SOME ASPECTS OF THE PHYTOGEOGRAPHY OF
TROPICAL AFRICA1
J. P. M. BRENAN2
ABSTRACT
The area covered follows political boundaries and amounts to just over 20,000,000 km2.
The flora is estimated at about 30,000 species and 2,497 genera. The phytogeographical
affinities of the families and genera are analyzed and discussed, as is the comparative poverty
of the flora. The main phytogeographical regions and domains are briefly described. The
phytogeography and endemism of the rain forest flora are analyzed and discussed and some
comparison is made with the savanna. The distribution of Indigofera species is analyzed and
conclusions drawn. An attempt is made to discuss and assess the endemism at specific and
generic levels shown by each country in the area, with representative examples and notes on
areas of special interest. Especially high concentrations are shown in Cameroon and Gabon,
Za�re, Ethiopia, Somalia, Tanzania, and Angola. A preliminary map is given to show relative
densities of endemism.
Africa has many features in its plant life and elsewhere that combine to
make the continent outstanding if not unique among other tropical landmasses.
However, these features are not at first sight obvious: the general shape of
Africa is undistinguished if not rather dull; so is its topography-there are no
great mountain masses to compare with the Andes or Himalayas, and its highest
peak, Kilimanjaro, is an isolated volcanic cone that would be no more than "a
considerable protuberance" in those great ranges; the flora is not especially rich,
and indeed often more noteworthy for its poverty. Yet its fascination is peren-
nial, and not without cause did Richards (1973) entitle a paper "Africa, the
'Odd Man Out'."
Poverty and change are perhaps too often the keynotes. Almost everywhere
evidence of change is only too obvious: relics of forest disappearing in a sea of
grass; senile trees without offspring on the edge of semidesert; lean and hungry
livestock pursuing the retreating remains of grazing. Sometimes the pace is
slower; only occasionally is ecological change for the better. The remarkable
pictorial record of Shantz & Turner (1958) is often revealing. The flora itself
and its distribution constantly show evidence of gigantic and destructive cli-
matic changes in the past-evidence in the form of floristic poverty, spectacular
1 Some acknowledgements are made in the appropriate place in the paper (p. 459). I
would, however, also like to express my thanks especially to Mr. P. S. Green, Deputy Director
and Keeper of the Herbarium, Royal Botanic Gardens, Kew, Dr. R. Polhill and Dr. G.
Wickens for much help unstintedly given and for the opportunity of discussing problems.
I am also grateful to Mrs. J. Carter and Mr. T. A. Harwood for help with photography. To
my wife, Jean, and my daughter, Miss Mary Brenan, my especial thanks for their help and
patience during the work of preparation. I am also grateful to Dr. F. Demaret and Dr. P.
Bamps for kind permission to reproduce the map of Za�re showing vegetation boundaries
originally published in the Flore du Congo Belge et du Ruanda-Urundi. It will of course be
obvious that certain place names on this map are now superseded. Last, but far from least,
my gratitude and thanks to Dr. Peter Raven for providing the opportunity and stimulus for
this paper.
2 Director, Royal Botanic Gardens, Kew, Richmond, Surrey TW9 3AB, England.
ANN. MIssoURI BOT. GARD. 65: 437-478. 1978.
0026-6493/78/0437-0478/$04.25/0


ANNALS OF THE MISSOURI BOTANICAL GARDEN
disjunctions in distribution, islands of floristic similarity now distantly isolated,
close affinities among species now widely separated geographically, etc. The
details about how these changes occurred are still mostly unknown, but they
seem the only possible explanation of many of today's problems.
Especially during the last fifty years, investigation of the flora and vegeta-
tion of Africa, both in the field and the herbarium, has been intense. Numerous
countries have participated, and that this research has been carried out in a
spirit of cooperation rather than rivalry during the past quarter-century has in
large measure been due to the vision of those who founded the Association pour
l'�tude Taxonomique de la Flore d'Afrique Tropicale (A.E.T.F.A.T.), still flour-
ishing and with a scientific message as relevant today as when it started.
Although very much still remains to be done and many areas are still poorly
explored botanically or even quite unknown, nevertheless Africa must now rank
among the comparatively better known tropical land-areas in terms of flora
and vegetation.
Few attempts, however, have been made to assess or analyze the flora as a
whole. The available knowledge of the taxonomy and distribution of the Afri-
can flora is now probably sufficient, although much of it is still unpublished,
to allow assessments to be made of its size and richness; of its geographical
affinities; of its endemism; and even of the distribution of endemism in relation
to phytogeographical regions and countries. The speed of change in the vege-
tation of Africa is accelerating rather than diminishing, and the urgency for
action over conservation has become even more acute since the publication of
the proceedings of the A.E.T.F.A.T. Symposium held in Uppsala in 1966 on the
Conservation of Vegetation in Africa South of the Sahara (Hedberg & Hedberg,
1968). In the present paper, with its manifest omissions and imperfections, an
attempt is made to discuss and analyze some of these features of the African
flora, in the hope that in the future a more complete general assessment may
become feasible.
BOUNDAmIES OF AREA
Although tropical Africa appears in the title of this paper, the boundaries of
the area under discussion here do not strictly follow the lines of the tropics. So
much of the available data are based on areas demarcated by political rather
than biological or climatic boundaries, that it seemed necessary to accept this
situation. Accordingly, discussion took place with my cocontributors to this
Symposium, Dr. P. Goldblatt of the Missouri Botanical Garden and Professor P.
Qu�zel of the Universit� de Droit, d'�conomie et des Sciences d'Aix-Marseille,
to determine the working boundaries to the south and north respectively.
To the south, all countries to the south of Angola, Zambia, Rhodesia, and
Mozambique are covered by Dr. Goldblatt's paper. To the north, Professor
Qu�zel deals with ail territories to the north of Senegal, Mali, Niger, Chad,
and Sudan. The only predominantly tropical country thus excluded is Mauri-
tania in the north. The total area covered is just under 20,000,000 km2 (7,700,000
mi2).
438
[VOL. 65


BRENAN-TROPICAL AFRICA
SIZE AND POVERTY OF THE TROPICAL AFRICAN FLORA
The relative poverty of the tropical African flora has repeatedly been noted
as long ago as, for example, Mildbraed (1922: 103). A recent more detailed
analysis is given by Richards (1973) who considers tropical Africa as phytogeo-
graphically an "odd man out" in comparison with the other main tropical areas
of the world. He considers three differences as most significant:
(1) The relative poverty of the African flora compared with those of other
comparable tropical areas,
(2) The wide areas of distribution of numerous African species, and
(3) The poverty (or absence) in tropical Africa of certain plant groups.
Let us deal more fully with each of these three points:
In assessing the relative poverty of the tropical African flora one encounters
as a major obstacle a corresponding poverty of reliable statistics! Good (1974:
170) has been bold enough to estimate the total number of species in the tropical
African flora at 30,000. Some regional estimates of numbers of species have also
been made (Table 1).
On a basis of the approximate expected degree of endemism and of the num-
ber of species still to be discovered, Good's estimate is probably not far off the
true total. It should not be forgotten that modern critical taxonomy tends to
reduce numbers of species as much as or even more than it adds. However, the
African flora is still very far from being fully known. L�onard (1975) pointed
out that 7,478 new species were described from Africa (including Madagascar)
in the 21 years from 1953-1973, nearly one per day!
How do these figures compare with those for other tropical regions of the
world?
Richards (1973) quotes figures of 20,000 for the Flora Malesiana region and
9,000 for the Malay Peninsula, an area smaller than Ghana, only one of the
several component countries of the area of the Flora of West Tropical Africa.
Good (1974: 170) estimates the total for Brazil as 40,000 and for India, Pakistan
and Burma at 20,000, and on a basis of these and other figures considers that
the species density for tropical Africa is much lower than for tropical America
and regions of tropical Asia. The species density figure is obtained by dividing
the area of a given region in square miles into the number of species occurring
in that region.
It is also significant to compare the tropical African figures quoted above
with those for Madagascar and Southern Africa. The specific richness of both
these areas is much greater than for any comparable area of tropical Africa.
The frequency with which species in tropical Africa show wide geographical
ranges is familiar to those who have worked on the botany of this region, but it
is difficult to obtain statistics. Hepper (1965) analyzing the affinities of the Flora
of West Tropical Africa, did a sample analysis of 2,000 species (out of a total
of just over 7,000). The result showed that more than 80% were endemic to
tropical Africa as a whole, 37% to the area of the Flora of West Tropical Africa,
and 17% were widespread in Africa (including 1-2% reaching Madagascar).
An analysis of 63 accepted species of Acacia in the area of the Flora of Trop-
ical East Africa (Brenan, 1959) shows 9 to reach Egypt or Asia, 12 to reach
1978]
439


ANNALS OF THE MISSOURI BOTANICAL GARDEN
TABLE 1. Regional estimates of numbers of species in tropical and South Africa and
Madagascar.
Area of Flora of Tropical East Africa (Kenya, Uganda, Tanzania) (Milne-Red-
head, 1971; Polhill, 1976) 10-11,000
Area of Flora Zambesiaca (Zambia, Rhodesia, Malawi, Mozambique, Botswana)
(Exell, 1971) 6,000
Ethiopia and Somalia (Jardin Botanique National de Belgique, 1976) 6,323
Cameroon (Letouzey, 1976) 6,500
Area of Flora of West Tropical Africa (Hepper, 1976) 7,072
Za�re (L�onard, 1971) 10,000
Madagascar (Keraudren-Aymonin, 1976) 8,200
Southern Africa (South Africa, Lesotho, Swaziland, South West Africa and
Botswana) (Killick, 1976) 17,000
(Goldblatt, this symposium) 18,532
West Africa, 17 to reach South Africa, and 33 more or less widespread in eastern
Africa. White (1965) in a study of the phytogeography of the woodlands of
the Sudano-Zambezian Region found 79 out of 426 tree species common to
West Africa and Zambia. Numerous similar wide distributions of Sudano-Zam-
bezian and Saharo-Sindian Region species are given in the fine series of maps
published by Lebrun (1977). It would be easy to multiply further examples.
The poverty in or complete absence from tropical Africa of many plant
groups, which from their distribution elsewhere might be expected to be well
represented in Africa, is striking. There are certain families, for example, which
might well be expected but which are quite absent, notably Magnoliaceae,
Fagaceae, and Symplocaceae. A fuller discussion will be found in Aubr�ville
(1955). Some examples follow, from Brenan (1954), of families much more
poorly represented than they should be:
Theaceae. 3 genera in tropical Africa, 2 of them endemic. Far better repre-
sented, both in numbers of genera and species, in Asia and America. Ternstroe-
mia, for example, has 2 out of 100 species in Africa.
Myrtaceae. Very richly represented in Asia and America. In tropical Africa
only 2 nonendemic genera. Eugenia has about 1,000 species of which probably
not more than 50-60 occur in Africa.
Melastomataceae. Much more numerous in Asia and especially America
than in Africa.
Araliaceae. 2-3 genera in tropical Africa, out of 55. Much more numerous
in America and Asia.
Monimiaceae. 1 genus in tropical Africa out of about 20.
Lauraceae. One of the large tropical families with 32 genera and 2,000-
2,500 species, mainly in Asia and Brazil. In tropical Africa 5 genera and perhaps
50 species.
Palmae. Moore (1973) gives instructive figures:
Africa, Europe, Arabia 17 genera, 117 species
Madagascar, Mascarene Islands, Seychelles 29 genera, 132 species
South America 64 genera, 837 species
Eastern Tropics 97 genera, 1,385 species
440
[VOL. 35


BRENAN-TROPICAL AFRICA
It is noteworthy that palms, often thought of as almost "indicator plants" of the
tropics, are by no means rare in Africa except perhaps in rain forest. Their
frequency is due to large numbers of individuals of few species. This goes with
some very wide distributions in tropical Africa, e.g., Phoenix reclinata (Senegal
and Kenya to South Africa), Hyphaene thebaica (Senegal to Arabia), Borassus
aethiopum (widely spread), Calamus de�rratus, etc. The paucity of genera and
species is certainly not due to lack of suitable habitats now.
Other examples of families and genera unexpectedly poorly represented in
Africa would not be hard to find. Aubr�ville (1955) mentions the genus Wein-
mannia, only found in Madagascar as far as Africa is concerned but well repre-
sented elsewhere. It is relevant that African poverty is not confined to plants.
Amadon (1973) made a comparison between the avifaunas of the Congo and
Amazon forests and concluded that bird diversity in the Congo forest is at any
one locality usually well below that of Amazonia.
What are the reasons for this biotic poverty in tropical Africa? The problem
has been discussed by Raven & Axelrod (1974: 607-608) who bring together
various likely factors:
(1) Past elimination of taxa by drought.
(2) Major increases in altitude, particularly since the Miocene, accompa-
nied by a cooler and drier climate.
(3) The development of the cold Benguella Current, bringing a drier cli-
mate to the West African coast.
(4) Major fluctuations in Quaternary climate, causing corresponding fluc-
tuations in forest areas.
The rich present flora of Madagascar may on this basis be regarded as relict,
representing a degree of richness probably shared by much of continental Africa
before the violent changes mentioned above took effect. In this way the rela-
tive richness of the Indian Ocean islands in, e.g., Palmae is explicable, as also
the remarkable links between Madagascar and East Africa with America, e.g.,
Trigoniaceae, Rheedia, Oplonia, and the remarkable Hymenaea-Trachylobium
link discussed by Langenheim (1973).
In this way too the wide disjunctions both at the generic and specific levels
between East and West Africa become explicable. For example, the genus
Coleotrype with one species in West Africa, Za�re and Uganda, a second in
Mozambique and Natal, and others in Madagascar; and the extraordinarily dis-
continuous distribution of Mansonia (Chatterjee & Brenan, 1950), with gaps
between Cameroon and Tanzania, and between Tanzania and Assam and Burma
(Fig. 1).
After many years' work on the flora of tropical Africa, I am left most strongly
with the impression that drastic climatic changes in the past are the only means
of explaining distributions that by their discontinuity and unpredictability are a
constant warning against the over hasty formulation of theories to explain them!
PHYTOGEOGRAPHICAL REGIONS OF AFRICA
As knowledge has increased about the distribution within the tropics of
Africa of plant species and the types of vegetation they compose, various at-
1978]
441


ANNALS OF THE MISSOURI BOTANICAL GARDEN
FIGURE 1. Afro-Asian disjunction. Distribution of the genus Mansonia (Sterculiaceae):
1. M. altissima.-2. M. nymphaeifolia.-3. M. diatomanthera.-4. M. dipikae.-5. M. gagei.
(After Chatterjee & Brenan, 1950.)
tempts have been made to define phytogeographical regions within tropical
Africa and to arrange them schematically. It is not necessary to describe and
discuss these schemes in detail here. A recent summary has been given by
Schnell (1976: vol. 3: 47-60). Although Good (1974) in his excellent book on
the geography of flowering plants divided Africa, with the rest of the world, into
floristic regions, his scheme has been justly criticized by Schnell (1976: vol. 3:
58) on the grounds that it takes too little account of the general ecology and
fails to give due prominence to disjunct regions linked with higher altitudes.
One of the most recent schemes, and the one accepted here, is set out by
Wickens (1976: 40-48), but as he states, it is itself based on two earlier works
(White, 1965; Chapman & White, 1970).
Wickens, dealing with the whole of Africa, recognized nine regions, of which
five are represented in the area dealt with here. Three of the regions are them-
selves subdivided into domains (Figs. 2, 3). A brief summary of the regions
and domains may be helpful.
1. SUDANO-ZAMBEZIAN REGION
This region corresponds to the tropical savanna and is by far the largest in
tropical Africa, extending both north and south of the equator, but physically
continuous only by a comparatively narrow isthmus in east Africa.
442
[VOL. 65


1978] BRENAN-TROPICAL AFRICA 443
]0 S 10 I- x0 * x5 b .5 U .� 1
Western "Mediterranean C . Cetral ' Sb
S- irano-Anatolin Sub-relon regon
Mtuit) reo n - . rgS'h.
SIRNO TUMEDITERRANEANIAN
' '; ---  . . REGI.,ON - . -f-. - ; I
S.TMiddle Saharu - -ndin Sub- regio
. Weste. Sah o Sindian Sub - gon
! �,"" "SAHARO-SINDIAN REGION
SN � , L/: . South Arabian
~  <'r Sudan.an Domaine �......-- . / , / f �
Guin.a Domain GUINEO-CONGO REGION Doai
r , ,
Congo Domain
* J - -:Usa'' mbaro- Zululand
� \.I ADAGASCAN REGION
KARROO NAMIB
REGION .
_ REGION
To the north it is bounded by the deserts or semideserts of the Saharo-Sindian
Region (No. 5). In its central part it embraces towards the west the great forest
area of the Guineo-Congo Region (No. 2). To the south it is bounded by South
West Africa and South Africa (the Karroo-Namib and Cape Regions, not dis-
cussed here).
The region is characterized by a strongly seasonal climate reflected in a
range of vegetation types from poor thorn scrub to quite rich deciduous wood-
land. There are very few endemic families. Although Wickens (1976: 40)


ANNALS OF THE MISSOURI BOTANICAL GARDEN
/ iY 4<
ACACIA SIEBERANA
A
i
( i.u. : ~ .� L�~
i� ~-�I~1
t� >'I-)
t,
ACACIA PENTAGONA
C D
FIGURE 3. Diagrams of some representative distribution patterns of tropical African
plant species.-A. Acacia sieberana of the Sudano-Zambezian Region; note outlying area in
Angola.-B. Acacia pentagona, mainly in the Guineo-Congo Region, but also occurring in
forest outliers elsewhere. C. Ccrastium octandrum of the Afromontane Region, showing typi-
cal disjunctions corresponding to mountain arcas.-D. Cornulaca monacantha of the Saharo-
Sindian Region, with typical extensions to Arabia and eastwards. (A & C after Wickens,
1976, D after Lebrun, 1977.)
implies that there are few endemic genera, the evidence indicates that this may
not be true. It is certain that there are a large number of endemic species and
that some genera (e.g., Crotalaria, Indigofera) have speciated profusely. Some
of the more characteristic genera will be mentioned under the component do-
mains, using many of those cited by Wickens (1976). The flora of the region
as a whole is discussed more fully on pp. 456-459.
444
[VOL. 65


BRENAN-TROPICAL AFRICA
(a) Sahelian Domain.-From Mauritania and Senegal on the Atlantic to the
Red Sea coast of the Sudan Republic, forming a belt bounded on the north by
the Indo-Turanian Region. Certain widespread species of Acacia, A. senegal, A.
seyal, and A. nubica, are characteristic, as is the absence of Terminalia. White
(1965: 657) remarks that it does not have a distinctive woody flora and that in
West Africa there are only eight tree species which do not also occur in adjacent
vegetation zones to the south, and that ail eight are extensively distributed in
the Oriental Domain (here called Afroriental).
(b) Sudanian Domain.-From Senegal to the Sudan Republic (Ethiopian
frontier). This domain forms a rather wider belt than the last, by which it is
bordered to the north; to the south it passes into the Guineo-Congo Region.
White (1965: 662) states that most woody species of this domain have wide dis-
tributions and that there are no well-defined restricted areas of endemism. Char-
acteristic species of this domain are Isoberlinia doka and Khaya senegalensis.
(c) Afroriental Domain.-In contrast to the Sahelian and Sudanian domains,
this area is floristically much richer. Wickens (1976: 41) cites five endemic
genera, but this is very far from a complete tally. Certain genera show high
speciation, e.g., Acacia, Commiphora, Combretum, the cactiform Euphorbiae,
Grewia, etc.
(d) South Arabian Domain.-Apparently an impoverished extension of the
Afroriental Domain, extending into southwestern and southern Arabia. It is
little known, and to be noted but not further discussed here.
(e) Zambezian Domain.-This comprises the remainder of the Sudano-Zam-
bezian Region lying to the south of the Afroriental Domain. It is floristically
rich and characterized by such genera as Brachystegia, Julbernardia, Isoberlinia,
and Colophospermum.
2. GUINEO-CONGO REGION
This region represents the main evergreen or partly evergreen forests of
Africa, extending southwards to Angola and eastwards to Ruwenzori. There are
a number of endemic families and a high proportion of generic endemism. How-
ever, in general prolific speciation is absent.
Three domains are recognized.
(a) Guinea Domain.-Extending along the northern coast of the Gulf of
Guinea from Gambia to central Nigeria. Poor in endemism compared with (b)
the Congo Domain.
(b) Congo Domain.-From eastern Nigeria to Angola and the Congo. The
richest area in endemism.
(c) Usambara-Zululand Domain.-The scattered relics of rain forest along
or near the eastern coast of Africa, extending from southern Kenya southwards
into South Africa. This is the most distinct of the three domains, now very lim-
ited in area, but with a high degree of specific endemism, becoming gradually
impoverished southwards.
It is proposed to discuss the relationships of the forest flora of tropical Africa
in a separate section (pp. 451-456).
445
1978]


ANNALS OF THE MISSOURI BOTANICAL GARDEN
3. AFROMONTANE REGION
This region corresponds to the Montane Forest and Ericaceous belts of the
tropical African mountain regions. As one might expect, and as can be seen
from the map, this region is widespread and notably disjunct, with strong con-
centrations in eastern Africa, but dispersed elsewhere southwards to Malawi
and westwards to Cameroon and other parts of West Africa.
Characteristic species are: Juniperus procera, Podocarpus milanjianus, Olea
africana, Prunus africana, Hagenia abyssinica, and Hypericum lanceolatum in
the montane forest; species of Philippia, Erica, Stoebe kilimandscharica, and
Adenocarpus mannii in the ericaceous zone. Although widespread, this region is
not floristically rich and is floristically strikingly uniform.
No general analysis of the tropical African montane flora has been published.
Morton (1972) analyzed the West African montane flora. Of 718 montane spe-
cies, 47% are West African endemics, and 53% occur also elsewhere, especially
on the mountains of East and central Africa. The latter are thus nearly ail strik-
ingly disjunct in distribution, a general feature of the flora of the African moun-
tains. Ayodele Cole (1974) gives some useful lists of endemics. A general dis-
cussion of the montane forest plants of Africa is given by Hamilton (1976).
4. AFROALPINE REGION
This is the zone above the Ericaceous belt of the Afromontane Region. It is
characterized by giant Senecio species, giant Lobelia species, low shrubby spe-
cies of Alchemilla, and tufted grassland. It is floristically poor, though with
visually striking components, and is strongly endemic, particularly at the spe-
cies level.
It is almost confined to the high mountains of eastern Africa, from Ethiopia
to Tanzania. Although the Cameroon Mountain, for example, is high enough
probably to suggest an Afroalpine belt, there is no indication of its development
there.
A systematic analysis of the species inhabiting the Afroalpine Region was
given by Hedberg (1957). Later refinements (Hedberg, 1965, 1969) resulted
in a total of only 278 species, which he analyzed according to their phytogeo-
graphical affinities. However, no less than 81% of the total are endemic to the
high mountains of East Africa. The flora appears to be ancient and of very
mixed derivation, both local and distant, from the Cape to Europe and the
Himalayas.
Clayton (1976) analyzed the chorology of African montane grasses and
found the maximum concentration of single-station endemics (6) in Ethiopia,
with surprisingly few elsewhere. A radial movement of species from an Ethi-
opian reservoir was suggested. Certainly the linkages between ecologically iso-
lated mountains are clear, and even with the geographically remote Cameroon
Mountain.
5. SAHARO-SINDIAN REGION
This region of desert and semidesert, extending from Morocco to India, is
barely represented in tropical Africa except along the northernmost fringe of
446
[VOL. 65


BRENAN-TROPICAL AFRICA
the area. The climate is harsh, with high but often extremely fluctuating tem-
peratures, and low erratic rainfall. The flora is poor and scanty. Wickens (1976:
46) estimates it as only about 1,500 species in ail. Endemism is low, indeed
almost nonexistent at the family level. Characteristic species include Calotropis
procera, Salvadora persica, Panicum turgidum, etc.
ANALYSIS OF ENDEMISM AND GEOGRAPHICAL AFFINITIES OF FAMILIES
In order to analyze the endemism and geographical affinities of the flower-
ing plant families of tropical Africa, a survey was made, using as a basis Willis's
(1973) well-known Dictionary, ed. 8. Although some may maintain that the
family limits have been defined rather narrowly, nevertheless for the present
purpose this has certain advantages, and the work as a whole is valuable as a
recent recension of families as critically viewed by a single taxonomist.
Let us first review those families endemic to continental tropical Africa.
There are nine: Barbeyaceae, Dioncophyllaceae, Hoplestigmataceae, Huaceae,
Medusandraceae, Napoleonaeaceae, Octoknemaceae, Pentadiplandraceae, and
Scytopetalaceae.
Ail the above are strictly endemic, except that Barbeyaceae just reaches adja-
cent Arabia, though otherwise confined to northeastern tropical Africa-a region
well known to be a center of endemism of genera and species.
All the other eight are nearly or quite restricted to the Guineo-Congo Region
of tropical Africa and are predominantly inhabitants of the rain forest. Napo-
leonaeaceae (sometimes considered as part of Lecythidaceae) is entirely a for-
est family, except for Napoleona gossweileri, a tiny suffrutex of the Kalahari
Sand savanna in Angola and Zambia-surely an adapted relie and a pointer to
drastic climatic changes in the past.
The families are all quite small, the largest being Scytopetalaceae and Napo-
leonaeaceae, with 5 genera and 20 species, and 2 genera and 18 species respec-
tively. Perhaps this again is indicative of the low level of speciation already
noted in the Guineo-Congo Region.
It is instructive to widen the survey of families to cover those occurring in
tropical Africa, but restricted to the continent as a whole. Four additional fami-
lies occur both in tropical and South Africa, but not elsewhere: Kirkiaceae, Meli-
anthaceae, Oliniaceae, and Wellstediaceae.
Of these families only Melianthaceae occurs in West Africa where it is rep-
resented by the genus Bersama, a taxonomically difficult genus still apparently
actively evolving and probably a recent element in the rain forest. The others
are represented in eastern tropical Africa, with the Wellstediaceae disjunct be-
tween southwestern and northeastern Africa, where it extends to Socotra.
Certain families are restricted to Africa and Madagascar:
Androstachyaceae : S.E. tropical Africa (extending to Natal) and Mad-
agascar
Montiniaceae : E. tropical and S.W. Africa and Madagascar
Myrothamnaceae : E. tropical, South Africa and Madagascar
Ptaeroxylaceae : E. and S. tropical, South Africa, and Madagascar
447
1978]


ANNALS OF THE MISSOURI BOTANICAL GARDEN
Uapacaceae : Tropical Africa and Madagascar
Hydrostachyaceae : Tropical and South Africa and Madagascar
Ail the families so far discussed are small. The largest are Uapacaceae and
Hydrostachyaceae, each with 1 genus with 50 and 30 species respectively. Of
the six families endemic to Africa and Madagascar all occur in East Africa, and
only one, Uapacaceae, occurs in West Africa, but in Guinea savanna, not rain
forest.
These distributions emphasize the separation of the West African rain-forest
flora from both South Africa and Madagascar. Cola (Sterculiaceae), perhaps
the largest mainly rain-forest genus in tropical Africa with 125-150 species, is
absent from Madagascar. Conversely there is positive evidence of closer links
at the family level between Madagascar and East Africa.
Using again the evidence of Willis (1973) there are 14 families endemic to
South Africa and 13 to Madagascar. Both groups are thus considerably larger
than the group of families endemic to tropical Africa, and underline the richness
of the flora of these two regions, or perhaps rather the poverty of the tropical
African flora.
It is instructive to broaden the picture to include families restricted to Africa
and either America or Asia. Twelve families are African/American and nine
African/Asian. At the family level there is no great discrepancy between these
two groups; at the generic level there emphatically is.
It is instructive to tabulate these results together and compare them with
some derived from other parts of the world as in Table 2.
ANALYSIS OF ENDEMISM AND GEOGRAPHICAL AFFINITIES OF GENERA
As careful data on geographical distribution are given for each genus in
Willis's (1973) Dictionary of the Flowering Plants and Ferns, ed. 8, I did a
survey of ail tropical African genera, using the information in that work supple-
mented by personal knowledge. This gave the following results: a total of the
genera occurring in tropical Africa; a total of the genera endemic to tropical
Africa; genera confined to tropical and South Africa; genera confined to Africa
and Madagascar; American/tropical African genera; Asian/tropical African gen-
era; and others. It was possible to subdivide the endemic genera, though far
from completely or satisfactorily. In the end it was of course possible to express
the results in terms of percentages. It should be noted that the figures relate to
flowering plants only: the pteridophytes are excluded. The results are given in
Table 3.
These figures are, of course, subject to various sources of error. The concept
of genera frequently alters with increasing taxonomic knowledge, and with it,
of course, the totals in Table 3 are altered. The geographical analysis has had to
be made in very broad terms, to conform to the standard of data available. In
particular the subdivisions of the endemic tropical African genera must be
treated with caution. Both of the first two groups are likely to be augmented
from the third. However, in spite of these cautions, the figures are likely to be
sufficiently near the truth to allow various points to be made. The figures given
448
[VOL. 65


BRENAN-TROPICAL AFRICA
TABLE 2. Distribution patterns and endemism of angiosperm families restricted to
Africa, South America or tropical Asia, or restricted to two of these areas.
Number of
Region Families
Endemic to tropical Africa 9
Confined to tropical and South Africa 4
Confined to tropical Africa and Madagascar (and sometimes South Africa) 6
Endemic to South Africa 14
Endemic to Madagascar and/or Mascarenes 13
Confined to tropical Africa and America 12
Confined to tropical Africa and Asia 9
Endemic to tropical Asia 31
Endemic to South America 34
by Good (1974: 99, 145), although based on different geographical areas not
too easy to equate, must clearly be modified. In most instances they appear to
be considerable underestimates.
The following points may be made on the basis of Table 3.
(1) Between one-third and one-half of the total genera are endemic to
tropical Africa, and this is the single most important element. If those also
occurring in South Africa are added, it is seen that nearly 60% of the total are
endemic to continental Africa. The uniqueness of the African flora is suffi-
ciently attested.
(2) West and central tropical Africa are richer in endemic genera than the
east and south (19.62% against 12.62%). Although the evidence is inadequate
for exact analysis, it is likely that the majority of the western and central endem-
ics are forest genera, the generic endemism outside the forest areas of West
Africa being small.
(3) The percentages of tropical African genera extending to South Africa and
Madagascar (8.81% and 7.05%) are surprisingly low proportions of the total of
tropical African genera.
(4) No less than 464 genera (18.59%) extend to Asia and/or Australia. This
element is numerically next in size to the tropical African endemic one, and is
a clear indication of the importance of the phytogeographical links between
Africa and Asia.
(5) Conversely, the tropical African/American element is small (88 genera-
3.52%).
(6) The relative significance of the pantropical and temperate elements is
noteworthy.
It may be instructive to compare the figures given in Table 3 to those in
Table 4 giving some parallel percentages obtained from a separate study of the
geographical relationships of the genera of Leguminosae occurring in tropical
Africa (Brenan, 1965).
The similarity between the proportions in Table 4 and those in Table 3 are
striking. The tropical African-Madagascan element is considerably less, and the
pantropical more, the latter perhaps reflecting the weediness and wide distribu-
tions of a number of genera in the Papilionioideae.
1978]
449


ANNALS OF THE MISSOURI BOTANICAL GARDEN
TABLE 3. Distribution patterns of tropical African genera. Total
occurring in tropical Africa: 2,497 (100%).
number of genera
Region
Endemic to tropical Africa
W. and central tropical Africa
E. and S. tropical Africa
General or unspecified
Endemic to tropical and South Africa
Endemic to tropical Africa and Madagascar
(with sometimes South Africa)
Restricted to tropical Africa and Asia
(plus sometimes South Africa)
Restricted to tropical Africa, Asia and Australia
or tropical Africa and Australia alone
Restricted to tropical Africa and America
Pantropical
Mainly temperate but occurring in tropical Africa
Total
Genera
1,081
490
315
276
220
176
Percentage
43.33
19.62
12.62
11.05
8.81 1
15.86
7.05
351 14.06
113
88
303
148
4.53
3.52
12.13
5.92
18.59
This is perhaps a convenient place to discuss more fully the relative impor-
tance of the links with America and Africa. The analysis of family distributions
(Table 2) shows that at this level there is little significant difference numerically
between the two elements, but at the generic level the balance is very strongly
weighted in favor of Asia. Thorne (1973) has analyzed the floristic relation-
ships between Africa and tropical America and has carefully tabulated the fami-
lies and genera in common. His total of 111 African/American genera is higher
than the 88 given here, because he takes Africa in a wide sense, including
South Africa, Madagascar, and the Mascarenes. As Thorne states, the floristic
links between tropical Africa and tropical America are undeniable, and some
are spectacular (Fig. 4). The occurrence in Africa of Pitcairnia feliciana (Brome-
liaceae), Kissenia (Loasaceae), Erismadelphus (Vochysiaceae), Maschalocephalus
(Rapateaceae), Sacoglottis (Humiriaceae), Mayaca baumii (Mayacaceae), Cy-
licomorpha (Caricaceae), etc., have attracted wide and justified attention.
However, Thorne draws attention, quite justly, to the absences as well as the
presences, and tabulates (p. 35) a considerable number of large or very large
African and American genera which might have been expected to occur in
both continents, but do not. As he says, this is a factor which must be consid-
TABLE 4. Distribution patterns and proportions of genera of Leguminosae occurring in
tropical Africa. Total number of genera: 229.
Total
Genera
Region
Percentage
Endemic to tropical Africa
Endemic to tropical and South Africa
Endemic to tropical Africa and Madagascar
Restricted to tropical Africa and Asia and/or Australia
Restricted to tropical Africa and America
Pantropical
Mainly temperate or Mediterranean but
occurring in tropical Africa
41.92
6.99
1.74
15.28
4.37
18.34
14 6.11
~~--~~--~--
450
[VOL. 65


BRENAN-TROPICAL AFRICA
135 120 105 90 75 60 45 30 15 0 15 30
FIGURE 4. Afro-American disjunctions. Distribution of Bromeliaceae (white area) and
of Rapateaceae (black area). The only African members of these families are Pitcairnia feli-
ciana and Maschalocephalus dinklagei respectively. (After Hepper, 1965.)
ered. He concludes that a comparison of the floristic links between tropical
Africa, tropical America, and Asia would appear to rule out continental drift as
a valid explanation of these widely disjunct ranges, and I agree with him that
long-distance dispersal would seem quite adequate to explain the floristic rela-
tionships between tropical Africa and tropical America. A similar conclusion
was reached by Ayensu (1973), in considering the remarkable Afro-American
family Velloziaceae with 47 species in Africa, 3 in Madagascar and about 200 in
South America.
ENDEMISM IN THE TROPICAL AFRICAN RAIN FOREST
In 1954 I carried out a detailed study of endemism in the rain forest areas of
tropical Africa (Brenan, 1954). Since most of this has not been published and
in spite of the lapse of time, the conclusions still seem relevant and valid, some
1978]
451


452
ANNALS OF THE MISSOURI BOTANICAL GARDEN
[VOL. 65
TABLE 5. Area subdivisions of the tropical African rain forest. Areas numbered; area-
groups indicated by letters.
A. Western: Sierra Leone-Ghana
1. Sierra Leone
2. Liberia
3. Ivory Coast
4. Ghana
B. West and central Nigeria
5. Nigeria (west and central)
C. Cameroon, Gabon and Congo-Brazzaville
6. Nigeria (eastern)
7. Cameroon (northwestern) and Fernando Po (Macias Nguema)
8. Cameroon (remainder) and Principe
9. Gabon, Equatorial Guinea, Sao Tom�, Annobon
10. Cabinda
11. Congo-Brazzaville
D. Za�re
12. Za�re
E. Uganda
13. Uganda
F. Tanzania
14. Usambaras
15. Ngurus and Ulugurus
G. Northern Forest Outliers (Republic of Guinea to Sudan Republic)
H. Southern Forest Outliers (Angola eastwards to Zambia and Rhodesia)
I. Eastern Forest Outliers [Kenya, Tanzania (except as under F), and Mozambique]
of the main ones are described here. The analysis was based entirely on a per-
sonal examination of herbarium material at the Royal Botanic Gardens, Kew,
supplemented by field experience in East and West Africa.
The rain forest area was taken in its widest sense, to include not only the
main block in West and central Africa, but also the easterly extension into
Uganda, the outlying island areas in Tanzania, and the other outlying areas
from the Republic of Guinea to the Sudan and southwards to Rhodesia and
Mozambique.
The area subdivisions, updated nomenclaturally, are listed in Table 5.
It will be seen that the total region defined corresponds (except for various
outliers) roughly to the Guineo-Congo Region, including the Usambara-Zulu-
land Domain.
Let us now discuss the results. Of 133 families present, 8 (6.01%) are
endemic to the rain forest areas as defined. Of a total of 1,121 genera, 492
(43.9%) are endemic to the rain forest region and 629 (56.1%) nonendemic.
Among the nonendemic families, the degree of generic endemism varies from
0-100%. Among larger families or subfamilies with high proportions of ende-
mism are the following:
Myristicaceae 5 genera 100%
Sapotaceae 26 genera 83.9%
Annonaceae 25 genera 73.2%
Melastomataceae 16 genera 72.7%
Leguminosae-Caesalpinioideae 41 genera 67.8%
Apocynaceae 24 genera 61.5%


BRENAN-TROPICAL AFRICA
TABLE 6. Numbers and proportions of genera endemic to the tropical African rain forest
occurring in single-numbered areas, or area-groups G, H, and I as defined in Table 5.
Proportion of
Total Generic
Number of Endemism in
Area Endemic Cenera Each Area
1. (Sierra Leone) 1 0.6
2. (Liberia) 1 0.7
3. (Ivory Coast) 5 3.9
4. (Ghana) 2 1.25
5. (Nigeria, N., W. and C.) 1 0.6
6. (Nigeria, E.) 4 2.1
7. (Cameroon, N.W. and Fernando Po) 5 2.5
8. (Cameroon, remainder and Principe) 26 8.9
9. (Gabon, Equatorial Guinea, and islands) 28 9.5
10. (Cabinda) 1 0.6
11. (Congo-Brazzaville) 3 5.8
12. (Za�re) 19 7.9
13. (Uganda) 0 0
14. (Usambaras) 4 7.7
15. (Ngurus, Ulugurus) 7 25.9
G. 4 3.6
H. 7 7.1
I. 2 3.0
It should be noted that the above analysis of Sapotaceae was made at a time
when generic fragmentation in the family had not reached its present-day state.
Probably now both the number of genera and the proportion of endemism would
be higher. It is also noteworthy that none of these larger families is represented
by an outstandingly large number of endemic genera-another facet of the flo-
ristic poverty of tropical Africa referred to elsewhere. Taking only those families
possessing some proportion of endemism among their genera, the average of
endemism per family is 55.12%.
An analysis of the numbers of genera in single-numbered areas and area-
groups G, H, and I (see Table 5) is given in Table 6.
It can be seen from these figures that areas 1-4 are collectively rather poor
in endemic genera, these possessing large rain forest areas. The contrast be-
tween the two parts of Nigeria, areas 5 and 6, is noteworthy and is decidedly
confirmed by field experience. There is clearly a phytogeographical discontinu-
ity here, located somewhere between the Niger and Cross rivers. The relative
richness in endemic genera shown in Cameroon and Gabon emerges very clearly
from the above table. In each area the number of endemic genera is much
greater than that of Za�re, in spite of the much more extensive rain forest in
Za�re. This contrast is surely attributable to the effect of past climatic changes;
see Chapin (1936).
Passing to East Africa, the absence of any endemic genus in Uganda is inter-
esting in comparison with Tanzania, and suggests that the rain forest in Uganda
is no more than a depauperate eastward extension of the rain forest of eastern
Za�re. The generic endemism of the Usambaras, Ulugurus, and Ngurus in Tan-
zania is striking.
1978]
453


ANNALS OF THE MISSOURI BOTANICAL GARDEN
TABLE 7. Analysis of endemic tropical African rain forest genera occurring in 3 or more
area-groups as defined in Table 5.
Number of Endemic Proportion of Total
Area Croups Genera of Endemic Genera
1) ABC 27 5.49
2) ABCD 54 10.97
3) ABCDE 34 6.9
4) ABCDEF 17 3.45
5) BCD 10 2.04
6) BCDE 3 0.61
7) CDE 17 3.45
8) CDEF 6 1.22
An analysis was made of genera occurring in three or more area-groups (see
Table 5) and this is instructive (Table 7). Discontinuities, which are striking,
will be discussed separately below.
Out of a total of 492 endemic genera, 167 or 33.94% have widespread ranges,
in the sense adopted above. The rather high number of genera in (1), which fail
to reach Za�re, is notable (Table 7). If the total numbers of widespread endemic
genera are taken for each area-group, the following figures are obtained: A-
144, B-145, C-168, D-141, E-82, F-32.
In each of the areas A-D, the totals are roughly similar, with C slightly
higher than the rest. This is in very marked contrast to the figures given in
Table 6 of narrowly endemic genera restricted to numbered areas, where area-
groups C and to a lesser extent D are very much richer in narrowly endemic
genera than other area-groups. This contrast is hard to explain. It may be that
these widespread genera represent the ancient core of a widespread tropical
African rain forest flora, which has in the course of time given rise to the
narrowly endemic genera. On the other hand, these widespread genera may
have attained their ranges more recently, through such means as effective long-
distance dispersal or abundant seed production. Both causes may well have
had their effect, though I suspect that the second may have been more signifi-
cant than the first. The prevalence of wind-dispersed emergent trees in the
Nigerian rain forest is significant, and also the importance of clearings in their
regeneration.
Mention has already been made of the discontinuities in distribution that
are a feature of the rain forest flora of tropical Africa. The analysis given in
Table 8 was made.
The first group, found in the area Sierra Leone-Ghana but absent from
western and central Nigeria, but appearing again in eastern Nigeria or other
countries to the south, is a large one with 73 genera. This discontinuity is also
found at the specific level: e.g., Cola chlamydantha, Protomegabaria macro-
phylla, Thecacoris stenopetala, Pararistolochia mannii, etc. Similar examples
among mammals are given by Rosevear (1953: 35). This discontinuity is dis-
cussed by Guillaumet (1967: 145-168) and corresponds to Groups IV and V of
his "esp�ces Sassandriennes," with 47 species discontinuous between the Ivory
Coast or Ghana and the Cameroon-Congo area. On this evidence Guillaumet
454
[VOL. 65


BRENAN-TROPICAL AFRICA
TABLE 8. Discontinuities between area-groups as defined in Table 5 shown by genera
endemic to the tropical African rain forest.
Number of Genera Percentage of Total
Discontinuity Involved of Endemic Cenera
1. Between A and C 73 14.63
(absent from B)
2. Between A and D 3 0.61
3. Between A and E 1 0.20
4. Between A and F 1 0.20
5. Between A and I 1 0.20
6. Between C and E 12 2.43
7. Between C and F 9 1.83
8. Between C and H 2 0.41
9. Between C and I 10 2.04
10. Between D and F 10 2.04
11. Between D and I 11 2.24
postulates "refugia" of forest in Sierra Leone and Liberia in a past period of
arid climate.
Discontinuity 6, involving 12 genera missing from Za�re, is probably more
apparent than real and may well reflect poor knowledge of the Za�rian rain
forest.
The discontinuities between C and F and between D and F are important.
Although some may eventually be found in Uganda (E), the latter country is
comparatively well-known botanically. The evidence indicates that the rain
forest flora in the Usambaras, Ulugurus, and Ngurus in Tanzania has long been
isolated, showing affinities with West Africa at the generic rather than the spe-
cific level, and was probably derived from a previous rain forest flora in which
many of the widespread genera present today in Uganda and Za�re were absent.
The rain forest in F may thus represent the fragmented relies of a primitive and
formerly more widespread forest flora not clearly recognizable elsewhere in
Africa today. Moreau (1952) suggests that the eastern African lowland rain
forests may have been separated from the western at least since the mid-Pleisto-
cene, and possibly a good deal longer-a period of 500,000 years or more.
The presence of endemic species of predominantly rain forest groups in habi-
tats other than rain forest in East Africa is also evidence pointing in the same
direction, e.g., Mansonia diatomanthera (Chatterjee & Brenan, 1950), Berlinia
orientalis, and the East African species of Tessmannia (Brenan, 1967). There
are many other examples.
So far the endemic genera have been treated as units, with little or no refer-
ence to their size. An analysis shows that 231 out of the 492 endemie genera are
monotypic, 82 have two species each, 48 have three species each, and so on-
in general the more species in the genus the fewer the genera become. Thus,
at the other end of the scale, there are single genera each with 25, 27, 28, 34 and
36 species; none with more. Thus, none of the genera endemic to the tropical
African rain forest region is large. Although some genera in the region may be
represented by as many as 100 species, e.g., Cola, Rinorea, Psychotria, and Ficus,
1978]
455


ANNALS OF THE MISSOURI BOTANICAL GARDEN
none is confined to rain forest, and all except Cola are found in other continents
than Africa.
THE SAVANNA AND WOODLAND REGIONS OF TROPICAL AFRICA
The savanna and woodland areas of tropical Africa, corresponding in general
to the vast Sudano-Zambezian phytogeographical Region (Fig. 2), has unfortu-
nately not been subjected to the same detailed phytogeographical analysis as
has the Guineo-Congo rain forest Region. Of necessity, then, any general com-
ments must be tentative and subjective, liable to reexamination when the grad-
ual accumulation of evidence permits. As can be seen from the map (Fig. 2),
the continuous tracts of savanna, particularly north of the equator, are far
greater than those of any other vegetation type in Africa.
Although it is evident that there has been in the past floristic interchange
between the forest and savanna (cf. particularly the presence of species of
forest affinity in the savannas of southern Africa, e.g., Napoleona gossweileri,
Caloncoba suffruticosa, Syzygium huillense, Magnistipula eglandulosa), never-
theless the "violent contrast" between the savanna and forest floras of tropical
Africa has been emphasized by Monod (1971: 377).
White (1965) studied the individual distributions of 426 Sudano-Zambezian
savanna-woodland tree species. Generic endemism was found to be slight (15%),
but specific endemism high within the region, most of the 426 species being
endemic, only 2 occurring also in the Guineo-Congo Region. Both at the generic
and specific levels the relative richness of the Zambezian Domain within the
region emerges. In West Africa there are 171 species, in Zambia 334. Of the
Zambian species 34% are confined to three clearly defined centers of endemism,
the Katangan, Barotse, and Zambezi. These results for tree species are readily
paralleled in other life forms.
Although the disjunctions of distribution so obvious in the Guineo-Congo
Region have been less publicized in the Sudano-Zambezian Region, they are
there and give clear indication of climatic fluctuations and changes in the past
as severe as those affecting the rain forest. Lebrun (1971) has given and mapped
some remarkable examples of disjunction between the savanna regions of north-
ern and southern tropical Africa. De Winter (1971) has focussed attention on
the disjunctions linking the more arid areas of northern and southern Africa.
Such disjunctions, although remarkable, emphasize the basic unity of the Su-
dano-Zambezian Region. Remarkable examples of such distributional disconti-
nuities as Kissenia (South West Africa, Somalia, and Arabia) and Wellstedia
(South West Africa, Socotra, and northeastern Africa), at first sight quite anom-
alous, fall into place against a background of numerous discontinuities, many of
them less extreme and passing into examples of more continuous distribution.
Gillett (1958) revised the genus Indigofera in tropical Africa. This genus is
represented in the Sudano-Zambezian Region of tropical Africa by numerous
species, in excess of 250, is widespread, and barely penetrates the forest areas.
Gillett set out the geography in some detail, and it therefore seemed appropriate
and useful to analyze its distribution (Table 9) in terms of its occurrence in the
various domains of the Sudano-Zambezian Region, to make an assessment of its
[VOL. 65
456


BRENAN-TROPICAL AFRICA
TABLE 9. Analysis of distribution patterns of African species of Indigofera, using data
from Gillett (1958). Total number of species; 279.
Number of
Region Species
Non-African (or occurring only outside continental Africa) 5
Occurring in more than one region and not analyzed further 29
Confined to Cuineo-Congo Region 2
Confined to Karroo-Namib Region 9
Confined to Sudano-Zambezian Region - Usambara-Zululand Domain 232
Region doubtful 2
Sudano-Zambezian Region (232)
In more than one Domain 83
N. and S. of Equator 44
N. of Equator 13
S. of Equator 26
In Sahelian Domain only 2
In Sudanian Domain only 30
Narrow endemics:
Guinea Republic 3
Nigeria 3
Sudan Republic 1
Central African Republic 2
In Afroriental Domain only 45"
Narrow endemics:
Ethiopia 5
Somalia 5
Socotra 2
Kenya 5
Tanzania 12
In Usambaro-Zululand Domain only 17
Narrow endemics:
Tanzania 1
Mozambique 6
In Zambezian Domain only 55
Narrow endemics:
Za�re 3
Mozambique 1
Zambia 3
Rhodesia 1
Angola 8
Transvaal 2
a Of the species other than narrow endemics 11 are confined to N. of the Equator, only one to the S. and
only one occurring both N. and S.
local endemism, and to compare the savanna areas north and south of the
Equator.
A number of significant conclusions emerge from these figures.
(a) The large number of species occurring in more than one domain, and
also north and south of the Equator.
(b) The small number of species apparently confined to the Sahelian Domain.
(c) Except in the Afroriental Domain, which is a special case, 45 species
are restricted to north of the Equator, and 143 to the south.
(d) North of the Equator (except in the Afroriental Domain) narrow ende-
mism is mainly towards the west (Nigeria and Guinea), but even here the
numbers are very low compared with elsewhere in the Region.
1978]
457


ANNALS OF THE MISSOURI BOTANICAL GARDEN
(e) In the Afroriental Domain, the high endemism of the "Horn of Africa"
emerges clearly. Of the 5 Kenya endemics, 4 are confined to the northeast. The
richness of Tanzania is also evident.
(f) In the Zambezian Domain, the endemism is less than one might expect in
Zambia, but high in Angola.
Although there are obvious possible sources of error in utilizing the above
data (e.g., lapse of time and difficulty in establishing clear boundaries of re-
gions and domains), nevertheless it is unlikely that the general pattern and the
deductions drawn will be much altered. Most of the deductions can be readily
paralleled by experience of other taxonomic groups in tropical Africa. For ex-
ample, in the genus Acacia there are very few endemics indeed in the Sudanian
and Sahelian domains. Although the genus is well represented there, most of
the species are more or less widespread in tropical Africa and not infrequently
extend into Asia. The "Horn of Africa" has at least a dozen endemic species of
Acacia, and no other part of Africa shows as many. Secondary areas of concen-
tration occur in Tanzania and in the area Rhodesia-Botswana-Mozambique-
northern Transvaal.
Further studies, both general and local, of the phytogeography of the Su-
dano-Zambezian Region of tropical Africa would be most welcome. However,
this region as well as being the largest in Africa is also I suspect considerably
more heterogeneous than the rest. It includes a wide range of vegetation types
ranging from thorn scrub and grassland to savanna and rich woodland. Further-
more the phytogeographical origins and history of these various vegetation types
may be by no means always the same. These differences are not, I think, ade-
quately reflected in the domains composing the Sudano-Zambezian Region. For
example, the Afroriental Domain includes the area of northern Kenya-Somalia-
Socotra-Ethiopia, in which are to be seen the remains of a highly endemic,
probably comparatively ancient flora very different from that of the rest of
Kenya and Tanzania, thus raising the question of whether the domain is really
a single one or whether it would be better divided.
The variety of vegetation types in the Zambezian Domain is indicated by
the works of Wild & Fernandes (1967) and Fanshawe (1969). One of the
remarkable features of this domain is the prevalence over huge areas of wood-
land or savanna woodland in which the genera Brachystegia and Julbernardia
are conspicuous or dominant-the so-called "miombo." Typically this occurs in
no other domain. Attempts have been made to parallel "miombo" in West Africa,
but at best the differences are profound.
In terms of numbers of woody species, this is the richest vegetation type in
Zambia (Fanshawe, 1969: 43), and this richness is particularly evident also in
the herb layer. Many of the local endemics so frequent in Zambia, Angola, and
southern Za�re occur in "miombo." A remarkable feature is the richness of suf-
frutices, often relatives of large trees that have adopted a habit nearer to that
of a perennial herb. Carcasson (1964) notes that the "Zambesian Zone," roughly
corresponding with the tropical part of the Zambezian Domain, is the richest
of the southern areas recognized by him in species of butterfly and that there
are numerous endemics.
[VOL. 65
458


BRENAN-TROPICAL AFRICA
The "miombo" is linked with a relatively high rainfall (700-1,600 mm per
annum) and occurs on a plateau of about 1,000-1,500 m elevation. Fanshawe
(1969) reproduces a map showing the hypothetical vegetation of Zambia at a
rainfall 500 mm per annum less than at present, entailing a spectacular decrease
in the present area of "miombo." Although Fanshawe (1969: 42) states that
"miombo" woodland is a "vigorous invasive vegetation type," I have seen in
Zambia "miombo" woodland areas closely similar in their tree cover, but very
different indeed in their understory of shrubs and herbs. In fact it may be
desirable to analyze separately in terms of composition and behavior the tree
and understory layers of "miombo" woodland. An invasive tree layer may
spread over lower layers floristically and historically different.
The "miombo" stands in strong contrast to the tree savannas in which there
is a prevalence of Acacia and Combretum, well represented in the Zambezian
Domain, usually at lower elevations and lower rainfall, but also much more
widespread in Africa and recognizably occurring in other regions and domains.
In the Zambezian Domain there are thus vegetation types as floristically
distinct and probably also as different in their history and origin as those in the
Afroriental Domain. In any analysis or comparison of floras or endemism it is
important that these differences are clearly recognized.
ENDEMISM OF AREAS AND INDIVIDUAL COUNTRIES
During the course of this study an attempt has been made to assess the rela-
tive richness in endemics of individual areas and countries in tropical Africa.
The areas correspond with the regional groupings of countries adopted in the
herbarium of the Royal Botanic Gardens, Kew, while the countries correspond
with the usual political boundaries. The only noteworthy exception is Came-
roon. The northwestern part (corresponding to the old British Cameroons) has
been completely dealt with in the second edition of the Flora of West Tropical
Africa (Hutchinson & Dalziel, 1954-1968), while the definitive Flore du Came-
roun is still very incomplete. It has thus seemed worthwhile to treat these two
areas separately, though recognizing that they are politically one. Approximate
areas in square kilometers are given after the name of each country. I am greatly
indebted for many of the data used in this assessment to the unstinted help of
the Conservation Unit at the Royal Botanic Gardens, Kew, especially Mr. G. Ll.
Lucas and Mr. John Hartshorne. Mr. Hartshorne in particular has been respon-
sible for most of the detailed and arduous extraction of data from published
works, particularly in 10A, 10D, and Za�re.
WEST AFRICA: AREA OF FLORA OF WEST TROPICAL AFRICA
Analysis of the total of taxa (Hutchinson & Dalziel, 1954-1968), species and
varieties being considered as separate taxa, shows the results enumerated in
Table 10.
These figures emphasize the regional or more narrow endemism characteris-
tic of so large a proportion of the flora of tropical West Africa. The families
with the largest number of taxa endemic to one country (over 25) are as follows:
1978]
459


ANNALS OF THE MISSOURI BOTANICAL GARDEN
Rubiaceae-107, Compositae-33,
Papilionaceae-52, Caesalpiniaceae-31,
Orchidaceae-48, Sterculiaceae-27,
Melastomataceae-38, Annonaceae-26,
Gramineae-37, Cyperaceae-25.
Euphorbiaceae-34,
In the following account totals of specific endemism are given for each
country, i.e., species endemic to that country alone, together with some remarks
on areas of special concentration and some representative or outstanding exam-
ples. Percentages are given of the total species of the Flora of West Tropical
Africa (7,072). These will of course be much higher if considered as percentages
of the country floras concerned.
Senegal (200,000 km2).-26 endemic species (0.37%). Examples: Abutilon
macropodum, Acalypha senegalensis, Berhautia senegalensis, Cyperus lateriticus,
Eriocaulon inundatum, Laurembergia villosa, Rhynchosia albiflora, Salicornia
senegalensis, S. praecox, Vernonia bambilorensis. Note the absence of forest
genera and predominance of herbs. Berhautia is apparently an endemic genus.
Clayton & Hepper (1974) identify "Senegal/Mali" as a center of endemism for
West African grasses.
Gambia (10,700 km2).-One endemic species, Rhinopterys spectabilis.
Guinea-Bissau (36,100 km2).-Apparently no endemism at the specific level.
Guinea Republic (250,000 km2).--88 endemic species (1.24%). Examples:
Adamea stenocarpa, Cailliella praerupticola, Djaloniella ypsilostyla, Fleurydora
felicis, Guyonia tenella, Impatiens bennae, Pitcairnia feliciana, Rinorea djalo-
nensis, Stonesia (3 species). Schnell (1968) has pointed out certain especially
important areas: the Nimba Mountains, Fouta Djallon, etc. The Guinea Repub-
lic is an important center of endemism: the number of endemic taxa is high,
and the genera Adamea, Cailliella, Djaloniella, Fleurydora, and Stonesia ail ap-
pear to be endemic. In addition the only African bromeliad, Pitcairnia feliciana,
occurs in this country. Clayton & Hepper (1974) identify "Guin�e/Sierra Le-
one" as an important center of endemism for West African grasses, although
"endemic" is by them equated with "endemic to the area of the Flora of West
Tropical Africa."
Mali (1,500,000 km2).-11 endemic species (0.15%). Examples: Combretum
nioroense, Gilletiodendron glandulosum, Letestuella chevalieri, Teclea ferrugi-
nea. In spite of its area Mali is outside the Guineo-Congo Region and is poor
in endemism.
Sierra Leone (72,300 km2).-74 endemic species (1.04%). Examples: Acioa
whytei, Afrotrilepis jaegeri, Byttneria guine�nsis, Clerodendrum whitfieldii,
Dovyalis afzelii, Gilbertiodendron aylmeri, Habropetalum dawei, Paepalanthus
pulvinatus, Tricalysia trilocularis. Somewhat less rich than the Guinea Republic,
and with apparently only one endemic genus, Habropetalum. Ayodele Cole
(1974) lists the few endemic species of the Loma Mountains/Tingi Hills in
Sierra Leone.
460
[VOL. 65


BRENAN-TROPICAL AFRICA
TABLE 10. Endemism in West Africa: area of the Flora of West Tropical Africa. Total
number of taxa: 7,172.
Region Number Percentage of total
Endemic to 1 country 1,001 13.96
Endemic to 2 countries 545 7.6
Endemic to 3 countries 328 4.57
Endemic to more than 3 countries 593 8.27
Endemic to Flora area plus 1 adjacent country outside 414 5.77
Occurring more widely 4,290 59.83
Liberia (111,500 km2).-59 endemic species (0.83%). Examples: Ancistro-
cladus pachyrhachis, Dichapetalum linderi, Dinklageodoxa scandens, Guibourtia
dinklagei, Hymenocardia intermedia, Jasminum dinklagei, Tetracera dinklagei.
The only endemic genus is the remarkable Dinklageodoxa, the only scandent
member of the Bignoniaceae indigenous to Africa.
Ivory Coast (322,000 km2).-41 endemic species (0.58%). Examples: Coffea
lemblinii, Cola attiensis, Cymnostemon zaizou, Hemandradenia chevalieri, Ho-
malium aubrevillei, Impatiens nzoana, Macaranga beillei, Sapium aubrevillei.
The only endemic genus appears to be Gymnostemon. The alleged endemic
genus and species, Vilbouchevitchia atro-purpurea A. Chev., is highly suspect
(see Hutchinson & Dalziel, 1954-1968, vol. 2: 74). Guillaumet (1967) discusses
the species endemic to the Ivory Coast and the region Ghana-Sierra Leone ("es-
p�ces sassandriennes"). Group I (17 species) is confined to the western Ivory
Coast (and often Liberia); Group II (75 species) to the region Ivory Coast-
Sierra Leone.
Ghana (238,300 km2).-43 endemic species (0.61%). Examples: Afrothismia
pachyantha, Aneilema mortonii, Cola umbratilis, Homalium angustistipulatum,
Millettia irvinei, Ochthocosmus chippii, Talbotiella gentii. Apparently no en-
demic genera.
Togo (56,000 km2).-20 endemic species (0.28%). Examples: Adenia triloba,
A. pulcherrima, Eugenia togoensis, Gutenbergia foliosa, Jaundea baumannii, Ri-
norea bussei, Streptocarpus kerstingii. No endemic genera.
Dahomey (112,600 km2).-11 endemic species (0.15%). Examples: Hibiscus
lonchosepalus, Jatropha atacorensis, Lepidagathis chevalieri, Polygala atacoren-
sis, Raphia humilis. No endemic genera.
Niger (1,247,000 km2).-Two dubious endemic species (Ipomoea ardissima,
Vigna marchalii).
Nigeria (877,000 km2).-It is instructive to consider this country under three
regional headings:
(a) Northern (as defined in Hutchinson & Dalziel, 1954-1968) (662,263 km2).
This is mainly in the Sudano-Zambezian Region (see Fig. 2). There are 39
endemic species (0.55%), examples of which are: Dissotis graminicola, Habe-
naria nigerica, Huernia nigeriana, Indigofera latisepala, Protea argyrophaea, Pen-
nisetum dalzielii, Psychotria dalzielii, Trochomeria dalzielii, Vernonia bauchi-
ensis. The lack of generic endemism, together with the prevalence of endemism
1978]
461


ANNALS OF THE MISSOURI BOTANICAL GARDEN
in widespread African genera, many extending to South Africa (e.g., Protea,
Huernia), is noteworthy.
(b) Western and Central (see Table 5) (125,656 km2). 38 endemic species
(0.54%). Examples: Begonia salisburyana, Brachystegia nigerica, Dissotis idan-
re�nsis, Memecylon meiklei, Psammetes nigerica. The low number of endemic
species, notwithstanding the fact that this area is mainly in the Guineo-Congo
Region, with rain forest well represented, is very significant, and is in strong
contrast with (c) Eastern below. The genus Psammetes is endemic.
(c) Eastern (see Table 5) (76,364 km2). 128 endemic species (1.81%). Ex-
amples: Allexis obanensis, Ancistrocladus uncinatus, Butumia marginalis, Cola
gigas, Crateranthus talbotii, Globulostylis talbotii, G. minor, Guaduella humilis,
Hibiscus grewioides, Pohliella flabellata, Talbotiella eketensis. The genera Bu-
tumia, Crateranthus and Globulostylis are endemic, and some of the species,
e.g., Cola gigas and Hibiscus grewioides are outstandingly distinct. This evi-
dence speaks for itself of the importance of eastern Nigeria as an area of high
endemism, and of the remarkable contrast with the preceding regions of Nigeria.
The area around Oban appears to be especially rich in endemics.
Cameroon (N.W.) (88,300 km2).-Out of 156 endemic species (2.20%), about
45 appear to be confined to the Cameroon Mountain (some also extending to
Bamenda). Examples of endemics are: Cameroon Mountain: Afrardisia oli-
gantha, Anthospermum cameroonense, Asparagus longipes, Begonia jussiaecarpa,
Camptostylus ovalis, Deschampsia mildbraedii, Genyorchis macrantha, Helichry-
sum cameroonense, Hypseochloa cameroonensis, Mikaniopsis maitlandii, Pepero-
mia vulcanica, Silene biafrae, Streptocarpus elongatus, Succisa trichotocephala,
Uebelinia hispida; other regions of Cameroon: Bafutia tenuicaulis, Cylico-
morpha solmsii, Inversodicraea keayii, Medusandra richardsiana, Oxyanthus se-
tosus, Vincentella brenanii. The genera Hypseochloa and Bafutia appear to be
endemic.
The richness of eastern Nigeria is thus continued into Cameroon. It is
remarkable that some of the outstanding rain forest endemics in each area have
not been found in the other. Probably this does not just reflect undercollecting
but is a real indication of narrow areas of distribution apparently not linked
with any major topographical barriers.
The presence on the Cameroon Mountain of endemic species of genera of
northern affinity is interesting, e.g., Deschampsia, Silene, Succisa, but probably
reflects endemism in a more generalized African mountain flora rather than any
special northern affinity particular to the Cameroon Mountain. The endemics
on the mountain are found at ail altitudes, from the rain forest on the lower
slopes upwards, and this is reflected in the mixture of floristic elements repre-
sented in the examples given.
Fernando Po (Macias Nguema) (1,000 km2) .-49 endemic species (0.69%),
a high figure considering the small area. Examples: Cyathula fernandopoensis,
Leptonychia (4 species), Melothria fernandensis, Psychotria crassicalyx, P. epi-
phytica, Sabicea urbaniana, Streptocarpus insularis. Exell (1944: 51) quotes a
figure of 99 endemic species for Fernando Po, but this was before the thorough
revision undertaken for the second edition of the Flora of West Tropical Africa.
462
[VOL. 65


BRENAN-TROPICAL AFRICA
CENTRAL AND WEST-CENTRAL AFRICA
No single flora or enumeration has been written to cover all this area and
therefore evidence has had to be assembled from a number of different works,
mostly more or less incomplete, and these are cited under each country.
Chad (1,284,000 km2).-I have not found any complete survey of the flora
of Chad. Special studies have been made of the Saharan mountain ranges of
Tibesti and Ennedi. Wickens (1976: 79) has examined the Tibesti specimens
collected by Qu�zel and as a result estimates 38 endemic species from this range
(7.2%). Gillet (1968) estimates 1% endemism for Ennedi. Elsewhere endemism
is likely to be very low or none.
Central African Republic (493,000 km2).-Sillans (1958: 197-203) estimated
the total flora at about 3,600 species. Of these about 1,000 occur in the rain
forest, with about 10 endemic (1%) and 2,600 in the savanna, with about 90
endemic (3.5%). The latter figure especially is probably too high. Examples of
species endemics are: Centaurea tisserantii, Combretum tisserantii, Monadenium
chevalieri, Oryza tisserantii. The two allegedly endemic genera, Heimodendron
and Tisserantodendron, are neither now maintained.
Cameroon (475,000 km2).-The Flore du Cameroun (Aubr�ville & Leroy,
1963-1975) is still very incomplete. Out of an estimated total of 6,500 flowering
plants, 882 species have been described-about 15%. Of these, 110 (12.47%)
are endemic. Multiplying this by a factor of 5.7, an approximate total figure of
627 endemics is obtained for the whole country. Generic endemism appears to
be surprisingly low. In the estimated 15% of the flora dealt with, only one
endemic genus, Oriciopsis, appears. Eurypetalum is almost endemic, being con-
fined to the Cameroon and Equatorial Guinea. The country has great tracts of
rain forest, and most of the endemic species are in the forest. Some examples
are: Balsamocitrus camerunensis, Aulotandra kamerunensis, Telfairia batesii,
Dialium zenkeri, Talbotiella batesii, Brachystegia cynometroides, Pimpinella le-
dermannii, Strychnos mimfiensis, Octoknema dinklagei, Ocotea angustitepala,
Scyphocephalium chrysothrix and no less than 26 species of Beilschmiedia. The
last-named is a most extraordinary concentration and deserves further study.
Equatorial Guinea (Rio Muni) (28,000 km2).-Unfortunately this very inter-
esting country has not received the critical botanical study that it merits. Guinea
L6pez (1946) gave a preliminary catalogue of the plants, but, as he admitted,
it included many species that might occur but were not so far actually known
to do so. Clearly, statistics of the known flora are not yet possible, but it is likely
to be rich. One genus at least, Chonopetalum (Sapindaceae), is believed to be
endemic. Of three new genera established by Guinea L6pez (1946) only Des-
mogymnosiphon has not been sunk.
S�o Tom� (with Principe 964 km2).-Of a total flora of 556 species (Exell,
1944) 108 are endemic (19.4%). Heteradelphia paulowilhelmia is a monotypic
endemic genus. Endemic species include: Begonia (6 species), Calvoa (6 spe-
cies), Crossandra thomensis, Impatiens (2 species), Lobelia barnsii, Philippia
thomensis, Pilea manniana, Podocarpus mannii, Rinorea (3 species), Staudtia
pterocarpa, Thunbergianthus quintasii.
463
1978]


ANNALS OF TIE MISSOURI BOTANICAL GARDEN
Principe (with S�o Tom� 964 km ).-Out of 276 species 35 are endemic
(12.7%) (Excll, 1944). Principina grandis is a monotypic endemic genus. En-
demic species include: Agelaea (4 species), Anthocleista stenantha, Casearia
mannii, Chrysophyllum calophyllum, Dracaena monostachya.
Annobon (17 km-2).-Out of 115 species 17 are endemic (14.8%), (Exell,
1944). No endemic genera, but species include: Agelaea annobonensis, Calvoa
uropetala, Fagara annobonensis, Lachnopylis annobonensis, Rhynchelytrum
reynaudioides.
Gabon (267,000 km ).-The 23 parts of the Flore du Gabon so far published
(Aubr�ville & Leroy, 1961-1972) (excluding the ferns) contain a total of 1,333
species, of which 243 are endemic (22.29%), a remarkably high figure. Unfor-
tunately no estimate of the total flora is available, but Gabon is evidently, in
terms of endemism, one of the richest countries in tropical Africa, although
less so than Angola. In some groups, e.g., Leguminosae-Caesalpinioideae and
Sapotaceae, the percentage endemism is much higher, 29.93% and 52.94%
respectively.
In the parts so far published there are no less than 13 endemic genera: Le-
testua, Lecomtedoxa (5 species), Tulestea (3 species), Iridosma, Temnopteryx,
Neochevalierodendron, Sindoropsis, Augouardia, Paraberlinia, Pseudartabotrys,
Coleactina. The other endemic species include: Cola (8), Dacryodes (5), Im-
patiens (5), Commelinidium gabunense, Costus (6), Beilschmiedia (7), Ocotea
gabonensis, Gilbertiodendron (8), Monopetalanthus (5), Uvaria (9), Anonidium
floribundum, A. le-testui, Polyceratocarpus pellegrinii, and very many others,
mostly in the forests.
Congo (Brazzaville) (342,000 km2n-).-Insufficient evidence for assessment.
However, the country is likely to be comparable in endemism to the Gabon. The
total number of species occurring has been estimated at about 4,000 (Bouquet,
1976).
Cabinda (7,770 km2).-Exell & Gon�alves (1973) analyzed a sample of 257
species and found 31 endemic (12.1%). This proportion is considerably lower
than those for the Gabon and Angola and suggests that the richness of the rain
forest flora may be decreasing southwards from the Gabon through Congo
(Brazzaville) to Cabinda. Endemic species include: Begonia mayombensis,
Memecylon (6 species), Acioa dawei, Impatiens gossweileri, Crudia gossweileri,
Chytranthus angustifolius.
Za�re (including Rwanda and Burundi) (2,345,000 km-).-This vast area has
an estimated flora of about 10,000 species. Of these, 3,921 have been described
in the first 10 volumes of the Flore du Congo Belge et dlu Ruanda-Urundi and
the Flore d'Afrique Centrale (Robyns, 1965, supplemented by personal count-
ing). A sample of about 39% is thus available. Of the total of 3,921, 1,280 are
endemic to Za�re (including Rwanda and Burundi), a remarkable proportion of
32.64%-apparently the highest proportion of endemism of any area in tropical
Africa. Exell & Gon�alves (1973), working on a much smaller sample (1,218
species), estimated 38.2% for Za�re. In this sample twelve endemic genera are
included: Hyalosepalum, Toussaintia, Afroguatteria, Gilbertiella, Thonnera, Ro-
464
[VOL. 65


BRENAN-TROPICAL AFRICA
Territoires phytog�ographiques par W. ROBYNS
1. C�tier. - 2. Mayumbe. - 3. Bas-Congo. - 4. Kasai. - 5. Bas-Katanga.
6. Forestier Central. - 7. Ubangi-Uele. - 8. Lac Albert. - 9. Lacs �douard
et Kivu. - 10. Ruanda-Urundi. - 11. Haut-Katanga.
FIGURE 5. Phytogeographical regions of Za�re, Rwanda and Burundi as recognized by
Robyns (1965). From Flore du Congo, du Rwanda et du Burundi Vol. 10 (1963).
bynsiophyton, Santaloidella, Pseudonwcrolobium, Pynaertiodendron, Mildbrae-
diodendron, Karina, Lebrunia. Ail or alnost ail are monotypic.
The endemism of Za�re, Rwanda and Burundi repays closer analysis, as it is
very far from being evenly distributed throughout the area. Robyns (1965)
recognized 11 regions (Fig. 5) and I have tabulated the distribution of the 1,280
endemics on this basis in Table 11.
Of these areas, 2 and 6 are mainly rain forest, the others mainly savanna.
Area 2 is small, 6 large. The high number of endemic species in area 6 is note-
465
1978]


ANNALS OF THE MISSOURI BOTANICAL GARDEN
TABLE 11. Za�re, Rwanda and Burundi endemic species under the regions recognized
by Robyns (1965).
Number of
Region Endemic Species
1. C�tier 6
2. Mayumbe 4
3. Bas-Congo 45
4. Kasai 80
5. Bas-Katanga 36
6. Forestier Central 233
Number of
Region Endemic Species
7. Ubangi-Uele 11
8. Lac Albert 10
9. Lacs �douard et Kivu 62
10. Rwanda-Burundi 26
11. Haut-Katanga 312
12. In more than one of areas 1-11 275
worthy, but not unexpected, but the very high figure for Haut-Katanga is re-
markable. This latter area, a mixture of forest and savanna in the Zambezian
Domain is clearly a noteworthy center of endemism. There are for example, 69
endemic species of Crotalaria alone and 23 of Cissus (including Cyphostemma).
It may well be that this area of high endemism is a continuation of the high
endemism found in Angola. No such prolific generic speciation is to be met in
area 6, Forestier Central. Here the highest numbers of endemic species under
individual genera are Salacia (16), Dichapetalum (14), Trichoscypha (5). Cola,
which one might expect to be well represented, has only one species endemic
to area 6.
It is evident that Rwanda and Burundi have only a modest endemism. Le-
walle (1975) discusses 8 endemic species in a mountain valley in Burundi near
Lake Tanganyika.
NORTHEASTERN TROPICAL AFRICA
This area comprises the Sudan, Ethiopia, and Somalia. The island of Soco-
tra, well known for its numerous and remarkable endemics, is not dealt with in
the present paper. No single flora or checklist covers the entire region. There is
a flora of the Sudan (Andrews, 1930-1956) and a checklist for the remainder
(Cufodontis, 1953-1972). Unfortunately Andrews (1950-1956) does not give
any data on general distribution, and Cufodontis (1953-1972), reflecting politi-
cal changes during the course of the work, changed the boundaries of the areas
under which species were recorded when more than halfway through the work.
This results in the distribution data published after the change being often not
directly comparable with those before, and also that very frequently it is not
possible easily to assign species dealt with before the change to Ethiopia or
Somalia as the boundaries then do not correspond with the boundaries now. It
therefore seemed best to limit the present statistical survey to the sample of
2,638 accepted species published subsequent to the change, which are assignable
to present-day political entities. The total flora of Cufodontis's region is esti-
mated at 6,323 species (Cufodontis, 1953-1972: 1624).
Sudan (2,500,000 km2 ).-Andrews (1950-1956) enumerated 3,137 species in
all. For reasons given above, any exact estimate of endemism is not feasible. A
cursory personal inspection of the flora leads me to estimate that the total num-
ber of species endemic to the Sudan is unlikely to exceed 50, which would be
466
[VOL. 65


BRENAN-TROPICAL AFRICA
1.6% of the above total. The work of Andrews is uncritical and incomplete and
the percentage is likely to be reduced rather than increased. The Jebel Marra
is an interesting and isolated mountain massif, but Wickens (1976) enumerated
only 11 endemic species (1.2% of the total flora of Jebel Marra). Species endemic
to Jebel Marra include: Plectranthus jebel-marrae, Celsia sudanica, Kickxia
dibolophylla and Gnaphalium marranum. Other Sudan endemics include Eu-
phorbia consobrina, Jatropha gallabatensis, Combretum kabadensis. However, a
new critical flora, with a reassessment of the alleged endemics, would be well
worthwhile.
Ethiopia (including Eritrea) (1,000,000 km2).-Of the 2,638 species ana-
lyzed (see above), 553 were endemic to Ethiopia (20.96%). There is no doubt
that Ethiopia, with its size and very diversified topography including great
mountain ranges, has a rich endemic element in its flora. However, Cufodontis
did not revise the flora critically and until this has been done, the proportions
remain suspect and probably too high. There are a number of endemic genera
including Pseudozoysia, Simenia, Leptagrostis, Odontelytrum, Erythroselinum,
Gymnosciadium, Lamellisepalum, Afrovivella, Sabaudiella, etc.
Djibouti (Territory of the Afars and Issars) (23,000 km2).-This small coun-
try, formerly French Somaliland, has apparently very few endemic species.
Only one, Kohautia gracillima, was found among the 2,638 mentioned above.
Somalia Republic (700,000 km2).-Of the 2,638 species analyzed (see above)
259 are apparently endemic (9.82%). However, the 2,638 species include a very
large number not found in Somalia, so that the percentage is against a regional
total comprising other countries and not of the Somalia flora as such, which is
likely in total to be much smaller than that of Ethiopia. For this reason the true
percentage of endemism is likely to be considerably higher, though against this
must be set the fact that the Somali flora is as much in need of critical revision
as that of Ethiopia, and there is no doubt that a significant proportion of the
allegedly endemic species will not be maintained. Nevertheless the flora of the
Somali Republic is a remarkable one with very many outstandingly distinct
species found nowhere else. Significant proportions of the flora seem to be
restricted to the Somali Republic but also occur in Harar Province of Ethiopia
(21 out of 2,638: 0.8%), or Socotra (10 out of 2,638: 0.38%), or Arabia (38 out
of 2,638: 1.44%). Furthermore, a very considerable number of species are re-
stricted to the Somali Republic and northem Kenya, though I do not have exact
figures for this.
Genera endemic to the Somali Republic include: Lagenantha, Pleuropter-
antha, Pseudodigera, Neocentema, Puccionia, Cordeauxia, Dicraeopetalum, Scas-
sellatia, Hypseloderma, Sennia, Rumicicarpus, Lithocaulon, Rhytidocaulon, Hy-
peraspis, Ghikaea, Chamaeacanthus, Lindauea, Golaea, Eionitis, Paolia. This
remarkable list, which will probably be lengthened, is a testimony to the interest
and uniqueness of the flora of the Somali Republic.
EAST AFRICA: AREA OF FLORA OF TROPICAL EAST AFRICA
The published parts (to 1977) of the Flora of Tropical East Africa account
for an estimated 40% of the total flora. However, this is already a sufficiently
1978]
467


ANNALS OF THE MISSOURI BOTANICAL GARDEN
TABLE 12. Endemism in East Africa: area of the Flora of Tropical East Africa. Total
number of species: 4,412 (X 2.5 = 11,030).
Region Numnber Percentage of 4,412
Endemic to 1 country 897 20.33
Endemic to 2 countries 199 4.51
Endemic to 3 countries 55 1.25
Endemii to more than 3 countries 10 0.23
Occurring more widely 3,253 73.73
large sample to give interesting evidence about the distribution of endemism in
tropical east Africa without further adjustment. By multiplying by an appropri-
ate factor (2.5) an approximation to totals can be obtained. In Table 12 per-
centages are given representing the relationship of the total endemism of each
country to 4,412, the size of the regional sample.
The mention of "country" or "countries" in Table 12 means countries within
the area of the Flora of Tropical East Africa. The high proportion of species
endemic to one country contrasts with the much smaller figures for those species
endemic to the area of the Flora, yet more widely spread than in one country.
In other words, this emphasizes the prevalence of narrow species ranges in East
Africa (cf. the proportion for West Africa in Table 10), and it is probably corre-
lated with the much more diverse topography of East Africa, the frequency of
isolated mountain ranges and peaks, and the fragmentation and isolation of
forest areas.
A more detailed discussion of the endemism follows under each country.
Uganda (236,800 km ").-12 endemic species (0.27%) (x 2.5 =30). These
are distributed among the provinces recognized in the Flora of Tropical East
Africa as follows: Ul, Northern: 3; U2, Western: 3; U3, Eastern: 1; U4, Bu-
ganda: 3. Two species occur in more than one province. Generic endemism is
apparently nil or very small. The poverty in endemism of the Uganda flora is
in very marked contrast to Kenya and Tanzania.
Kenya (582,600 km2).-106 endemic species (2.40%) (x 2.5 = 265). The dis-
tribution according to the provinces recognized in the Flora of Tropical East
Africa is as follows: Kl, Northern Frontier: 22; K2, Turkana: 1; K3, Rift Valley:
3; K4, Central: 26; K5, Nyanza: 2; K6, Masai: 4; K7, Coast: 23.
In addition, 25 are endemic to Kenya but found in more than one province.
The above figures bring out the richness of K1, mainly because of the remark-
able Somali element in the Kenya flora, especially in the northeast; the poverty
of western Kenya (K2, 3, 5); the strong endemism of K4, in considerable mea-
sure due to the presence of Mount Kenya; and the richness of the coastal flora
of K7. Generic endemism is low in Kenya but Loewia with two species and
Dibrachionostylus kaessneri (K4) are to be mentioned.
The coastal flora in K7, although with a number of endemies, is really a
northwards extension of a coastal element going far south into Tanzania and
often extending on to Zanzibar. Of the 40% sample 77 species are endemic to
K7 and also extend into the coastal areas of Tanzania and Zanzibar. Some gen-
468
[VoL.. 95


BRENAN-TROPICAL AFRICA
era are endemic to this general area: Angylocalyx, Asteranthe, Lettowianthus,
Mkiluea, Ophrypetalum. Lucas (1968) gives a useful list of species found in
the Kenya coastal forests, including most of the endemics, which unfortunately
are not distinguished as a separate category.
Tanzania (939,400 km-).-449 endemic species (10.18%) (x 2.5= 1,122).
The distribution according to the provinces recognized in the Flora of Tropical
East Africa is as follows: T1, Lake: 13; T2, Northern: 20; T3, Tanga: 56; T4,
Western: 24; T5, Central: 22; T6, Eastern: 77; T7, Southern Highlands: 70;
T8, Southern: 56.
The figures show that Tanzania is outstandingly rich in endemics and that
every province has a significant quota, though there seems a tendency for the
numbers to be lower towards the northeast. T2 is noteworthy for the presence
of Kilimanjaro, the highest peak in Africa; T3 contains the Usambara Mountains,
with their sadly depleted but remarkably rich forests; T6 contains the Uluguru
Mountains, as rich as the Usambaras, if not more so; T7 contains a rich endemic
element in the high altitude grasslands of the mountain regions of southwestern
Tanzania; and T8 contains a remarkable area towards the Mozambique fron-
tier, very rich in endemism though apparently not isolated topographically and
badly in need of further investigation in the field.
Polhill (1968) gave very useful lists of the known endemics from the Usam-
bara Mountains (T3), the Uluguru and Nguru Mountains (T6), and the Lindi
District (T8). It is therefore perhaps unnecessary to repeat here lists of en-
demic species, but mention should be made of generic endemism:
Usambaras: Cephalosphaera usambarensis, Dolichometra leucantha, Engler-
odendron usambarense, Platypterocarpus tanganyikensis.
Ulugurus: Adenoplusia ulugurensis, Dionychastrum schliebenii, Pseudoneso-
hedyotis bremekampii, Rhipidantha chlorantha.
T8: Farrago racemosa, Primularia pulchella, Vismianthus punctatus.
Elsewhere in Tanzania: Apochiton burttii, Peterodendron ovatum, Stuhl-
mannia moavi.
This total of 15 genera, which will probably be increased, may be compared
with 6 for Nigeria and 5 for Guinea.
Zanzibar (1,658 km2) .-As one might expect from its size and proximity to
Tanzania, there is little endemism. Ipomoea zanzibarica and Aeschynomene
zanzibarica are said to be endemic in the sample of the East African flora ana-
lyzed (2 species, x 2.5 = 5) (0.05%).
SOUTHERN AND SOUTH-EAST TROPICAL AFRICA: AREAS OF FLORA ZAMBESIACA
AND CONSPECTUS FLORAE ANGOLENSIS
The area numbered 10E is altogether within the area of the Flora Zambesi-
aca (Exell, Wild & others, 1960-1971), except for Angola. Unfortunately the
floras are still far from complete. However, a sufficiently large proportion of
each is in print to allow useful statistical deductions to be drawn.
Of Flora Zambesiaca the first two volumes are finished, together with part 1
of volume 3 and part 1 of volume 10. The pteridophyta, although published,
have not been included here as the factors governing their distribution seem
1978]
469


ANNALS OF THE MISSOURI BOTANICAL GARDEN
TABLE 13. Statistics based on Flora Zambesiaca.
Approximate total of species in the whole flora (Exell, 1971) 6,000
Total of species, native and naturalized, in published sample surveyed 1,402
Percentage surveyed 23.37%
Species endemic to the area of the flora but not confined to one country 50 (3.57%)
Species endemic to single countries in the area of the flora 142 (10.13%)
often rather different from those of flowering plants, and their inclusion might
well be a distorting factor in the rather limited samples dealt with here. Table
13 summarizes some of the main points.
It will be evident that the general percentages of endemism in the Flora
Zambesiaca area are considerably lower than those for West Africa (10A) and
East Africa (10D). As we shall see below, they are also considerably lower than
for Angola. It is possibly worth noting that the total percentage endemism for
the Flora Zambesiaca area is in close agreement with a figure of 13.2% quoted
by Exell & Gon�alves (1973: 109) on a smaller sample.
The statistics drawn from the Con.spectus Florae Angolensis will be consid-
ered under Angola among the accounts which follow of the individual countries
in the area. Under each country the number of endemic species is given as
revealed in the sample of 1,402 species surveyed, together with an estimate of
the total arrived at by multiplying the number in the sample by a factor of 4.3,
which represents the conversion of 1,402 to the estimated total of 6,000.
Mozambique (785,000 km2).-51 endemic species (3.64% of sample); esti-
mated total (51 x 4.3) of 219. Examples are: Xylopia torrei, Bombax mos-
sambicense, Glyphaea tomentosa, Impatiens suffulta, Dombeya leachii, Entada
mossambicensis, Xylia mendoncae. Generic endemism is low. Two monotypic
genera occur: Krauseola and Thespesiopsis, of which the former may well be
found also in northern Natal. Fryxell (1968) does not, however, consider the
latter genus separable from Thespesia.
Mozambique has marginally more narrow endemics than any other country
dealt with in Flora Zambesiaca (Zambia being next). Much of this comparative
richness is probably due to a southerly extension of the high endemism noted
above as a feature of the coast of Tanzania. The Usambara-Zululand Domain
extends southwards from Tanzania along the coast to South Africa. However,
in particular there is a region in northern Mozambique, extending to the Lindi
District in Tanzania where there is a remarkable concentration of local ende-
mism. The exact boundary of the area and the explanation of its existence still
remain to be discovered. During the survey of the published parts of Flora
Zambesiaca, at least 14 species were noted as restricted to northern Mozam-
bique and southeastern Tanzania. These include such species as Capparis orth-
acantha, Nectaropetalum carvalhi, Salacia orientalis, Pseudoprosopis euryphylla,
Mimosa busseana, Acacia latistipulata. This area is much in need of further
study in the field.
Malawi (93,900 km2).-16 endemic species (1.14% of sample); estimated
total (16 x 4.3) of 69. Examples are: Dasylepis burtt-davyi, Dovyalis spinosis-
470
[VOL. 65


BRENAN-TROPICAL AFRICA
sima, Dombeya calantha, Impatiens quisqualis, Helichrysum whyteanum, Clutia
brassii, Rhus monticola. The low level of general endemism is noteworthy and
also the apparent total absence of endemic genera. Of the 16 species, six are
apparently restricted to Mt. Mlanje. Wild (1964) lists 30 species endemic to
Mlanje, and this has been commented on by Chapman & White (1970: 74) who
note "the small number and critical nature" of the tree species represented and
state that in the Afromontane tree flora "endemism is very weak and vicariism
scarcely occurs." The affinity of the endemic element on Mt. Mlanje is in my
view to be sought as much in South Africa as in tropical Africa. Similar opinions
are voiced by Wild (1964) for the Chimanimani Mountains in Rhodesia.
Rhodesia (389,300 km2).-22 endemic species (1.57% of sample); estimated
total (22 x 4.3) of 95. Examples are: Homalium chasei, Dianthus chimani-
maniensis, Cyphostemma milleri, Bersama swynnertonii, Rhus wildii, Acacia
chariessa.
As for Malawi, the general level of endemism is low. The level of generic
endemism is difficult to estimate, but probably very low, though including
the outstandingly distinct Triceratella from the Limpopo River basin. Unfortu-
nately only one gathering has been made (Brenan, 1961).
A considerable proportion of the total of endemic species in Rhodesia is
likely to be confined to the Chimanimani Mountains. Wild (1964) lists 41 spe-
cies endemic to this range, representing 4.6% of the total of 859 species occur-
ring above 1,220 m (4,000 ft). Edaphic factors seem of especial importance in
controlling the occurrence of endemism in Rhodesia. Wild (1964) notes that ail
or nearly ail of the Chimanimani endemics are confined to areas of quartzite
rock, and not found in the areas of shales and schists also well represented in
this range. Wild (1965) listed 20 taxa, including at least 11 species endemic to
the serpentines of the Great Dyke in Rhodesia. That serpentine rock often sup-
ports an unusual or unique flora is a well-known and widespread occurrence.
Wild adduces evidence that high concentrations of nickel and chrome may be a
leading factor. Aristida hispidula and Rhus wildii, for example, are endemics
apparently linked with chrome.
Zambia (729,900 km2).-49 endemic species (3.5% of sample); estimated
total (49 x 4.3) of 211. Examples are: IlHpericum oligandrum, Monotes dis-
color, Dombeya brachystemma, Triumfetta (3 species), Biophytum richardsiae,
Cissus fanshawei, Cyphostemma (6 species), Sorindeia rhodesica, Entada doli-
chorrhachis, Ozoroa (4 species). Generic endemism appears to be low. The
remarkable grass genus Richardsiella was described from the Abercorn area and
the monotypic genus Rastrophyllum has been recently described (Wild & Pope,
1977) from Mwinilunga. The passifloraceous genus Viridivia, although not en-
demic to Zambia, appears only to occur elsewhere in southwestern Tanzania.
The distribution of Ageratinastrum seems similar. The evidence is at present
inadequate to comment clearly on the local distribution of endemism in Zambia,
but out of the 49 endemics, 6 were restricted to the Solwezi-Mwinilunga area
and 8 to the Abercorn area near Lake Tanganyika. It is possible, though un-
likely, that this is partly due to both areas having been comparatively well
collected.
471
1978]


ANNALS OF THE MISSOURI BOTANICAL GARDEN
Angola (1,246,700 km2).-Exell & Gon�alves (1973) have carried out a very
useful survey on the endemism of the flora of Angola, using as a basis volumes
2-4 of the Conspectus Florae Angolensis (Exell et al., 1937-1970), published
during the period 1954-1970. The first volume was not used as being "now
considerably out-of-date." The sample comprised 1,379 species. Ferandes
(1971) estimated that the first four volumes of the Conspectus Florae Ango-
lensis represented 40% of the total; roughly 10% of the total flora is thus dealt
with in each volume, and Exell & Gon�alves's sample is thus roughly 30%.
Out of the 1,379 species, 378 (27.3%) were found to be endemic. This is a
remarkable proportion-the more so in comparison with the other countries
under 10E (3.64% for Mozambique and 3.5% for Zambia being the highest).
The endemics were found to be most numerous in Huila, Benguela, and Bi�
districts, respectively. It was considered that information was insufficient to list
"district" endemics. Cabinda was, incidentally, not included in this exercise.
In the first four volumes of the Conspectus, eight genera are given appar-
ently endemic to Angola (excluding Cabinda): Mischogyne, Sedopsis, Caulo-
carpus, Carrissoa, Aizoanthemum, Spuriodaucus, Aframmi, Pseudoselinum. As-
suming on the basis given above that this is roughly 40% of the total, Angola
appears to be one of the most important centers of endemism in southern trop-
ical Africa.
COUNTRY ENDEMISM IN TROPICAL AFRICA
SUMMARY AND CONCLUSIONS
One of the objectives in the preceding pages has been to assemble the com-
ponent parts in such a way that they might be put together into a more general-
ized coherent picture of the geographical distribution of endemism in tropical
Africa.
The evidence set out in the last section of this paper is obviously very
uneven. Sometimes it is based on a complete survey, but more often on more
or less partial ones; sometimes the survey has been made for a single country,
sometimes for a region. The deficiencies are only too manifest. Nevertheless it
seemed worth while to make an attempt to bring the data together on a common
basis.
Numbers of endemic species are important, but are inadequate alone. The
Sudan with about 50 endemics is numerically richer than Principe with 35, but
the concentration in the latter is far greater. Area is thus also important. Some
combination of number and area seemed necessary. Evidence exists, as we have
seen, either for exact figures or at least for approximate estimates of the total
numbers of endemic species in most of the countries considered here. By divid-
ing the number of endemic species (E) into the total area in square kilometers
(A) of the country an Area-Endemism Index can be obtained varying from 0
to infinity.
The results are obviously unreliable in detail. In particular, probably no
country has an even area spread of endemism-some areas in cach country are
richer than others. Yet the results are probably of the right order of magnitude
and may serve to provide an objective though imperfect basis for assessing rela-
472
[VOL. 65


BRENAN-TROPICAL AFRICA
TABLE 14. Area-Endemism Index for tropical Africa. Figures followed by (e) represent
an estimate based on a sample.
Country No. of Endemic Species A/E Index
Senegal 26 7,692
Guinea Bissau 0 -
Guinea Republic 88 2,840
Mali 11 136,363
Sierra Leone 74 977
Liberia 59 1,890
Ivory Coast 41 7,854
Ghana 43 5,541
Togo 20 2,800
Dahomey 11 10,236
Niger 2 ? 623,500
Nigeria
N. 39 16,981
W. & C. 38 3,307
E. 128 596
Cameroon (N.W.) 156 566
Fernando Po (Macias Nguema) 49 2,041
Chad 50 (e) 25,680 (e)
Central African Republic 100 (e) 4,930 (e)
Cameroon 627 (e) 757 (e)
Equatorial Cuinea ?
Sio Tom�e 143 7
Principe j
Annobon 17 1
Gabon 1,115 (e) 239 (e)
Congo (Brazzaville) ? ?
Cabinda 310 (e) 251 (e)
Za�re 3,200 (e) 733 (e)
Sudan 50 50,000
Ethiopia 1,105 (e) 905 (e)
Djibouti 2 (e) 11,500 (e)
Somali Republic 518 (e) 1,351 (e)
Uganda 30 (e) 1,893 (e)
Kenya 265 (e) 2,198 (e)
ranzania 1,122 (e) 837 (e)
Zanzibar 5 (e) 327 (e)
Mozambique 219 (e) 3,584 (e)
Malawi 69 (e) 1,360 (e)
Rhodesia 95 (e) 4,097 (e)
Botswana 17 (e) 33,823 (e)
Zambia 211 (e) 3,459 (e)
Angola 1,260 (e) 989 (e)
tive endemism. In Table 14 the results are set out. The figures for endemism when
followed by "(e)" represent an estimate based on a sample.
The basis of certain of the above figures in the left-hand column nceds
stating. For Ethiopia it is assumed that 553 endemics (see p. 467) represents a
sample of about half the total, and the same applies to Somalia (see p. 467 when
the figure 259 is multiplied by 2 for the above table).
The distribution of endemism in Nigeria is so uneven (see p. 461-462) that an
overall figure would be misleading and the A/E Index is therefore given for the
three regions individually recognized.
473
1978]


ANNALS OF THE MISSOURI BOTANICAL GARDEN
[VOL. 65
SC
S' '
I I1I t f l l l
1 1 1 ' .� I I I I ' 1 f' " ' � 1 1 �
o . 1 1 1 1 1 1 1 l l l l l l l l l l 'r l l l l l l l l l l l
1 1 1 1 1 1 1 ( I I I I 1 1 1 (. I I I I I I I I 1 1 1 1
11 1 11 1 1 1 1 1 '1 111 ( 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1111 i 11 11'l 11 It 1 11 1 1 1 11 1 1 1 1 1f 1 1 1I
1 1 1.1 1 1, ,1, II 1 1 1 I L I I I I I I I i l I I I I I I I I I I
pp jjll a'~ i,'ill IIIIIIIIIII'Ilfllllll1111 IIIIIIIIII:
1 U l �1 , 1 lt1 11 1�t r11�1 I I I F 1 1 t 1 11 1 11 1 11 1 11 1 11 ( II I ,II I , , I I I I I I 1 11 11II I I I I IIII I I I111 11
' IIIII��III��I 1j1 1j1 'l lll 111 (11 11 111 111 ' 'f1 1f I'l' '1{ 111 111 111 111 11' 111 111
( 1, (1 1 1 1 11 1 1 1 1 1 1111 11 1 1 � 1 1 1 1 1 1 1 1 1 t1111 11 11 11 1 1 11 11 111111 1 1
� , a' 1 1 1 1 1 1 1 1 1 1 1 1 1, .1 1 1 1 1 1 1 1 1 1 1 1 1
I IInl l Ii ,, I I � I I I ,1 I I 1 "i 11 I 11 1 1 1 1 . r 1 1 1 1 1 11 1 1 1 1
1 '''I '11 ll l iIIII I I ' I l1 1 1' 1 I
` � � � ., c,", "~ ~ " ee ", ( 1,11111 1 1111 I I .
",, 111111 c o Il
var hi h � No , 1 1 I I I I I , I I I I � I. i
I I I I ' I Ii ( 1 11 1111 1111 1 11, 1
unknow 1 :11111,ii Iii I i I i i I ii i 1 1N
\ o'
FIGURE 6. Categories of endemismn in tropical Africa assessed by countries based on
Table 12. Although Za�re (Congo on map) on this basis is a country of high endemism, the
high overall assessment reflects two separate areas especially rich: Bas-Katanga joiining Angola
and Tanzania; and Forestier Central adjoining Congo-Brazzaville which is probably rich in
endemism as is the adjacent Cabon.
The above figures enable a rough categorization to be made:
(a) Very Low
(b) Low
(c) Medium-High
(d) Very High
: 10,000 or more
: 4,000-10,000
: 1,000-4,000
: 0-1,000
474
ri io i
a ^aF
1M r
K � w
I ,.,
rars nrrii�iic i
,, L
+ ., so ss


BRENAN-TROPICAL AFRICA
The distribution of the categories within tropical Africa is given on the
accompanying map (Fig. 6).
It is hoped that the various incomplete regional floras or lists for tropical
Africa may be finished as soon as possible so that a clearer and more exact pic-
ture may be obtained of this aspect of its flora. On the basis its future manage-
ment and preservation may be undertaken on a basis of better evidence than
has hitherto been available.
With the speed of destruction or modification of tropical vegetation types
under the pressures of increasing population and modern technology, the need
for urgent conservation measures is increasingly pressing, but the evidence upon
which to establish priorities has been often imperfect or lacking. Obviously the
degree of endemism (or uniqueness) in any given area is a most important fac-
tor. Where endemism is low the danger of irreparable loss is lessened, and vice
versa. The best and most economical way of conserving threatened species is
to conserve the habitat. In the past the establishment of reserves or national
parks has often been made with insufficient account taken of the uniqueness of
what is preserved and deserving areas have been neglected. A knowledge of
local endemism will help to make a better basis for future policy.
LITERATURE CITED
AMADON, D. 1973. Birds of the Congo and Amazon forests: a comparison. Pp. 267-277,
in B. J. Meggers, E. S. Ayensu & W. D. Duckworth (editors), Tropical Forest Ecosys-
tems in Africa and South America: A Comparative Review. Smithsonian Inst. Press,
Washington, D.C.
ANDREWS, F. W. 1950-1956. The Flowering Plants of the [Anglo-Egyptian] Sudan. Vols.
1-3. T. Buncle, Arbroath, Scotland.
AUnR�VILLE, A. 1955. La disjonction africaine dans la flore foresti�re tropicale. Compt.
Rend. Sommaire S�ances Soc. Biog�ogr. 278: 43-49.
& J. F. LEROY. 1961-1972. Flore du Gabon. Mus�um National d'Histoire Natu-
relle, Paris.
& . 1963-1975. Flore du Cameroun. Mus�um National d'Histoire Naturelle,
Paris.
AYENSU, E. S. 1973. Phytogeography and evolution of the Velloziaceae. Pp. 105-119, in
B. J. Meggers, E. S. Ayensu & W. D. Duckworth (editors), Tropical Forest Ecosystems
in Africa and South America: A Comparative Review. Smithsonian Inst. Press, Wash-
ington, D.C.
AYODELE COLE, N. H. 1974. Climate life forms and species distribution on the Loma mon-
tane grassland, Sierra Leone. Bot. J. Linn. Soc. 69: 197-210.
BOUQUET, A. 1976. Etat d'avancement des travaux sur la Flore du Congo-Brazzaville. Bois-
siera 24: 581.
BRENAN, J. P. M. 1954. Endemism in the flora of the tropical African rain-forest. Thesis,
unpublished.
.1959. Flora of Tropical East Africa: Leguminosae Subfam. Mimosoideae. Crown
Agents, London.
.1961. Triceratella, a new genus of Commelinaceae from Southern Rhodesia. Kirkia
1: 14-20.
.1965. The geographical relationships of the genera of Leguminosae in tropical
Africa. Webbia 19: 545-578.
1967. Flora of Tropical East Africa: Leguminosae Subfam. Caesalpinioideae.
Crown Agents, London.
CARCASSON, R. H. 1964. A preliminary survey of the zoogeography of African butterflies.
E. African Wildlife J. 2: 122-157.
CHAPIN, J. P. 1936. Birds of the Belgian Congo. Bull. Amer. Mus. Nat. Hist. 65: 1-756.
CHAPMAN, J. D. & F. WHITE. 1970. The Evergreen Forests of Malawi. Commonwealth
Forestry Institute, Univ. of Oxford, England.
475
1978]


ANNALS OF THE MISSOURI BOTANICAL GARDEN
CHATTERJ EE, 1). & J. P. M. BRENAN. 1950. Mansonia dipikae Purkayastha. Hooker's
Icon. Pl., ser. 3, 5: tab. 3484.
CLAYTON, W. D. 1976. The chorology of African mountain grasses. Kew Bull. 31: 273-
288.
- & F. N. HEPPEn. 1974. Computer-aided chorology of West African grasses. Kew
Bull. 29: 213-233.
CUFOIXNTIS, C. 1953-1972. Enumeratio Plantarum Aethiopiac Spermatophyta. Bull. Jard.
Bot. Bruxelles and Bull. Jard. Bot. Nat. Belgique. Reprint ed. 2 vols. Jardin Botanique
National de Belgique, Bruxelles.
EXELL, A. W. 1944. Catalogue of the Vascuilar Plants of S. Tom� (with Principe and
Annobon). British Museum (Natural Ilistory), London.
S1971. Flora Zambesiaca. Mitt. Bot. Staatssamunl. Miinchen 10: 69-70.
- & M. L. CON�ALVES. 1973. A statistical analysis of a sample of the flora of Angola.
Garcia de Orta, S�r. Bot. 1(1-2): 105-128.
, F. A. MEDON�A, A. FEINANDES & E. J. MENI)ES (editors). 1937-1970. Conspectus
Florae Angolensis. Vol. 1-4. Junta de Investiga�oes do Ultramar, Lisboa, Portugal.
-- , I. WVILD & OTHERS. 1960-1971. Flora Zambesiaca. Vol. 1-Crown Agents for
Overseas Covernments and Administrations, London.
FANSHAWE, D. B. 1969. The Vegetation of Zambia. Forest Research Bulletin No. 7. Kitwe,
Zambia.
FERNANDES, A. 1971. Rapport sur le Conspectiis Florae Angolensis. Mitt. Bot. Staatssamml.
Minchen 10: 72-74.
FLORE DU CONGO BELGE ET DU RUANDA-URIUNDI. 1948-. [Publications de ]'Institut Na-
tional pour l'�tude Agronomique du Congo Belge] & Jardin Botanique National de
Belgique, Bruxelles.
FRYXELL, P. A. 1968. A re-definition of the tribe Goss!pic'a. Bot. Caz. (Crawfordsville)
129: 296-308.
CILLET, H. 1968. Le peuplement v�g�tal du massif de l'Ennedi. M�m. Mus. Natl. Hist.
Nat., S�r. B, Bot. 17: 1-206.
GILLETT, J. B. 1958. Indigofera (Microcluaris) in Tropical Africa. Kew Bulletin, Add.
Ser. 1: 1-166.
Goo�, R. D'O. 1974. The Ceography of the Flowering Plants. Ed. 4. Longman, London.
GUILLAUMET, J.-L. 1967. Recherches sur la V�g�tation et la Flore de la R�gion du Bas-
Cavally (C�te-d'Ivoire). O.R.S.T.O.M., Paris.
CUINEA LOI'Ez, E. 1946. Ensayo Ceobotinico de la Cuinea Continental Espanola. Direc-
ci6n de Agricultura de los Territorios Espanoles del Colfo de Guinea, Madrid.
HAMILTON, A. 1976. The significance of patterns of distribution shown by forest plants
and animais in tropical Africa for the reconstruction of Upper Pleistocene palaeoenviron-
ments. In E. M. Van Zinderen Bakker (editor), Palaeoecology of Africa. Vol. 9: 63-97.
A. A. Balkema, Cape Town.
HEDBERG, I. & O. HEI)IEHG (editors). 1968. Conservation of Vegetation in Africa South of
the Sahara. Acta Phytogeogr. Suec. 54: 1-320. Uppsala.
IEDHERG, O. 1957. Afroalpine Vascular Plants. A Taxonomic Revision. Symb. Bot. Upsal.
15: 1-411.
1965. Afroalpine flora elements. Webbia 19: 519-529.
S1969. Evolution and speciation in a tropical high mountain flora. Biol. J. Linn.
Soc. 1: 135-148.
HEPPER, F. N. 1965. Preliminary account of the phytogeographical affinities of the Flora
of West Tropical Africa. Webbia 19: 593-617.
. 1976. Flora of West Tropical Africa. Boissiera 24: 563-564.
HUTrcIUNSON, J. & J. M. DALZIEL. 1954-1968. Flora of West Tropical Africa. Ed. 2.
Revised by R. W. J. Keay & F. N. Hepper. Vols. 1-3. Crown Agents, London.
JARDIN BOTANIQUE NATIONAL )E BELGIQUE. 1976. Reprint of "Enumeratio Plantarum
Aethiopiae Spermatophyta" by C. Cufondontis. Boissiera 24: 589.
KERAUDREN-AYMONIN, M. 1976. Progr�s accomplis dans la connaisance de la flore de
Madagascar et des Comores. Boissiera 24: 635-639.
KILLICK, D. J. B. 1976. Report on progress with Flora of Southern Africa. Boissiera 24:
633-634.
LANGENHEIM, J. H. 1973. Leguminous resin-producing trees in Africa and South America.
Pp. 89-104, in B. J. Meggers, E. S. Ayensu & W. D. Duckworth (editors), Tropical
476
[Voi.. 65


1978]
BRENAN-TROPICAL AFRICA
477
Forest Ecosystems in Africa and South America: A Comparative Review. Smithsonian
Inst. Press, Washington, D.C.
LEHRUN, J.-P. 1971. Quelques phan�rogames africaines � aire disjointe. Mitt. Bot. Staats-
samml. Minchen 10: 438-448.
.1977. El�ments pour un Atlas des Plantes Vasculaires de l'Afrique S�che. Institut
d'El�vage et de M�d�cine V�t�rinaire des Pays Tropicaux, Maisons Alfort, France.
L�ONARD, J. 1971. Progr�s accomplis dans l'�tude de la Flore du Congo, du Rwanda et
du Burundi de 1966 � 1970. Mitt. Bot. Staatssamml. Minchen 10: 46-47.
.1975. Statistiques des progr�s accomplis en 21 ans dans la connaissance de la
flore phan�rogamique afrique et malgache (1953-1973). Boissiera 24: 15-19.
LETOUZEY, R. 1976. Flore du Cameroun. Boissiera 24: 571-573.
LEWALLE, J. 1975. End�misme dans une haute vall�e du Burundi. Boissiera 24: 84-89.
LUCAs, G. LL. 1968. Kenya. Pp. 152-163, in I. & O. Hedberg (editors), Conservation of
Vegetation in Africa South of the Sahara. Acta Phytogeogr. Suec. 54: 1-320.
MILDBRAED, J. 1922. Wissenschaftliche Ergebnisse der Zweiten Deutschen Zentral-Afrika-
Expedition 1910-11 unter Fihrung Adolf Friedrichs, Herzogs zu Mecklenburg. Vol. 2.
Klinkhardt & Biermann, Leipzig.
MILNE-REDHEAD, E. 1971. Progress on the Flora of Tropical East Africa. Mitt. Bot. Staats-
samml. Minchen 10: 66-67.
MONOD, T. 1971. Remarques sur les sym�tries floristiques des zones s�ches nord et sud en
Afrique. Mitt. Bot. Staatssamml. Miinchen 10: 375-423.
MOORE, H. E. 1973. Palms in the tropical forest ecosystems of Africa and South America.
Pp. 63-88, in B. J. Meggers, E. S. Ayensu & W. D. Duckworth (editors), Tropical Forest
Ecosystems in Africa and South America: A Comparative Review. Smithsonian Inst.
Press, Washington, D.C.
MOREAU, R. E. 1952. Africa since the Mcsozoic: with particular reference to certain bio-
logical problems. Proc. Zool. Soc. London 121: 869-913.
MORTON, J. K. 1972. Phytogeography of the West African mountains. Pp. 221-236, in
D. H. Valentine (editor), Taxonomy, Phytogeography and Evolution. Academic Press,
London.
POLILL, R. M. 1968. Tanzania. Pp. 166-178, in I. Hedberg & O. Hedberg (editors),
Conservation of Vegetation in Africa South of the Sahara. Acta Phytogeogr. Suec. 54:
1-320.
.1976. Progress on the Flora of Tropical East Africa. Boissiera 24: 607-608.
RAVEN, P. H. & D. I. AXELROD. 1974. Angiosperm biogeography and past continental
movements. Ann. Missouri Bot. Gard. 61: 539-673.
RICHARDs, P. W. 1973. Africa, the "Odd Man Ont." Pp. 21-26, in B. J. Meggers, E. S.
Ayensu & W. D. Duckworth (editors), Tropical Forest Ecosystems in Africa and South
America: A Comparative Review. Smithsonian Inst. Press, Washington, D.C.
RoBYNs, A. 1965. On the state of the Congo Flora. Ann. Missouri Bot. Gard. 52: 427-431.
ROSEVEAR, D. R. 1953. Checklist and Atlas of Nigerian Mammals. Nigerian Government,
Lagos.
SCHNELL, R. 1968. Guin�e. Pp. 69-72, in I. Hedberg & O. Hedberg (editors), Conserva-
tion of Vegetation in Africa South of the Sahara. Acta Phytogeogr. Suec. 54: 1-320.
.1976. Introduction � la Phytog�ographie des Pays Tropicaux. Vol. 3-4. La Flore
et la V�g�tation de l'Afrique Tropicale 1-2. Gauthier-Villars, Paris.
SHAN-Z, H. L. & B. L. TURNER. 1958. Photographie Documentation of Vegetational
Changes in Africa over a Third of a Century. University of Arizona, College of Agri-
culture Report 169.
SILLANS, R . 1958. Les Savanes de l'Afrique Centrale. P. Lechevalier, Paris.
THORNE, R. F. 1973. Floristic relationships between tropical Africa and tropical America.
Pp. 27-46, in B. J. Meggers, E. S. Ayensu & W. D. Duckworth (editors), Tropical Forest
Ecosystems in Africa and South America: A Comparative Review. Smithsonian Inst.
Press, Washington, D.C.
WHITE, F. 1965. The savanna woodlands of the Zambezian and Sudanian Domains. Webbia
19: 651-681.
WICKENs, G. E. 1976. The flora of Jebel Marra (Sudan Republic) and its geographical
affinities. Kew Bull. Add. Ser. 5: 1-368.
WILD, H. 1964. The endemic species of the Chimanimani Mountains and their signifi-
cance. Kirkia 4: 125-157.
.1965. The flora of the Great Dyke of Southern Rhodesia with special reference to
the serpentine soils. Kirkia 5: 49-86.


478 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VOL. 65
-- & A. FERNANDES. 1967. Vegetation Map of the Flora Zambesiaca Area. Flora Zam-
besiaca, Supplement. Salisbury.
- & G. V. POPE. 1977. New and interesting Compositae from South-Central Africa,
3. Kirkia 10: 309-338.
WILLIS, J. C. 1973. A Dictionary of the Flowering Plants and Ferns. Ed. 8. Revised H. K.
Airy Shaw. Cambridge Univ. Press, Cambridge, England.
WINTER, B. DE. 1971. Floristic relationships between the northern and southern arid areas
in Africa. Mitt. Bot. Staatssamml. Miinchen 10: 424-437.