Figure 26. KSH08-KSH03-K679-KSH02 fractionation model results 50 Figure 27. K2225-K803 fractionation model results 52 Figure 28. KSH01-K802 fractionation model results 52 Figure 29. K813-K820-K825 fractionation model results 54 Figure 30. K361-K897-K894 fractionation model results 55 Figure 31. Zr/Hf and Nb/Ta versus Mg number diagrams 57 Figure 32. Magma mixing model path 58 Figure 33. Backscanned image of KSH05 clinopyroxene 1, with LA-ICP-MS
results showing oscillatory zonation 59
Figure 34. Magma mixing results normalised to KSH11 60 Figure 35. Chondrite normalised REE and primitive mantle normalised
multi-element spider diagrams for magma mixing model 61 Figure 36. Paths produced from addition of equilibrium olivine 65 Figure 37. Chondrite normalised REE and primitive mantle normalised
IV
multi-element spider diagrams for equilibrium olivine addition 65
Figure 38. Compilation of primitive samples plotted on CaO versus Mg number 66 Figure 39. Paths produced from addition of clinopyroxene and olivine 67 Figure 40. Chondrite normalised REE and primitive mantle normalised
multi-element spider diagrams for addition of clinopyroxene and olivine 67 Figure 41. MELTS models of ‘primary’ fractionation corrected magmas
using pressure = 0.5kb, H2O = 0.2% and fO2 = QFM. 69 Figure 42. REE and primitive mantle normalised multi-element spider diagrams
of reverse modal equilibrium batch melting models 73 Figure 43. REE and primitive mantle normalised multi-element spider diagrams
of reverse non-modal equilibrium batch melting models 73 Figure 44. Primitive mantle normalised multi-element spider diagram of forward
Figure 46. Model of the genesis and evolution of Mt Kilimanjaro and the Shira region 77
LIST OF TABLES
Table 1. Analytical precision of EDS microprobe results 15 Table 2. Analytical precision of ICP-AES major element results 16 Table 3. Shira volcanic rock group classification 18 Table 4. Samples analysed by EDS microprobe and LA-ICP-MS 18 Table 5. Representative microprobe analyses 19 Table 6. Group 1 geochemical results 34 Table 7. Group 2 geochemical results 35 Table 8. Group 3 geochemical results 36 Table 9. Group 4 geochemical results 37 Table 10. Partition coefficients used in modelling 46 Table 11. Microprobe results used in modelling 47 Table 12. Fractional crystallisation models 49 Table 13. Magma mixing model 60 Table 14. Compositions of ‘primary’ fractionation corrected magmas 68
LIST OF APPENDICES
Appendix A. Fractional crystallisation models 92 Appendix B. Magma mixing model 97 Appendix C. Primary magma compositions 99 Appendix D. Reverse partial melting models 101 Appendix E. Forward partial melting models 103
V
STATEMENT OF ORIGINAL AUTHORSHIP
The work contained in this thesis has not been previously submitted for a degree or
diploma at any other higher education institution. To the best of my knowledge, this
contains no material previously published or written by another person except where
due reference is made.
Signed:…………………………………..
Date:…………………………………..
VI
ACKNOWLEDGMENTS
Several people have provided invaluable assistance through the duration of my
project which I would sincerely like to thank. Firstly, I would like to gratefully
acknowledge the time, work, fieldwork assistance and financial assistance (through
numerous jobs) of my supervisor Associate Professor David Gust.
I would also like to acknowledge Dr Michael Carpenter and Dr Sally Gibson
(Cambridge University) for providing access to samples from the Sheffield University
Kilimanjaro rock collection. Furthermore, thankyou to Professor Richard Arculus
(Australian National University) for performing trace element analyses of all
Kilimanjaro samples and for his assistance when I went to Canberra for LA-ICP-MS
analysis.
Thankyou to the QUT technical staff, in particular Bill Kwiecien and Loc Duong, and
to Dave Purdy for showing me the ropes on all the machines at QUT. Thanks must
also go to Franco (Kilimanjaro guide) and our porters for not leaving us stranded on
the mountain or telling the national parks about our “souvenir rocks”, and also to
Luke for his endless supply of music and entertainment.
Finally, special thanks must go to my family, and Therese for their support and
encouragement and for tolerating me over the last two years.
1
INTRODUCTION
The petrogenetic modelling of primitive mafic, alkalic rocks provides valuable
information on large parts of the earth’s interior which are otherwise
inaccessible. When combined with geophysical studies, the geochemical
studies of alkalic rocks may hold the key to understanding the composition and
evolution of the Earth’s mantle (Spath et al., 2001). Geochemical and
mineralogical studies yield valuable information concerning the magmatic
evolution and magma chamber dynamics of melts once segregated from their
source. Extensive geochemical and mineralogical studies have been
performed on numerous tectonic settings, including arc volcanics, ocean island
volcanics and continental flood basalts (e.g. MacDonald et al., 2001). However,
the source, production and evolution of large mixed-association, off rift axis
stratovolcanoes remains enigmatic within the studies of continental rifting and
has only recently been addressed (e.g. Spath et al., 2001). This thesis
investigates the geochemistry and mineralogy of the Pliocene to Pleistocene
Shira Volcanics, Mt Kilimanjaro to determine the processes responsible for their
evolution as well as speculate on their source and conditions of partial melting.
Continental rifting, in which voluminous alkalic magmatism is commonly
associated, has been the subject of geochemical investigations for decades
Olivine phenocrysts occur in all groups and are compositionally homogenous or
normally zoned. Olivine phenocrysts from Group 1 samples have higher
forsterite contents than those from other groups; phenocryst cores (Fo85-92) are
compositionally homogenous, with thin rims that have dramatically lower
forsterite content (Fo77-80). Group 2 samples vary from Fo75 cores to ~ Fo40 rims.
Olivine phenocrysts from Group 3 and 4 samples are more homogenous (Fo65
to Fo60 and Fo55 to Fo48, respectively). Compositions of groundmass olivine
varies, but is always less than the least forsteritic phenocryst rim composition;
groundmass olivine in Group 3 is considerably lower than the rim compositions
of the phenocrysts. Olivine phenocryst compositions are less than
compositions calculated to be in equilibrium with the bulk rock (Figure 9), with
groundmass analyses considerably lower.
30 40 50 60 70 80
10
30
50
70
90
Fo
%
Mg number of rock
Figure 9. Comparison of forsterite values of phenocryst and groundmass olivines withMg number calculated from bulk rock analysis. The dashed line represents the olivinecomposition in equilibrium with the bulk rock using K = 0.3. (arrows show typical
Fo change from core to rim where applicable).D(Ol/Liq)
Fe/Mg
Group 1 Phenocryst
Group 1 Groundmass
Group 2 Phenocryst
Group 2 Groundmass Group 3 Groundmass
Group 3 Phenocryst Group 4 Phenocryst
Group 4 Groundmass
rim
-co
re
22
Clinopyroxene
Clinopyroxene phenocrysts commonly occur in all four petrographic groups.
Phenocrysts range from large subhedral crystals with resorption rims,
numerous melt, apatite, olivine and spinel inclusions and complex zonation
patterns to small, unzoned or normally zoned euhedral crystals with few
inclusions. Most clinopyroxenes are aluminium diopsides (Figure 10a).
Cpx 4a (0.93 weight percent)); TiO2 contents reach 4.6 weight percent in some
rim / groundmass analyses. The majority of samples that plot in the “others”
quad (Figure 10b) are rim or groundmass analyses. According to the
boundaries defined by Aoki and Kushiro (1968) on an octahedral aluminium
(AlM1) versus tetrahedral aluminium (AlT) plot (Figure 10c), all clinopyroxene are
of low pressure origin.
Group 1 samples vary from chromian aluminium augite to ferrian sub-silic
aluminium wollastonite. Groundmass analyses dominantly plot towards the
diopside/wollastonite end of this band. Group 3 and 4 samples show similar
trends however span much smaller compositional bands, whilst Group 2
phenocrysts and groundmass compositions overlap.
Feldspar
Feldspar phenocrysts are well developed only in Group 2 samples. These
phenocrysts are sub- to euhedral, coarse to very fine-grained with a variety of
zonation patterns. Most phenocrysts are unzoned or normally zoned; An
content varies from An70 to An45. Some larger Group 2 phenocrysts show
oscillatory or reverse zonation. Group 1 and 3 samples contain sub- to
euhedral micro phenocrysts that are normally zoned from An80 to An60 and An70
to An50, respectively. Groundmass plagioclase generally overlaps phenocryst
rim compositions and extends to lower An contents. Group 2 samples contain
sanidine in the groundmass.
23
NaM2
AlT
CaTSJD
NaTi TiAlNaTiAl
Ti
enstatite ferrosilite
Hedenbergitediopside hedenbergite
Group 1 Phenocryst
Group 1 Groundmass
Group 2 Phenocryst
Group 2 Groundmass Group 3 Groundmass
Group 3 Phenocryst Group 4 Phenocryst
Group 4 Groundmass
Figure 10. A) Microprobe analyses of phenocryst and groundmass clinopyroxenesfrom Shira samples presented in the Mg-Ca-Fe (enstatite-wollastonite-ferrosilite) triangle(arrows indicate general trend from core to rim, circles indicate groundmass compositionregions) and
.
Samples have been split into the four petrographic groups based on phenocryst assemblagesas discussed in the text. C) Plot of octahedral aluminium (Al ) versus tetrahedral aluminium(Al ) in clinopyroxenes, and the pressure fields of Aoki and Kushiro (1968).
M1
T
B) “others” quadrilateral (JD = jadeite, CaTS = Ca-Tschermaks, TiAl = Ti-Alaugite, NaTiAl = Na-Ti-Al augite, NaTi = Na-Ti augite (Ti end member is “fictive” CaTiAl O )2 6
Figure 11. Microprobe analyses of phenocryst and groundmass feldspars from Shira samplesplotted as proportion anorthite-albite-orthoclase (An-Ab-Or), with arrow showing the general trendfrom core to rim, and circles showing the groundmass composition regions.
core
Spinel
Analyses of spinels are separated into inclusion, phenocryst and groundmass
phases. Inclusions occur dominantly in Group 1 olivine (one inclusion was
found in a Group 4 olivine). These spinel inclusions are dominantly
magnesiochromite spinels (Figure 12). Spinel phenocrysts and groundmass
phases in all groups are similar in composition, being dominantly titaniferous
magnetites.
25
Figure 12. Microprobe analyses of phenocrysts (including inclusions) and groundmassspinels of Shira samples. Samples have been plotted using Mg numbers (Mg/(Mg + Fe ))and Cr numbers (Cr/(Cr+Al)). Groundmass samples plot very close to the lower left cornerof each diagram or not at all due to 0% Mg or Cr (magnetite / titaniferous magnetite).
2+
Cr/
Cr+
Al
Mg/Mg+Fe2+
00 10 20 30 40 50 60 700
20
40
60
80
10 20 30 40 50 60 700
20
40
60
80
Cr/
Cr+
Al
Mg/Mg+Fe2+
0 10 20 30 40 50 60 700
20
40
60
80
Cr/
Cr+
Al
Mg/Mg+Fe2+
0 10 20 30 40 50 60 700
20
40
60
80
Cr/
Cr+
Al
Mg/Mg+Fe2+
inclusions
inclusion
Group 1 Inclusion
Group 1 Groundmass
Group 2 Phenocryst
Group 2 Groundmass
Group 4 Phenocryst
Group 4 GroundmassGroup 3 Groundmass
Group 3 Phenocryst
Feldspathoid
Nepheline phenocrysts occur in Group 3 and 4 samples; groundmass nepheline
is also present in these as well as in one Group 2 sample (K802). Nepheline
compositions of Group 3 samples have higher nepheline components (Ne70-80)
than Group 4 samples (Ne60-65) (Figure 13). Groundmass and phenocryst
nepheline compositions overlap.
26
50 60 70 800
10
20
Ks
Ne
Group 2 Groundmass
Group 4 Phenocryst
Group 4 GroundmassGroup 3 Groundmass
Group 3 Phenocryst
Figure 13. Nepheline phenocryst and groundmass analyses from Shira samples plotted as
percent nepheline (Ne) versus kalsilite (Ks).
Laser Ablation Results
Samples for LA-ICP-MS (Table 4) were all analysed by EDS prior to LA-ICP-MS
analysis in order to reduce data and gain quantitative results. Results are
presented with respect to phenocryst type, and their respective sample names.
Olivine
Olivine shows consistent trace element concentrations for almost all core to rim
traverses (Figure 14), despite a significant decrease in forsterite content at the
phenocryst rim. KSH05 and K2225 (Group 1) olivine phenocrysts have similar
forsterite contents (Fo80-90) and very similar trace element concentrations (Ni ~
2000ppm, Cr ~ 250-400ppm, V ~ 5ppm and Mn ~ 1500ppm). KSH03 (Group 1)
olivine has slightly lower forsterite content (Fo75-82), and significantly lower Ni (~
1000ppm) and Cr (~ 50ppm), but higher V (~ 7-9ppm) and Mn (~ 2500ppm).
KSH01 (Group 2) olivine has the lowest forsterite contents (Fo72-55) and much
27
lower Ni (178-379ppm) and Cr (1.8-20ppm), yet the highest V (10-33.16ppm)
and Mn (4500-7000ppm) concentrations of all groups.
Clinopyroxene
Traverses of clinopyroxene phenocrysts yield a range of zonation patterns from
relatively unzoned to complex and oscillatory zoned. Results are presented as
a series of chondrite-normalised (Sun & McDonough, 1989) REE diagrams,
with an inset diagram of Mg number (Mg #) and Sc variation from core to rim for
comparison (Figure 15). REE diagrams show smooth enriched curves,
increasing in degree of enrichment from La to Nd, then decreasing from Nd to
Lu. Many samples also show slight positive Gd anomalies. In general, Mg #
and Sc concentrations show opposing trends (increasing Mg # versus
decreasing Sc); low Mg #’s and high Sc concentrations correlate with greater
REE enrichment. Many samples show distinct steps in REE enrichment (i.e.
Figure 15, KSH05 cpx 2, 4 and 5) in which there is a drastic increase in REE
enrichment over a small increase in distance from the phenocryst core.
REE contents in individual clinopyroxene phenocrysts can vary by up to a factor
of 10, however in most cases, variation is restricted to a factor of ~3. The
overall degree of REE enrichment in clinopyroxene increases from Group 1
(samples KSH05, KSH03 and K2225) to Group 2 (KSH01) and Group 3 (K811).
28
Sc V Cr Mn Ni Fo%
Figure 14. Results of LA-ICP-MS core to rim analyses of 3 olivine phenocrysts from
each sample KSH05, KSH03, K2225 and KSH01 showing the variation of Sc, V, Cr,
Mn and Ni, along with the EDS determined forsterite content.
Three plagioclase phenocrysts from sample KSH01 were analysed, and are
presented in order of decreasing anorthite content (determined by EDS
microprobe analyses) (Figure 16a). Only minor changes in trace element
concentrations were noted between each analysis. Chondrite normalised (Sun
& McDonough, 1989) REE diagrams (Figure 16b) show decreases in degree of
enrichment from La (42 x chondrite) to Yb (0.14 x chondrite), with distinct
positive Eu anomalies (up to 22 x chondrite). Degree of enrichment increases
with decreasing An %.
Spinel
Due to the small size of spinel phenocrysts (samples KSH01 (Group 2) and
K811 (Group 3)) and inclusions (sample KSH05 (Group1)), traverses were
unable to be conducted. Analyses have instead been plotted against
decreasing Mg #’s as determined through EDS microprobe analyses in order to
show trace element variations (Figure 17). Distinct changes are observed with
the most notable being decreases in Cr and Ni, yet increases in Ti, Mn, V, Zr
and Nb. Zr/Nb is lowest in K811 analyses (0.56-0.61), and increases in KSH05
(0.93-1.33) and KSH01 (1.71).
30
KSH01 Plagioclase LA-ICP-MS Results
1.00
10.00
100.00
1000.00
10000.00
Pl7aPl7bPl1a
Analyses
Lo
g(p
pm
)
40.00
60.00
80.00
An
ort
hit
e%
(ED
S)
Ti Mn Ga Sr Ce An%
Figure 16. A) LA-ICP-MS analyses of Ti, Mn, Ga, Sr, Ce and An% in plagioclase phenocrystsfrom sample KSH01. Samples have been plotted in order of decreasing anorthite content asdetermined through EDS microprobe analysis. B) Chondrite normalised (Sun & McDonough,1989) REE diagram of analysed plagioclase phenocrysts.
KSH01 Plagioclase LA-ICP-MS Results
0.10
1.00
10.00
100.00
1000.00
La Ce Nd Sm Eu Gd Dy Er Yb Lu
REE
Ch
on
dri
teN
orm
ali
sed
RE
E
(Su
n&
McD
on
ou
gh
,1989)
Plag 1a Plag 7b Plag 7a
A)
B)
31
Figure 17. LA-ICP-MS analyses of spinel inclusions (KSH05 samples) and phenocrysts (K811and KSH01 samples). Samples have been plotted in order of decreasing Mg number asdetermined through EDS microprobe analysis.
Spinel LA-ICP-MS Results
1.00
10.00
100.00
1000.00
10000.00
100000.00
1000000.00
KS
H01-S
p3
K811-S
p2a
K811-S
p3a
K811-S
p4a
KS
H05-S
p3a
KS
H05-S
p4a
Analyses
Lo
g(p
pm
)
0.00
20.00
40.00
60.00
Mg
nu
mb
er
(ED
S)
Ti Mn V Cr Ni Zr Nb Mg#
Geochemical Results
All Shira samples were analysed for both major and trace elements. Results
are presented with respect to petrographic groups (Tables 6, 7, 8 and 9) and
graphically in Figures 18, 19 and 20. Mg #’s (Mg/Mg+Fe2+) were adjusted to a
FeO ratio of 0.85 (FeO/Fe2O3+FeO). CIPW normative mineralogy was
calculated using IGPET (Igpet32) petrologic software (Terra Softa Inc.).
Samples are classified using the total alkalis-silica (TAS) diagram (Le Bas et al.,
1986).
The Shira volcanic rocks are all strongly alkalic, ranging from nephelinite to
picro-basalt, basanite and trachybasalt (Figure 21) and are all nepheline
normative; Mg #’s vary from 77 to 36. The Shira samples have a limited range
in SiO2 content (40.46 wt % to 49.31wt %), a broad range in MgO content
(16.51wt % to 3.11wt %) and Al2O3 content (8.35wt % to 17.72wt %). CaO
abundances (15.76wt % to 7.09wt %) and CaO/Al2O3 (molecular proportions)
(0.73 to 3.71) have positive correlations with Mg # (Figure 22). Abundances of
Fe2O3, TiO2, K2O, P2O5, Na2O, Sr and Ba all show negative correlations with
32
Mg# (Figure 18), however both Fe2O3 and TiO2 show inflections at
approximately Mg# 45.
Groups identified on the basis of petrographic character are easily discernible
on most major element and trace element graphs (Figures 18, 19 and 21) and
normative mineralogy (Figure 23). Group 1 samples (picrites, basanites and
alkali-olivine basalts) are easily separated due to their much higher Mg #’s and
CaO contents, and much lower Al2O3, Na2O, P2O5 and K2O abundances (Figure
18). Group 1 samples generally show low incompatible element concentrations
(Figures 19 & 20), relatively high normative plagioclase compositions (Figure
23b) and low normative albite contents (Figure 23c).
Group 3 (nephelinites and basanites) samples, although having similar Mg #’s
to Group 2 and 4 samples, are distinguished by their high P2O5 and MnO, and
low SiO2 content (Figure 18). Group 3 samples also have higher CaO contents
and CaO/Al2O3 ratios (Figure 22a) at comparable Mg #’s to Group 2 and 4
samples, as well as higher Sr, Ce, Yb, Zr, Nb and Ta abundances (Figure 19).
Group 3 samples have the highest normative nepheline contents (Figure 23a),
high normative plagioclase compositions (Figure 23b), and low albite contents
(Figure 23c) at comparable Mg #’s to Groups 2 and 4.
Group 2 and 4 samples (trachy-basalts and basanites) cover broad, but similar
chemical composition ranges (Figures 18, 19, 22 and 23). Group 2 samples
have lower CaO/Al2O3 (Figure 22a), than Group 3 and 4 samples of similar Mg
#. Group 2 samples generally contain slightly higher Sr, Ba, Rb, Ce, Yb, Zr, Hf,
Nb and Ta contents at comparable Mg #’s (Figure 19) than Group 4 samples,
whereas the majority of Group 4 samples contain higher normative plagioclase
compositions (Figure 23b) and lower normative albite content (Figure 23c) than
Group 2 samples of comparable Mg #’s.
Although broad geochemical trends are apparent over the entire range of Shira
samples (negative trends for incompatible elements (i.e. Sr, Ba, REE, Zr, Hf, Nb
& Ta) and positive trends for Cr, Sc and V), smaller intra-group trends are also
apparent, with some intra-group trends opposing the broader Shira trend.
Group 2 samples show positive trends for Nb and Ta, whilst Groups 1, 3 and 4
show negative trends. Similarly, Groups 1 and 3 show positive Rb trends,
whilst Groups 2 and 4 show negative trends. Hf shows a negative correlation
33
for Groups 1 and 4, yet a positive correlation for Group 3 samples, and broad
scatter of Group 2 samples.
Chondrite-normalised REE patterns of Shira samples are light-REE enriched.
La concentrations range between 100 and 400 times chondritic levels, with Lu
concentrations approximately 10 to 20 times chondritic levels. Ce/Yb ratios
vary from 36 to 70. REE patterns shallow towards the heavy REE, with Ce/Sm
values between 9.9 and 16.3 and Sm/Yb values of between 3.17 and 4.41.
Chondrite-normalised REE patterns are smooth and near parallel (Figure 20),
with very minor Eu anomalies observed in only five Group 2 samples, three
Group 3 samples, and one Group 4 sample. The degree of REE enrichment
increases from Groups 1 to 3, with Group 4 covering a broader range. Groups
have distinct multi-element spider diagram trends when normalised against
primitive mantle values (Figure 20) (Sun & McDonough, 1989). All groups show
distinct K depletions, but uncharacteristically, Pb enrichments (not as
pronounced in Group 3 samples) (Figure 18). Group 1 and 2 samples have
similar characteristics, with Group 1 tending to be less enriched than Group 2.
Positive anomalies are shown for Pb, Nb, Nd, and Ti relative to neighbouring
elements, and negative anomalies are shown for P, K and Zr in Groups 1 and 2,
with larger anomalies in Group 1 than Group 2. Group 3 multi-element spider
diagrams are similar to Group 2, with larger negative K anomalies, but smaller
Pb anomalies. The multi-element diagram for Group 4 is very similar to that of
Group 1.
34
Table 6. Geochemical results of Group 1 samples (BSN=basanite,
Figure 18. ICP-AES major element resultsfor Shira samples in weight percent plottedagainst Mg numbers.
Group 1
Group 2
Group 3
Group 4
Shira Samples
39
40 50 60 70
200
600
1000
1400
Sr
Mg/(Mg+Fe2+)
40 50 60 70
200
400
600
800
1000
Ba
Mg/(Mg+Fe2+)
40 50 60 700.0
0.5
1.0
1.5
Cs
Mg/(Mg+Fe2+)
40 50 60 700
50
100
Rb
Mg/(Mg+Fe2+)
40 50 60 700
40
80
120
160
Ce
Mg/(Mg+Fe2+)
40 50 60 701
2
3
Yb
Mg/(Mg+Fe2+)
40 50 60 700
100
200
300
400
Zr
Mg/(Mg+Fe2+)
40 50 60 700
2
4
6
8
Hf
Mg/(Mg+Fe2+)
40 50 60 70 800
50
100
150
Nb
Mg/(Mg+Fe )2+
40 50 60 70 80
2
6
10
Ta
Mg/(Mg+Fe )2+
40 50 60 70 80
Mg/(Mg+Fe )2+
40 50 60 70 80
Mg/(Mg+Fe )2+
Figure 19. LA-ICP-MS trace element results (Sr, Ba, Cs, Rb, Ce, Yb, Zr, Hf, Nb & Ta) for Shirasamples in parts per million (ppm) plotted against Mg numbers.
Group 1
Group 2
Group 3
Group 4
Shira Samples
40
1
10
100
La Ce Pr Nd Pm Sm Eu Gd Tb Dy Ho Er Tm Yb Lu
Rock/Chondrites
1
10
100
La Ce Pr Nd Pm Sm Eu Gd Tb Dy Ho Er Tm Yb Lu
Rock/Chondrites
1
10
100
La Ce Pr Nd Pm Sm Eu Gd Tb Dy Ho Er Tm Yb Lu
Rock/Chondrites
1
10
100
La Ce Pr Nd Pm Sm Eu Gd Tb Dy Ho Er Tm Yb Lu
Rock/Chondrites
1
10
100
1000
CsRbBaTh U Nb K LaCePb Pr Sr P Nd ZrSmEu Ti Dy Y YbLu
Rock/Primitive Mantle
1
10
100
1000
CsRbBaTh U Nb K LaCePb Pr Sr P Nd ZrSmEu Ti Dy Y YbLu
Rock/Primitive Mantle
1
10
100
1000
CsRbBaTh U Nb K LaCePb Pr Sr P Nd ZrSmEu Ti Dy Y YbLu
Rock/Primitive Mantle
1
10
100
1000
CsRbBaTh U Nb K LaCePb Pr Sr P Nd ZrSmEu Ti Dy Y YbLu
Rock/Primitive Mantle
Figure 20. REE and multi element spider diagrams for all Shira samples (separated into
groups and normalised to Sun & McDonough, 1989 chondrite and primitive mantle respectively)
Group 1
Group 2
Group 3
Group 4
41
Figure 21. A)Total alkalis silica (TAS) diagram of all Kilimanjaro samples showing the
classification scheme of Le Bas , (1986) with inset showing sample numbers.
B) Plot of normative plagioclase composition (AN=(anx100)/(ab+an)) versus normative
nepheline and hypersthene for all Shira samples. The centre line represents the plane
of silica undersaturation, whereas the shaded band represents the approximate trace of
the plane of silica saturation (e.g. Best & Brimhall, 1974), showing that all Shia samples
and albite (C) versus Mg number in order to show normative differences
between groups.
Figure 22. A) Plot of Mg number versus CaO/Al O for Shira samples showing the broad
negative trend (decrasing with decreasing Mg number) and differing
values for groups 1 to 4. B) Plot of K O versus P O for Shira samples showing the regions
in which groups 1 to 4 plot.
2 3
2 2 5
CaO/Al O CaO/Al O2 3 2 3
40 50 60 70 800
1
2
3
4
Ca
O/A
lO
23
Mg/(Mg+Fe2+)
40 50 60 700
10
20
30
ne
Mg/(Mg+Fe2+)
40 50 60 7030
50
70
90
%AN
Mg/(Mg+Fe2+)
0 1 2 30.0
0.5
1.0
K O2
PO
25
Group 1
Group 2
Group 3
Group 4
Shira SamplesShira Sample
A) B)
A) B)
40 50 60 70 800
10
20
30
ab
Mg/(Mg+Fe2+)
Group 1
Group 2
Group 3
Group 4
Shira SamplesC)
43
DISCUSSION
Four distinct groups based upon petrographic characteristics are identified as
composing the Shira Volcanics of Mt Kilimanjaro. In general, these groups also
have distinctive petrographic, geochemical and geographic characteristics.
Group 1 (East Shira Hill and Shira Cathedral areas) are high-Mg strongly phyric
basanites, picrobasalts and alkali olivine basalts. Group 2 samples (from the
same area) are trachybasalts and basanites, Group 3 samples (nephelinites)
are from the main Shira Ridge near Klute and Kente peaks, and Group 4
samples (basanites and trachybasalts) are from Platzkegel.
Understanding the relationship between these different groups may help to
understand the petrogenesis of the Shira volcanic rocks and magma chamber
dynamics beneath Mt Kilimanjaro. The petrogenetic relationship of each group
is developed through modelling of magmatic processes such as fractional
crystallisation, mixing and assimilation, and ultimately, speculating on their
source composition and mineralogy, and melting processes.
Fractional Crystallisation Models
Major and trace element trends indicate that the diversity of the Shira suite can
be explained by the process of fractional crystallisation (Figure 24) commencing
from a suite of slightly different primary magmas.
Quantitative models of fractional crystallisation for the Shira lavas are based on
major element mass balance equations using inputs of major element
geochemistry and microprobe data. These models provide a basis for trace
element calculations using the Rayleigh fractional crystallisation formula
CL/CO=F(D-1) (Allegre et al., 1977; Allegre & Minister, 1978) and compilation of
relevant partition coefficients (Table 10). Modelling has been performed through
both IGPET petrologic software and the creation of Microsoft Excel
spreadsheets for comparing calculated daughter trace element concentrations
with observed parent and daughter concentrations.
Several criteria are used for selecting suitable solutions to major element mass
balance calculations. The first criterion is that the sum of the residuals squared
be less than 0.6 (using a weighting of 0.4 for Si, 0.5 for Al and Mn and 1 for Ti,
Comment [ADG1]: You will find that I have not changed anything in the discussion yet. This is not to say that it is perfect. It does need some changes some in presentation and some in science), but you certainly have the bulk of it there. Why I haven’t written anything is that after 2 readings, I am still mulling it all over in my mind. I am not quite sure what the issues are, but I have this slightly unsettled feeling that there is another angle that needs to be thought about. I will do this thinking over the weekend and get some comment back to you on the discssion by Monday.
44
Mg, Ca, Fe, Na, K and P). The second criterion is the plausibility of the
fractionating assemblage (whether the relative abundances of fractionated
phases from the model is suitable for the observed mineralogy of the samples)
whilst the third criterion is the match between calculated daughter trace element
concentrations and observed daughter concentrations. When these criteria fail
to discriminate between models, the model that uses the least number of
fractionating phases is preferred. Comprehensive major and trace element
results are presented in Appendix A, and simplified model results are shown in
Table 12.
Potential fractionating phases are limited to observed phenocrysts
(clinopyroxene, olivine, plagioclase, nepheline and spinel). In some cases, the
compositions of the selected phase vary significantly between core and rim.
The phase chemistries used in the fractional crystallisation models for each
group are presented in Table 11. The vectors produced by fractionation of the
wide range of phenocryst compositions are shown in Figure 25 on a CaO
versus Mg # diagram. The major observations that can be made from this
graphical presentation are that fractionation of olivine drives liquid compositions
to lower Mg #’s and higher CaO, whereas fractionation of clinopyroxene drives
liquid compositions to lower Mg #’s and lower CaO. In general it takes
approximately twice the amount of clinopyroxene fractionation to decrease the
Mg # the same amount that olivine fractionation would decrease it. Variation in
olivine composition and clinopyroxene compositions create a range of liquid
1 Spath et al. , 2001 (comp of McKenzie & O'Nions, 1991; Frey et al. , 1978; 4 Lemarchand et al ., 1987
Dalpe' & Baker, 1994; Chazot et al., 1996; Adam et al., 1993; 5 Fujimaku, 1984
Le Roex et al ., 1990 and references therein; Lemarchand et al ., 1987 6 Onuma et al ., 19812 Rollinson, 1993 (comp of Arth 1976; Pearce & Norry, 1979; Green et al., 1989; 7 Villemant et al ., 1981
Schock, 1979; Fujimaku, 1984; Dostal et al ., 1983; Henderson, 1982; 8 Schock, 1979
Leeman & Lindstrom, 1978; Lindstrom & Weill, 1978; Green & Pearson, 1987) 9 Irving & Frey, 19783 Kempton et al ., 1987 10 Arth, 1976
47
Oliv
ine
Ana
lys e
s U
sed
in P
etro
gene
tic M
odel
ling
Spi
n el A
nal
yses
Use
d in
Pet
rog
enet
ic M
ode
lling
Gro
up 1
Gro
up 1
Gro
up 1
Gro
up 4
Gro
up 4
Gro
up 1
Gro
u p 3
Gro
up 3
Gro
up 4
Gro
up 4
K
SH
05 o
l 8c
KS
H05
ol 5
c K
SH
03 O
l3b
K36
1 O
l1a
K36
1 O
l1b
KS
H0 3
CrS
pK
811 M
gTiS
pK
820M
gTiS
p K
361M
gTiC
r Sp
K89
4TiS
p av
g rim
av
g c
o re
r imS
iO2
39.4
339
.33
3 8.7
234
.97
34.0
6S
iO2
0.8
0.48
0.45
0.26
0 .75
TiO
20
00
0.05
0.02
TiO
24.
4617
.12
15.2
515
.81
20.4
3A
l 2O3
00.
150
00.
22A
l 2O3
1 9.3
87.
747.
975.
166.
98Fe
O14
.02
16.5
119
.88
36.9
940
.13
FeO
43.7
666
.39
64.6
863
.45
58.8
MnO
0.1
0.39
0.29
0.72
0.99
MnO
0.06
0.61
0.51
0.32
2.5
MgO
44.7
842
.89
40.4
526
.01
22.4
MgO
9.09
3.93
4.14
1.7
0.4
CaO
0.55
0.49
0.38
0.57
0.4 2
CaO
0.02
0.03
00.
030.
12N
a 2O
00
00
0N
a 2O
00
00
0.11
K2O
00.
10.
10
0.06
K2O
0.01
0.01
0.01
00
P2O
50
00
0.14
0.18
P2O
50.
120
0.01
00
Cr 2
O3
00
00.
010
Cr 2
O3
17.1
60.
10 .
045.
380.
11T O
TAL
98.8
899
.86
99.8
299
.46
98.4
8TO
TAL
94.8
696
.41
93.0
692
.11
90.2
# of
Cat
ions
3.00
3.00
3.00
3 .01
3 .00
# of
Cat
ions
3.00
3.00
3.00
3.00
3.0 1
Fo85
82.2
78.3
55.6
49.8
Mg#
39.5
14.3
15.9
6.7
1.5
Clin
opy
roxe
ne A
naly
ses
Use
d in
Pet
roge
netic
Mo d
ellin
gP
lag
iocl
ase
Ana
lyse
s U
sed
in P
etro
g ene
tic M
odel
ling
Gro
up 1
Gro
up 1
Gro
up 3
Gro
up 3
Gro
up 4
Gro
up 4
Gro
up 4
Gro
up 4
KS
H05
cpx
8c
KS
H03
Cpx
4b
K81
1 C
p x4c
K
820
Cpx
1cK
361
Cpx
1a
K89
4 C
px2c
K36
1 G
M P
lag1
K89
4 G
M P
lag1
avg
avg
r im ri
mco
re ri
m a
vg
avg
SiO
249
.66
48.3
243
.26
46.5
646
.39
42.1
9S
iO2
52.2
252
.9Ti
O2
0.77
1.5
2.82
1.62
1.75
3.41
TiO
20.
1 40.
18A
l 2O3
4.08
6.17
9.49
6 .44
5 .63
1 0.6
5A
l 2O3
26.3
927
.5Fe
O5 .
326.
369.
818.
677.
318.
81Fe
O0.
040.
15M
nO0.
10.
010.
210.
250.
140.
07M
nO0
0M
gO15
.21
13.9
79.
3111
.74
12.7
9.31
MgO
0 .02
0C
aO23
.38
22.6
122
.47
23.3
722
.26
2 2.2
CaO
9.77
10.1
1N
a 2O
0.1
00 .
310.
240.
480.
29N
a 2O
5.17
4.75
K2O
0.12
0.11
0.17
0.1
0.1
0.04
K2O
0.5
0.49
P2O
50
0.19
00
0.18
0.41
P2O
50
0C
r 2O
31.
20.
280.
070.
020.
160.
03C
r 2O
30
0.06
TOTA
L99
.94
99.5
297
.92
99.0
197
.197
.41
TOTA
L94
.25
96.1
4#
of C
atio
ns4.
034.
0 14 .
034.
044.
044.
01#
o f C
atio
ns20
.01
19.9
0E
n43
.39
4 1.3
429
.95
3538
.63
30.7
8A
b47
.44
44.5
6W
o4 7
.93
48.0
951
.96
50.0
848
.66
52.7
5A
n49
.54
52.4
1Fe
8.68
10.5
718
.09
14.9
212
.71
16.4
7O
r3.
0 23.
03
avg
= av
e rag
e co
mpo
sitio
n fo
r pha
se in
sa m
ple,
rim
= a
vera
ge r
im c
ompo
sitio
n fo
r pha
se in
sam
ple,
cor
e =
aver
age
core
com
posi
tion
for p
hase
in s
ampl
eN
umb e
r of c
atio
ns (
# of
Cat
ions
) cal
cula
ted
on th
e ba
sis
of 4
oxy
gens
for o
livin
e an
d s p
inel
, 6 o
xyge
n s fo
r clin
opyr
oxe n
e an
d 32
oxy
g ens
for p
lagi
ocla
se
Tabl
e 11
. R
epre
sent
ativ
e m
icro
prob
e an
alys
es u
sed
in f r
actio
nal c
ryst
allis
atio
n m
odel
s.
48
Groups 1 and 2
Although different with respect to phenocryst mineralogy, samples from Group 1
and 2 define a broadly coherent trend on variation diagrams that suggest they
are probably genetically related. Three plausible fractionation paths that
include samples from both groups are considered. Path A (samples KSH08-
KSH03-K679-KSH02), path B (samples K2225-K803) and path C (samples
KSH01-K802) (Figure 24) model variation from basanite (Mg # 78) to
trachybasalt (Mg # 43). Several other samples from these two groups are
difficult to reconcile with a fractional crystallisation origin and suggest magma
mixing.
Path A (Samples KSH08-KSH03-K679-KSH02)
Examination of variation diagrams for a number of trace elements indicate that
Sr, Ba, Rb, LREE, Y, and Ga are incompatible throughout the entire
fractionation path (decreasing Mg #), and Cr and Sc behave compatibly. This
pattern of trace element variation suggests fractionation of clinopyroxene ±
olivine. Inflections in V, Ti and Fe at KSH03 may indicate the onset of spinel
fractionation. Zr, Hf and Th behave incompatibly from KSH08 to K679, yet
compatibly from K679 to KSH02, possibly due to minor accessory phase
fractionation. Nb and Ta also show inflections at this point attributable to a
change in spinel fractionation. Quantitative fractional crystallisation models that
incorporate these suggestions are presented as Table 12.
The most primitive sample, a basanite (Mg# 78 (KSH08) probably contains
cumulate olivine and clinopyroxene, as evidenced by the high modal
abundance of these phases and high forsterite content in some of the olivines.
A fractional crystallisation model that uses this sample as the parental magma
requires significant fractionation of clinopyroxene and olivine. Fractional
crystallisation is dominated by clinopyroxene and olivine with the addition of
minor Cr spinel for the parent / daughter pairs KSH03 / K679 and for K679 /
KSH02.
Trace element calculations for Rayleigh fractional crystallisation based on each
model of Path A show a high degree of consistency between calculated and
observed values (Figure 26). Slight discrepancies between observed versus
calculated REE for the model that links the most evolved magmas
49
(trachybasalts; K679-KSH02) may reflect minor accessory phase fractionation
that was not included in the calculation.
Path B (Samples K2225-K803) and Path C (Samples KSH01-K802)
The two samples that constitute Path B are distinct from those that constitute
Path A in that they are richer in CaO than Path A samples with similar Mg #.
For the parent / daughter pair (K2225/K803) the general increase in the
abundance of a wide variety of trace elements (except for the transition
elements) suggests fractionation controlled by clinopyroxene ± minor olivine
and spinel. A quantitative fractional crystallisation model for this pair (Table 12)
has a sum of squares of 0.55, and a good agreement between calculated and
observed trace element abundances (Figure 27).
Path C is represented by samples KSH01 (Mg # 46) and K802 (Mg # 38). A
fractional crystallisation model that relates these two significantly evolved
magmas requires clinopyroxene, olivine and spinel fractionation (Table 12).
This suggestion is supported by the overall incompatible behaviour of Sr, Ba,
Al, Th, Pb, Na, Er, U and K, and the compatible behaviour of Sc and V
(indicating clinopyroxene, spinel ± olivine as fractionating phases). Trace
element calculations overlap between calculated and observed values for
almost all elements except Rb and Cs (Figure 28).
50
Table 12. Fractional crystallisation models and trace element calculations for
Group 1 and 2 (Paths A, B and C), Group 3 and Group 4.
approximately 70 to 90km. This stability range is consistent with the derivation
of the Shira magmas from a spinel lherzolite source that is part of the sub
continental lithosphere (to approximately 115km depth). Thus the Shira
volcanic rocks reflect melting of an enriched sub continental lithosphere rather
than an asthenospheric source.
Isotopic studies of numerous volcanic centres in northern Tanzania (Paslick et
al., 1995) suggest ancient (>1Ga) underplating and metasomatism of the
continental lithosphere by OIB melts is responsible for enrichment of the sub
continental lithosphere. Megacryst vein studies identify metasomatism by
alkaline silicate and possibly carbonatite melts beneath northern Tanzania
(Johnson et al., 1997). No constraint is available for the timing of this
metasomatic event. Asthenospheric-sourced carbonatite melts are identified as
responsible for metasomatism of peridotite xenoliths from Olmani cinder cone,
northern Tanzania (Rudnick et al., 1993) with numerous other authors
postulating metasomatism of the lithospheric mantle in other East African
regions (e.g. Vollmer & Norry, 1983a, 1983b; Cohen et al., 1984; Rogers et al.,
1992; Furman, 1995). Fractionation-corrected Shira samples have trace
element patterns similar to OIB-melts, although slightly less enriched and with K
and Rb anomalies (Figure 45). An age of >1Ga for OIB-like continental
lithosphere enrichment (Paslick et al., 1995) implies that source enrichment
occurred long before the onset of rifting (Figure 46a). Although additional
recent enrichment by plume-derived OIB melts may have occurred (i.e. Spath et
al., 2001), the source of the Shira magmas probably acquired its enrichment
signature from ancient (>1Ga) OIB underplating and metasomatism.
It is argued that ancient metasomatism created a sub continental lithosphere
that was capable of producing highly-enriched silica under-saturated magmas
by low degrees of partial melting. The ultimate cause of this melting event
however, remains unclear. The EAR is the result of ‘active’ rifting processes in
which a plume of asthenospheric mantle ascends, and undergoes
decompression melting (Kampunzu & Mohr, 1991; Spath et al., 2001).
78
Kilimanjaro is located well away from the rift axis, thus it is expected that the
impact of the asthenosperic plume may be less. The major effect however,
would be the provision of heat (Turner et al., 1996), and probably volatiles to
the sub continental lithosphere to induce partial melting. In particular, the
introduction of volatiles such as H2O and CO2 would both lower the solidus
(Brey, 1969; Brey & Green, 1976) of the sub continental lithosphere and
stabilise amphibole, as well as providing additional incompatible element
enrichment. Thus while the Shira magmas are sourced from the sub
continental lithosphere, the ultimate reason for their generation lies in the
interaction between the upwelling asthenospheric mantle and the
subcontinental lithosphere
.1
1
10
CsRbBaTh U Nb K LaCe Sr P Nd ZrSmEu Dy Y YbLu
Rock/OIB
Figure 45. Fractionation corrected “primitive” Shira samples normalised to Sun & McDonough
(1989) OIB values.
Upon segregation from their source, melts follow structural weaknesses in the
lithosphere (Figure 46b) before ponding at shallow depths (1-5kb) in small
volume magma chambers, being subject to significant ferromagnesian
fractionation (Figure 46c). Fractionation paths result from separate,
geochemically distinct ‘primary’ magmas, in which the fractional crystallisation
of basanites (samples K2225, KSH03 and K832a) is initially dominated by
olivine, becoming clinopyroxene dominated between Mg # 45 and 60.
Nephelinite fractional crystallisation is initially subject to both olivine and
79
clinopyroxene fractionation, becoming increasingly dominated by clinopyroxene
fractionation with evolution. Magma mixing by subsequent partial melts or
convective currents within magma chambers is thought to occur episodically,
resulting in oscillatory zoned phenocrysts.
The Shira volcanic centre has since collapsed, forming the present caldera, and
been intruded by late stage dykes, sills and vent infill (Platzkegel) before
cessation (Figure 46d).
Individual nephelinite cones in intraplate volcanic provinces (i.e. southeastern
Australia, various regions in East Africa, New-Mexico etc.) may simply
represent the extrusion of short-lived, single melts. Large mixed-association
continental rift stratovolcanoes however may represent longer-lived polygenetic
volcanoes, formed through the extrusion of many successive melts.
Polygenetic volcanoes may form due to their positions relative to magmatic
conduits (i.e. major faults in the case of Mt Kilimanjaro), raised geothermal
gradients or source region differences (i.e. extent of metasomatism,
mineralogy).
80
Dehydration melting and structural weakening forming the initialdepression and magmatic conduits to the crust
(i.e. Kilimanjaro depression).
Crust
B
Rising asthenospheric plume
0km
40km
105km
SCLM
Metasomatised SCLMVariable partial melts (4-10%)
Crust
Introduction of volatiles to the sub continentallithospheric mantle (SCLM), enriched by ancient
OIB underplating and melts
A
Rising asthenosphericplume
0km
40km
105km
The collapse of Shira, followed by the eruption of Platzkegellavas and intrusion of various dykes/sills. This is in turn followed
by the cessation of the Shira and Mawenzi volcanic centresresulting in Kilimanjaro’s present morphology
Crust
SCLM
ShiraKibo
Mawenzi
D
Platzkegel
Small volume magma chambers ponding and undergoingextensive fractional crystallisation in the crust. Differring degrees ofpartial melting are responsible for nephelinites and basanites, with
cumulate samples erupted upon either emptying of magma chamber oras a result of magma mixing. Numerous partial melts (most likely of short
lifespans) create a polygenetic volcano, resulting in the formation ofKilimanjaro and its three main volcanic centres (Shira, Kibo and Mawenzi).
Crust
SCLM
ShiraKibo
Mawenzi
Magma chambers(shallow fractional
crystallisation)
Figure 46. Proposed model for the genesis and evolution of Mt Kilimanjaro starting withA) Introduction of volatile phases to a previously enriched subcontinental lithospheric mantle resultingfrom upwelling asthenosphere. B) Lithospheric faulting and low degree partial melts of the SCLMfollowing structural weaknesses, with larger polygenetic volcanoes occurring along major faults, wheremultiple partial melts may easily ascend and smaller monogenetic volcanoes occurring along minorfaults or structural weaknesses. C) Fractional crystallisation of small volume melts occurringpredominantly in small, shallow magma chamber, erupting to form Shira, Kibo and Mawenz.D) Formation of Shira caldera and Platzkegel followed by the cessation of Shira and Mawenziresulting in the present morphology of Mt Kilimanjaro (not to scale).
C
CONCLUSION
The Shira volcanic suite consists of silica-undersaturated nephelinites,
basanites, picro-basalts and hawaiites (Mg #’s ranging from 35.5 to 77.2).
Groups identified on the basis of phenocryst assemblages and textures
correlate with geographic location. Samples (East Shira Hill) contain olivine
and clinopyroxene phenocrysts + microphenocrysts of plagioclase (Group 1), or
plagioclase and clinopyroxene phenocrysts + microphenocrysts of olivine
(Group 2). Samples with high Mg #’s contain abundant cumulate clinopyroxene
and olivine (Fo92-Fo85). Group 3 samples (Shira Ridge) contain nepheline
81
phenocrysts and Group 4 samples (Platzkegel) have distinct intergranular
textures.
Trends on many geochemical diagrams identify fractional crystallisation paths
reflecting fractionation of clinopyroxene ± olivine and spinel. Complex major
and trace element zonation in clinopyroxene and feldspar phenocrysts suggest
magma mixing.
Primitive samples corrected to ‘primary’ compositions indicate low degrees of
partial melting of between 4% and 10%. Trace element abundances similar to
OIB-melts, with negative K anomalies require retention of amphibole in the
source during partial melting. REE trends are compatible with an amphibole-
bearing spinel lherzolite source. When combined with local geophysical and
mineralogical studies, this conclusion implies a source in the sub continental
lithospheric mantle. The enrichment of this source reflects ancient (>1Ga) OIB
under-plating and metasomatism. Rising asthenosphere provides a source of
heat as well as volatiles, necessary in inducing melting and stabilising hydrous
K-bearing phases.
82
REFERENCE LIST
Abdullah, M. I., (1963). Chemical and mineralogical investigations of the
titanomagnetites in reference to the petrogenesis of Kilimanjaro igneous