ACTA PHYTOGEOGRAPHICA SUECICA

BD ID IT SVENSKA VAXTGEOGRAFISKA SALLSKAPET

49

FEATURES OF

AFROALPINE PLANT ECOLOGY

Avec un resume en frant;ais

BY

OLOV HEDBERG

UPPSALA 1964

A L M Q V I S T & W I K S E L L S B 0 K T R Y C K �- R I A B SVENSKA VAXTGEOG,RAFISKA SALLSI(APET

SOCIETAS PHYTOGEOGRAPHICA SUECANA

Adress: Vaxtbiologiska Institutionen, Villavagen 14, Uppsala, Sverige

Styrelse (council): prof. Hugo Sjors, ordf. (chairman); lab. Magnus Fries, v. ordf. (vice chairman); fil.lic. Bengt M. P. Larsson, sekr.; fil. kand. Folke Bjorkback, skattm. (treasur�r); doe. Mats Wrern, red. (editor); fil. lic. Gunnar Wassen, v. sekr.; doe. Olov Hedberg; prof. John Axel Nannfeldt; doe. Nils Quennerstedt.

Sallskapet har till andama.I att vacka och underhalla The object of the Society is to promote investigation intresse for vaxtgeografien i vidstracktaste mening, in flora and vegetation, their history and their eco­ att framja utforskandet_av flora och vegetation i Sve­ logical background. Through publication of mono­ rige och andra lander och att havda geobotanikens graphs, and other activities, the Society tries to praktiska och vetenskapliga betydelse. stimulate geobotanical research and its application Sallskapet anordnar sammankomster och exkur­ to practical and scientific problems. Membership is sioner och utger publikationsserien Acta phytogeo­ open to all persons, in Sweden and e]sewhere, who graphica suecica. Varje for sallskapets. syften och have a personal interest in the advancement of verksamhet intresserad person kan vinna intrade phytogeography. efter anmalan hos sekreteraren. Foreningar, bibliotek, The Society issues the Acta Phytogeographica laroanstalter och andra institutioner kunna inga som Suecica. Individual members and subscribers (so­ abonnenter. Arsavgift 24 kr. (18:- :£or studerande); cieties, institutes, libraries, etc.) receive the Acta standigt medlemskap 400 kr. for annual dues of 24 Skr plus postage. There are Medlemmar och abonnenter erhalla Acta mot additional fees in years when more than one volume postforskott pa arsavgift jamte porto och expedi­ are issued. Permanent membership 400 Skr. For tionskostnad. Da flera band utges under samma ar, membership please apply to the Secretary. uttages en tillaggsavgift for extra band. Postgiro 7 504 7. Back volumes may be obtained by members and Medlemmar och abonnenter kunna erhalla aldre subscribers at reduced cost (see list of publications, publikationer till reducerat pris (parentespriset i prices in brackets). To receive the Acta by exchange listan). for other publications, please apply to the Institute Librarian (address as above).

SVENSKA VAXTSOCIOLOGISKA SALLSKAPETS HANDLINGAR

I. H. OsvALD, Die Vegetation des Hochmoores Komosse. 6. TB. C. E. FRIES, Die Rolle des Gesteinsgrundes bei der 1923. 40: - (27: -}. Verbreitung der Gebirgspflanzen in Skandinavien. 2. G. E. Du RIETZ, Gotlii.ndische Vegetationsstudien. 1925. 3: - (2: -). 1925. 12: - (8: -}. 7. H. OsvALD, Zur Vegetation der ozeanischen Hochmoore 3. G. E. Du RIETZ & J. A. NANNFELDT, Ryggmossen und in Norwegen. 1925. 15:- (10: -). Stigsbo Rodmosse, die letzten lebenden Hochmoore 8. G. E. Du RIETZ, Die regionale Gliederung der skandi­ der Gegend von Uppsa.la.. 1925. 6: - (4: -). na.vischen Vegetation. 1925. 15: - (10: -) . 4. G. E. Du RIETZ, Zur Kenntnis der flechtenreichen 9. G. SAMUELSSON, Untersuchungen iiber die hohere Zwergstra.uchheiden im kontinenta.len Siidnorwegen. Wasserflora von Dalarne. 1925. 6:- (4: -). 1925. 12: - (8: -). 10. TB. C. E. FRIEs, En vaxtsociologisk huvudfraga. (A 5. TB:. C. E. FRIEs, Okologische Un.d phii.nologische Beob­ principal problem of phytosociology.) 1926. 0: 50 (0: 25). a.chtungen bei Abisko in den Jahren 1917-1919. I. 1925. 15: - (10: -).

ACTA PHYTOGEOGRAPHICA SUECICA

1. E. Al.MQUIST, Upplands vegetation och flora. (Vegeta­ their distribution in. Northwestern Europe.) 1932. 12: - tion and flora of Uppland.) 1929. 40:- (27: -). (8: -). 2. S. THUNMARK, Der See Fiolen und seine Vegetation. 5. H. OsvALD, Vegetation of the Pacific coast bogs of 1931. 21:- (14: -). North America. 1933. 12:- (8: -). 3. G. E. Du RIETZ, Life-forms of terrestrial flowering 6. G. SAMUELSSON, Die Verbreitung der hoheren Wasser­ plants. I. 1931. 15:- (10: -). pflanzen in Nordeuropa. 1934. 21: - (14: -). 4. B. LINDQUIST, Om den vildvaxa.nde skogsalmens raser 7. G. DEGELIUS, Das ozea.nische Element der Strauch­ och deras utbredning i Nordvii.steuropa. (Summary: und Laubflechtenflora von Skandinavien. 1935. 36: ;_ The races of spontaneous Ulmus glabra. Huds. and (24: -). ACTA PHYTOGEOGRAPHICA SUECICA 49

FEATURES OF

AFROALPINE PLANT ECOLOGY

avec un resume en fran�aise

BY

OLOV HEDBERG

UPPSALA 1964

Almqvist & Wiksells Boktryckeri AB Illustrations printed with contribution from Langmanska Kulturfonden Contents

Frontispiece Lobelia keniensis 59

Preface 5 Carduus keniensis 61

Introduction 7 Conclusion on "Giant rosette plants" 62

Frost heaving resistance 64 I. AFROALPINE ENVIRONMENTS 8 Xeromorphy 70 Macro climates 8 Reduced length of internodes 73 Atmospheric pressure 8 AFROALPINE LIFE-FORMS Wind conditions and cloudiness 8 Ill. 80

Precipitation . 10 Introduction 80

Insolation and (outward) long wave Current life-form systems 80

radiation . 15 Giant rosette plants . 84

Temperature conditions 16 Tussock grasses (and sedges) 86

Relative humidity . 19 Acaulescent rosette plants 87

Number of humid months 19 Cushion plants 88

Conclusion . 20 Sclerophyllous shrubs (and dwarf-shrubs) 89

Snow and glaciers 20 Other phanerogamic life-forms 90

Microclimates. 22 Solifluction floaters 92

Soils. 26 IV. WATER ECONOMY IN THE AFROALPINE Soil water 28 FLORA 93

Solifluction 29 V. PRINCIPAL AFROALPINE PLANT COMMU- Fire 35 NITIES 96

Biotic influences 3 7 Introduction 96

Dendrosenecio woodlands 96 Il. ADAPTIVE TRENDS IN THE .AFROALPlNE FLORA 41 Helichrysum scrub I 04

Introduction 41 Alchemilla scrub 110

Protection against radiation 42 Tussock grassland 114

Temperature insulation 45 Carex bogs and related communities 118

Big grass tussocks 45 Other plant communities 128

Giant leaf rosettes . 49 Vicarious plant communities 129

Senecio keniodendron 49 RESUME. 131

Senecio brassica . 52 EXPLANATION TO TABLES 7--10 132 Other Dendrosenecios; comparison with REFERENCES. 133 Espeletia . 52

Lobelia telekii 56 INDEX TO LATIN NAMES 140

Acta Phytogeogr. Suec. 49

Upper photo: Two leaf rosettes of Lobelia keniensis in night position with densely folded leaves, covered by hoar frost (cf. p. 59 and Fig. 50). The big grass tussocks are formed by Festuca pilger·i. Photo 0. Hedberg, Aug. 1948. Lower photo: Leaf rosette of Lobelia lceniensis in day position with patent leaves, between which appears the surface of the "water reservoir" described in the text (p. 59). Photo 0. Hedb rg, Aug. 1948. (Printed in Svensk Naturvetenskap, 1963.) Preface

The author's first notions about afroalpine plant elaboration had to be done in spare time between ecology were received through a lecture by Pro­ other projects and has therefore required a number fessor Carl Troll given in Uppsala in the autumn of of years. In the course of this work much valuable 194 7. The impulses received on that occasion were assistance has been received from various sources, released through a minor mishap less than a year only part of which can be acknowledged here. later, when I spent a few weeks exploring the flora First of all I am indebted to Dr. Ake Holm, Upp­ on the upper parts of Mt Kenya. A bad knee forced sala, the initiator and leader of the Swedish East me to remain for a few days more or less stationary Mrica expedition in 1948, for the opportunity to around my camp in the upper part of Teleki Valley. join the expedition as well as for much help and These days were spent in making some ecological cheerful comradeship during the field work. My observations, and a few analyses of the local participation in this expedition was rendered vegetation. This vegetation, growing at an altitude economicaJly feasible by a grant from the Swedish of more than 4000 metres not far from the equator, Natural Science Research Council. The completion is exposed to a peculiar and rigorous climate with of the present book was achieved at the Institute small seasonal variations but large diurnal changes. of Systematic Botany of the University of Uppsala. Heavy frost occurs almost every night, whereas Most of the meteorological data included here were the insolation in daytime is very intense. I soon obtained from the Meteorological Institute, and discovered that the weird Giant Lobelias and Giant much of the ecological literature was consulted at Senecios growing around my camp behaved in a the Institute of Plant Ecology, both of the same very peculiar manner at night. On closer study these University. At various stages of my work I have plants proved to display some quite remarkable profited from discussions with numerous colleagues adaptations to the inhospitable afroalpine climate, and friends, notably G. E. Du Rietz, J. B. Gillett, with "summer every day and winter every night". J. Leonard, J. A. Nannfeldt, S. Nilsson, H. Rufelt, Also several other plants were found to possess R. Santesson, and H. Sjors. For multifarious help peculiar adaptations to the perils of the afro­ and encouragement, including fast and sagacious alpine environment. Similar conditions obtain, of proof-reading, I am indebted to my wife, Mrs. course, on the other high East Mrican mountains Inga Hedberg. Some of the photographs reproduced as well, and the following account embodies also here were taken by Dr. A. Holm. Most of the en­ some observations from the latter. Because of in­ larged prints utilized in making the illustrations sufficient time and equipment my studies of these were skilfully prepared by Miss U.-B. Sahlstrom, matters had to be of a very preliminary nature. and the English text was checked by Mrs . C. The present publication cannot therefore claim to Horner. The printing was supported by grants give an exhaustive account of "afroalpine plant from "Langmanska Kulturfonden" and from the ecology" but only to demonstrate some of the pro­ Swedish Natural Science Research Council. I am blems encountered here, and to draw attention to also indebted to the Board of "Svenska Vaxt­ an interesting and almost virgin field for ecological geografiska Sallskapet" for accepting my book in research. The first draft for this book was made their series "Acta Phytogeographica Suecica", during my stay on Mt Kenya in 1948. The further and to Dr. M. Wrern for editorial assistance. 0LOV HEDBERG

Acta Phytogeogr. Sueo. 49 Fig. I. Orographic map of the SE part of Afric a, showing the location of the relevant high East Afric an mountains.

Acta Phytogeogr. Suec. 49 Introduction

The science of plant ecology studies the interrela­ Elgon, Aberdare, Mt Kenya, Kilimanjaro, and Mt tions between plants and their environment. Ob­ Meru-see Fig. l. The vegetation of these moun­ viously any plant species must be adapted to its tains differs very much from that of the surrounding natural environment in innumerable ways-other­ plains, exhibiting a marked altitudinal zonation wise it would not persist. Most of its adaptations (HEDBERG, 1951). Their uppermost parts harbour are likely to escape attention, however, either a peculiar flora with pronouncedly disjunct dis­ because they are difficult to observe, or are shared tribution, characterized by the occurrence of Giant with many other species of the same region, or Senecios, Giant Lobelias, shrubby Alchemillas and because their significance for the plant is not under­ other plants of remarkable life-form, and con­ stood. Easiest of observation are such morphologi­ taining representatives of many genera unknown cally manifested adaptations that can be correlated from the lowlands of tropical Africa. According to a to definite facets of the environment. A well known term coined by HAUMAN (1933a) this is called the example is provided by the famous life-form system afroalpine flora. Outside Tropical East Africa it is of RAUNKIAER in which plants are classified ac­ well represented also on the high mountains of cording to their way of surviving the unfavourable Ethiopia, with a few species extending to the Ye­ season-notably after the relation .to ground level men, Cameroons, South Tropical Africa, and South of their perennating buds. Other cases where Africa (HEDBERG, 1961). morphologically observable adaptations can be Ever since its discovery less than a century ago easily demonstrated occur in the floras of extreme this afroalpine flora has caught the interest of habitats, such as deserts, sea shores, and alpine or botanists, by its peculiar Giant Senecios and Giant arctic areas. In general it seems that the more ex­ Lobelias as well as by its vicarious species and its treme the habitat, the more conspicuous adapta­ phytogeographical problems in general. Inhabiting tions can be found to it. Few environments popula­ an extreme environment it is very poor in species­ ted by plants on this planet could have greater in a recent taxonomic revision only some 280 taxa claims to extremeness than the alpine belt of the of vascular plants were recognized as regular mem­ high East African mountains. It is therefore only bers of the afroalpine flora in Tropical East Africa natural to find here a number of quite remarkable (HEDBERG, 1957). A preliminary phytogeographical adaptations to this afroalpine environment. treatment of this flora was attempted in HEDBERG, The major part of Tropical East Africa consists of 1961; a fuller account was deemed impracticable vast uplands at altitudes of 1000-2000 m, dissected until the life-form of some of these plants and their by two series of mainly longitudinal rift-valleys. adaptations to the afroalpine climate had been Along those valleys lie scattered a number of isola­ studied in more detail. An attempt at such an ted high mountains, mostly of volcanic origin, the ecological treatment is provided in this book. Pre­ highest of which reach altitudes between 3800 and liminary reports have been given in two earlier 6000 m, viz. the Virunga Volcanoes, Ruwenzori, publications (HEDBERG 1963, 1964).

Acta Phytogeogr. Suec. 49 I. Afroalpine enviroiiments

MACRO C LIMATE S

The character of the afroalpine climate is funda­ Although of considerable importance to many mentally governed by two geographical circum­ animals (cf. e.g. MoREAU, 1957, p. 328) this de­ stances: the vicinity to the equator, and the high crease in overall air pressure will presumably have altitude above sea level. All the mountains con­ little direct influence on plants. Of greater impor­ cerned in this study lie within the latitudes 2° N and tance is probably the accompanying decrease in the 4° S (see Fig. 1), and the lower limit of the alpine partial pressure of carbon dioxide; this may provide belt, as defined by HEDBERG (1951, p. 167), falls one of the reasons for the very slow growth rate of at about 3500-4100 m, while the upper limit of most afroalpine vascular plants (cf. DuTT & GuHA­ vascular plants lies around 5000 m (op. cit.). This THAKURTA, 1938). The rarefied air at high level is entails, firstly, that the climate is of a diurnal type, much more transparent for radiation than that at where seasonal variations in temperature are less sea level and has a smaller heat capacity-both important than the diurnal ones. Secondly, the circumstances of considerable ecological importance. high level provokes low air pressure and tempera­ ture, intense radiation, etc. (cf. TROLL, 1959, Wind conditions and cloudiness p. 19 f.). Since unfortunately no detailed long-range meteorological records exist from the alpine belt "The two main air currents which affect British in East Africa no authoritative account can at East Mrica are the northeast (November to Feb­ present be given, but I shall attempt to summarize ruary) and southeast (May to September) mon­ below as far as possible the scattered data avail­ soons, which are seasonal in character .... The line able. The mountains projecting into the alpine belt of demarcation between the two monsoons is the are the Virunga Volcanoes, Ruwenzori, Elgon, Inter-Tropical Convergence Zone. The average Aberdare, Mt Kenya, Kilimanjaro, and Mt Meru depth of either monsoon is about eight thousand (cf. Fig. 1). A general account of their main physical feet. Above this level, and hence over most of the features and vegetation zonation may be found in inland areas, winds are in general easterly" (ANo­ HEDBERG, 1951. NYMUS, 1948a, p. 2). The highest summits in East Africa are also influenced by the anti-trade, blow­ Atmospheric pressure ing above an altitude of some 4000-4500 m (cf. KLUTE, 1920, pp. 11, 42; TROLL & WrEN, 1950, That climatic feature which changes most regu­ p. 269). larly with altitude is of course the atmospheric pres­ This overall pattern is of course modified by the sure. At an altitude of 4400 m this is about 600 mb local topography, and each high mountain has its (or 60 % of the value at sea level) and at 5900 m own system of local winds (cf. TROLL, 1952b; about 500 mb (or 50 %). Both annual and diurnal ScAETTA, 1934, p. 116). The most important local variations are insignificant as compared with those winds on the isolated East Mrican mountains are at temperate latitudes (cf. RoBERT, 1946, p. 261; the "Hangwinde" (TROLL, op. cit., p. 126). In day­ WALTER, 1948, p. 1; HEDBERG, 1951, p. 160). time the mountain slopes are heated by the sun-

Acta Phytogeog1·. Suec. 49 Mac roclimates 9

TABLE l. Annual rainfall as registered by raingauges in nine different stations on Kilimanjaro.

Values compiled from East Alrican Meteorological Department, Summary of Rainfall, for th e years 1946-1958. Some of the values are incomplete because readings for one or two months are lacki ng; th ose fig ures are placed in brackets and have not been used for the calculation of annual mean values. Th e rainfall is given in inches; 1 inch = 25.4 mm . Gauges No. 1-6 (1E-6E according to the source quoted) are situated on the SE slope of th e mountain, and Nos. 7- 9 (1W- 3W) on th eW slope. A dash means th at no measurements are available.-Cf. text.

Gauge No. 1 2 3 4 5 6 7 8 9 Altitude, m 2200 2850 3800 4250 4850 5800 3050 4000 47 50

1945 65.4 70.7 36.6 (7.8) 1. 5 ( 0.1 ) 1946 ( 44.2) 73.6 21.6 (10.3) 4.6 0.7 1947 70.2 21.2 3.4 ( 2.2) ( 0.3) 63.6 49.9 4.6 1948 (71.5) 91.7 25.0 11.0 50.6 0.5 66.4 47 .4 10.4 1949 ( 68.9) (58.2) ( 18.8) ( 3.3) (22.0) ( 0.1 ) 48.9 38.5 2.3 1950 150.2 (75.2) 51 .3 35.8 10.9 63.0 52.0 1.5 1951 ( 105.3) 33.8 7.0 55.0 52.4 42.2 0. 5 1952 85.0 29.0 4.7 29.7 19.9 16.8 0.3 1953 68.3 ( 24.5) 5.1 28.4 13.4 10.6 1954 62.9 14.9 4.6 9.5 1955 46.9 68.1 21.1 1.6 7.5 1956 57 .9 57 .1 15.8 3.4 6.5 1957 68.2 61.6 14.2 5.7 33.2 1958 48.0 39.7 20.3 5.9 10.3

Mean valuea 72.5 (9) 66.6 ( 8) 25.4 (12) 8.0 ( 11) 20.6 (12) 0.6 (2) 46.8 (7 ) 36.8 (7 ) 3.3 (6) Ratio of max. to min. annual rainfall 3.1 2. 8 3.6 22.4 38.0 5.0 4.9 40.0

a The number of annual values employed is given in brackets after each mean.

shine and the air near the ground becomes warmer ascending clouds pass close to the ground forming and moves upwards, following the slope. At higher a damp mist (Fig. 7, p. 16). As emphasized by level the ascending winds are cooled and part of TROLL (1952b, p. 126) this cycle of cloud formation their moisture contents is condensed, causing cloud creates a pronounced difference in insolation be­ formation. At night, on the other hand, the moun­ tween the eastern and western side of the same tain slopes rapidly lose heat by outward radiation, mountain, and hence a different altitude for the and the air near the ground gets cooler and flows snowline in East and West. From an ecological

downhill as a more or less distinct "Fohn". At the point of view this diurnal cycle of cloudiness is of same time there will often arise a straturn of clouds course very important, since the screening of direct round the base of the mountain (cf. TROLL, la c. cit.; sunshine causes a rapid decrease in temperature and TROLL & WrEN, 1950, p. 271 and Fig. 7). On some an increase in relative humidity (�f. SALT, op. cit., p. of the relevant mountains this daily cycle of cloud 415). One ecological expression of it may be seen formation is fairly regular, so that from about noon in the flowering of the bushes of Heli chrysum citri­ most of their upper parts will as a rule be capped in spinum var. hoehnelii on Kilimanjaro, which tend clouds, e.g. on Kilimanjaro (KLUTE, 1920, p. 44 f.; to flower much more profusely on their west­ SALT, 1954, pp. 380, 400; author's observations exposed side (Fig. 10, p. 24; cf. HEDBERG, 1957, in June 1948), and Mt Kenya (TROLL & WrEN, Pl. I 0 B). On other mountains the changes in 1950, p. 271; JEANNEL, 1950, p. 107; and author's cloudiness are less regular, e.g. on Ruwenzori. observations in July and August 1948). Often the The upper parts of that mountain seem to be cap-

Acta Phytogeog1·. S'uec. 49 10 Afroalpine environments

observations on seven different mountains). In 6 special localities we may, however, occasionally get fairly strong local winds at night, caused by convergence of descending cold air ("Fohn"), as

s described by ScA:ETTA (1934, p. lOO) from the Mu- 9 kuru saddle on Karisimbi (cf. also GEIGER, 1950, p. 216). Similar observations were made by the 4 present author at Peter's Hut (3800 m above sea 4000 8 level) on Kilimanjaro. The rarity or absence of strong winds in most of the vegetated part of the alpine belt is testified to by the appearance of the Giant Senecios and Lobelias (cf. below). 3000 7 2 Precipitation

The rainfall records available for the afroalpine

2000 belt are scarce, and most of them give only annual totals. The best known mountain in this respect is 20 50 100 20 50 Kilimanjaro, where nine different gauges have been Fig. 2. Variation in annual rainfall (h orizontal axis, in in­ in operation on the upper parts for periods of 5-12 ches) with altitude (vertical axis, in m) on Kilimanjaro. years (cf. East African Meteorological Department, The diagram to th e left corresponds to th e SE slope of th e mountain, th e right-hand one refers to th e W slope. Th e Summary of Rainfall, for the years 1946-1958). small circles correspond to mean va lues for a number of Some values computed from this source are listed years (between 1945 and 1958), whereas th e +s igns to th e in Tables l and 2, and illustrated in Figs. 2-4. As left and right of th em at th e same level indicate th e smal­ is evident from Fig. 2 the amount of precipitation lest and largest annual rainfall regis tered durin g th e same decreases from a maximum somewhere in the mon­ period. Th e hatch ed area between the maximum and mini­ mum curves illus trates th e variability in th e annual rain­ tane forest belt (cf. KLUTE, 1920, p. 17; MoREAU, fall at different levels (cf. text). Th e values employed are 1936, p. 846; ANONYMUS, l948b, p. l) up to an taken from Table 1, and th e figures 1-9 refer to th e number altitude of some 4250 m on the SE slope of Kibo of th e raingauges in that table. and 4750 m on the W slope. At 4900 m on the E slope of Kibo it is again higher. ped with cloud and mist most of the day throughout As regards rainfall distribution during the year, the year (cf. HAUMAN, l933a, p. 708; HEDBERG, the surrounding parts of Kenya and Tanganyika 1951, p. 149). have two rainy seasons, the "long rains" from On mountains at higher latitude high winds are March to May, and the "short rains" from Mid­ a factor of considerable ecological importance, ac­ October to Mid-December (ANONYMUS, l948b). ting through desiccation, abrasion, breaking, etc. The same seasons can be traced in all the high (cf. DAUBENMIRE, 1947, p. 282 f.). In parts of level stations on Kilimanjaro (Fig. 3). During the larger mountain systems in the tropics high winds dry season precipitation seems to occur more often may also be important locally (TROLL, 1952 b, p. at high level than in the lowlands, however, though 130 f.). But as regards East Africa "high winds are in small quantities only (KLuTE, op. cit., p. 48). uncommon, ... except when associated with thun­ Judging from the mean values in Table l and derstorm conditions and, except in these, speeds Fig. 2 the precipitation would seem to be moderately of over twenty-five knots would be unusually low (considering the low temperature level) at sta­ high" (ANONYMUS, l948b, p. 2). The same thing tions 3 (3800 m) and 8 (4000 m), low at stations appears to be true of the alpine belt (cf. HAUMAN, 4 ( 4250 m) and 9 ( 4650 m), and higher again at l933a, pp. 709, 715; SALT, 1954, p. 400; author's station 5 (4850 m). But this is not the whole story.

Acta Phytogeogr. Suec. 49 Macroc limates 11

TABLE 2. Mean monthly rainfall (in inches) as registered by raingauges in nine different stations on Kilimanjaro.

The numbers of the gauges are the same as in Table 1, and the values have been obtained from the same source. Since the readings were often not made at the beginning of the month the values obtained are approximate. The occasional extreme values obtained at high level have been included in the means. For further data see explanation to Table l.

Gauge No. 2 3 4 5 6 7 8 9 Altitude, m 2200 2850 3800 4250 4850 5800 3050 4000 475 0 No. of years 9- 12 6-8 10-12 9-12 9- 12 2-5 6-8 6- 8 5-7

January 3.6 5.0 1 .7 0.4 2.3 0.03 5.4 3.8 0.3 February 4.6 5.4 1 .4 0.4 1.8 0.03 3.8 3.0 0.0 March 1 1.5 10.2 4.4 1.6 5.0 0.18 8.3 5.7 0.8 April 16.0 13.3 5.0 1.4 6.1 0.09 8.6 7.8 1.2 Ma.y 10.0 6.1 3.0 0.4 2.7 0.04 5.2 4.3 0.5 June 5.4 3.9 1.1 0.2 0.4 0.00 1.4 1.1 0.0 July 4. 1 3.0 1 .0 0.03 0.1 0.00 1 .0 0.7 0.7 August 4.0 2.3 0.6 0.04 0.1 0.00 1.0 0.6 0.0 September 2.7 3.8 0.4 0.2 0.2 0.00 0.8 0.5 0.0 October 5.3 5.2 1.3 0.4 0.5 0.05 2.7 1.9 0.1 November 4.9 4.0 1.6 0.4 0.8 0.00 3.4 3.0 0.3 a December 6.1 6.8 2.4 2.8 2.4 0.05 3.8 3.0 0.1

a If one extreme value is excluded this becomes 0.4.

TABLE 3. l\1ean monthly and annual rainfall (in inches) as registered by raingauges in eight different stations on Mt Kenya in 1950-1952.

Compiled from unpublished figures kindly supplied by the Director of the East African Meteorological Department, Nai- robi, in a letter of 3 March 1960. Gauges No. 1-4 were situated along the Naromoru route on the WSW slope of the moun- tain; Nos. 5-8 along the Sirimon route on the NNE slope. The number of readings available is given in brackets after each mean value.

Gauge No. 2 3 4 5 6 7 8 Altitude, m 2450 3050 3650 4250 2450 3050 3650 4250

January 1.3 ( 3) 2.5 ( 3) 1.7 ( 3) 1.7 ( 3) 0.5 ( 2) l.l (3) 1.1 ( 2) 1.0 (I) February 0.8 ( 2) 2.8 ( 2) 1.5 ( 2) 1.5 (2) 0.4 ( 2) 0.6 ( 2) 0.3 ( 2) 0.0 (I) March 2.5 ( 2) 3.3 (2) 2.9 ( 2) 3.3 (2) 1.5 (2) 2.0 ( 2) 1.0 ( 2) 2.0 ( 1) April 11 .0 (2) 9.5 (2) I0 .8 ( 2) 6.3 ( 2) 2.2 ( 2) 4.5 ( 2) 3.3 ( 2) 3.5 (I) May 5.0 ( 2) 8.3 (2) 5.8 ( 2) 3.0 (2) 5.8 ( 3) 7.5 ( 3) 7.7 ( 3) 4.5 (l) June 3.5 ( 2) 4.8 ( 2) 4.0 ( 2) 2.8 ( 2) 1.5 ( 3) 2.4 ( 3) 2.7 ( 3) 3.0 (I) July l.8 ( 2) 5.3 (2) 3.3 (2) 2.0 (2) 2.7 (3) 4.3 ( 3) 2.8 ( 3) 2.3 ( l ) August 2.8 ( 3) 3.7 ( 3) 3.5 ( 3) 2.5 (3) 2. 4 (3) 2.3 ( 3) 2.2 (3) 2.5 ( l ) September 2.3 (3) 4.8 (3) 3.7 ( 3) 2.0 (3) 2.5 ( 3) 4.3 ( 3) 3.9 ( 3) 3.5 ( 2) October 3.0 ( 3) 5.7 (3) 5.0 (3) 3.3 ( 3) 3.2 (3) 4.9 ( 3) 3.5 (3) 2.5 ( 2) November 7.7 ( 3) 9.0 ( 3) 6.8 ( 3) 5.7 ( 3) 3.9 ( 3) 4.6 ( 3) 4.2 ( 3) 3.5 ( 2) December 6.2 ( 3) 6.2 (3) 5.3 (3) 2.8 (3) 1.7 ( 3) 2.8 (3) 2.8 ( 3) 3.1 ( 2)

Year (mean) 51.8 ( 2) 67 .8 ( 2) 54.I ( 2) 37 .8 ( 2) 30.3 ( 2) 43.0 ( 2) 35.1 ( 2) 35.3 (1) Year (1951 ) 68.0 82.0 68.0 52.0 40.0 51.5 44.3 Year (I952) 35.3 53.5 40.2 23.5 20.6 34.5 26.0 35.3

Acta Phytogeogr. Suec. 49 12 Afroalpine environments

A B c D

5

5000 m

9

. - 4 a

4500 m

8 - 4000m lL...]

3500 m 2 7

3000 m

2500 m

J"F'M A MJ"JA SOND

Fig. 3. Histograms illustrating the variation in mean monthly rainfall (in inches) throughout the year for 16 stations at different levels on Kilimanjaro (A= SE slope, B =W slope) and Mt Kenya (C =WS W slope, D = NNE slope). The common scale at the left margin gives the altitude above sea level in metres; the base of each histogram is inserted at a level cor­ responding to the altitude of the station concerned on this scale. All values employed are taken from Tables 2 and 3.

As is evident from Table 1 and Figs. 2 and 4 the keeps at 2.8-3.1, but in the ericaceous belt it variation in the annual mean is quite astounding ascends to 3.6-4.9, and in the alpine belt it reaches at all high level stations. Already in the surrounding values of 22-40 (cf. Fig. 4). In other words, the lowland the variability of rainfall is remarkably precipitation in most of the alpine belt on this high (cf. KLUTE, 1920, p. 18; MoREAU, 1938, p. mountain must be quite erratic-one year may give 480 f.). For seven stations selected from various a total of 50 mm rain, another year may provide parts of Tropical East Africa the ratios betvv·een 1250 mm in the same locality. This feature must the highest and lowest annual rainfalls observed for undoubtedly be of vast ecological significance for each station during a 25 year period fall between the local flora. Values obtained from rain gauges 1.8 and 3.2 (ANONYMUS, l948b, p. 3). In the mon­ at altitudes above some 4000 m cannot be trusted tane forest belt on Kilimanjaro the same ratio to be quite adequate, of course, since above that

Acta PhytogeogT. Suec. 49 Macroclimates 13 level much of the precipitation falls as snow and m hail, which cannot be adequately sampled by this 5000 5 type of instrument (cf. VoLKENS, 1897, p. 321;

RICHARD, 1945, Figs. 1 a and 9). The annual mean 9 of 14 mm obtained for the crater of Kibo at an altitude of 5700 m will certainly bear little rela- 4000 tion to the actual precipitation in this locality, which must occur essentially as snow and rime (cf. HEDBERG, 1951, p. 155). The effects of the low and erratic rainfall are enhanced by edaphic factors, 3000 7 so that much of the alpine belt on Kilimanjaro is 2 of desert-like appearance (cf. VoLKENS, 1897, p. 322; HEDBERG, 1951, p. 193; SALT, 1954, p. 402). The lack of soil water may perhaps be partly com­ pensated for by the frequent occurrence of cloud 2000 and mist, however (cf. p. 9 above and Fig. 7). 10 20 40 On the other high mountains the rainfall seems to Fig. 4. Variability in annual rainfall at different levels on be less erratic. Kilimanjaro. The vertical axis corresponds to the al titude Also from Mt Kenya monthly precipitation above sea level in metres, and the horizontal axis gives the ratio between the highest and the lowest annual rainfall records are available for a few stations, though recorded during the period 1945-1958. Values from Table only for a period of less than 3 years (1950-1952). 1; the figures 1-9 refer to the numbers of the raingauges in Monthly and annual mean values for the 8 gauges that table. operated on that mountain are given in Table 3. The altitudinal distribution of rainfall on Mt Kenya in 1951-1952 is shown in Fig. 5 (left part); for trachtliche Niederschlagsmengen auch in der comparison curves are given also for the rainfall Gletscherregion fallen". The short range of ob- distribution on Kilimanjaro during the same two servations available from Mt Kenya make a years (right part of the same figure). Comparison of detailed comparison with other mountains rather these curves suggests that the decrease of rainfall hazardous, but it seems safe to conclude that the with altitude is less marked on Mt Kenya than on rainfall on the upper parts of Mt Kenya is much more Kilimanjaro. At an altitude of 4250 m (14000 ft) abundant and more evenly distributed throughout the precipitation seems to be about the same on the the year (and between different years) than at the NNE slope of Mt Kenya as towards WSW, although same levels on Kilimanjaro (cf. also Tab. 6, p. 19). the lower slopes towards NNE are considerably For the other high mountains precipitation re­ drier than the western ones. cords are lacking, or consist of a few short-term The mean monthly rainfall for the relevant gau­ observations only. Mt Meru appears to be equally ges is shown also in Fig. 3 (right hand part). All dry as Kilimanjaro on its upper parts, but no re­ stations display the same pronounced maximum cords exist. The rainfall conditions on Aberdare in April-May as shown by the Kilimanjaro stations, seem to be largely the same as on Mt Kenya. No and there is also a (less distinct) maximum about record is known from the alpine belt, but a 12- November. But the dry seasons seem to be much less year mean from a gauge at 2450 m in the montane pronounced at high level on Mt Kenya than lower forest belt gives 70 in. (1780 mm), and a 7-year down; only on the NNE slope is there a short dry mean from a station in the moorland at 3050 m season. From their glaciological studies TROLL & gives 67.6 in. (1720 mm; values computed from WIEN (1950, p. 268) also concluded that "in der readings kindly supplied by the Director, East Hochregion des Mount Kenya langere regenlose African Meteorological Department, Nairobi). These Zeiten nicht vorhanden sind und dass sehr be- values are slightly higher than at corresponding

Acta Phytogeog1·. Suec. 49 14 Afroalpine environments

m A B c D sooo

4000 I

Fig. 5. Variation in annual rainf all (horizontal ax is, in inch es) with 3000 altitude above sea level (vertical b a axis, in m) on= NNEMt Kenya (A= I ) WSW slope, B =E slope) and b a b a Kilimanjaro (C S slope, D = W slope) for the years 1951 (a) and 2000 b a 1952 (b). Based on values from 20 so 20 40 20 50 100 20 so Tables 1 and 3.

altitudes on Mt Kenya. From the alpine belt on cerning the seasonal distribution of rainfall nothing Elgon no rainfall records exist. seems to be known from the alpine belt on these As regards Ruwenzori it was stated by DAwE mountains. To judge from the appearance of their (1906, p. 539): "On the upper region of the moun­ alpine vegetation the wet weather would perhaps tain rain falls almost daily in the form of heavy be a little less persistent than on Ruwenzori. mists, and the vegetation throughout the day is Below a level of about 4000 m on the moun­ usually dripping with moisture". Most later visitors tains the precipitation falls largely as rain, oc­ seem to have had similar experiences, although casionally as snow or hail. At higher levels the SYNGE (1937) met with a dry spell. The only frequency of snow increases, and above some 4500 precipitation records available from high level are m most of the precipitation seems to come as a those of the Belgian expedition in 1932 (cf. HAu­ rule in the form of snow. There is no sharp alti­

MAN, 1933a, p. 707), reporting from the West slope tudinal limit in this respect, of course. During a a total of 101 mm (4 in.) in 24 days (13/7-5/8 1932) march from Kitandara over Freshfield Pass to at an altitude of 4200 m, and 78 mm (3.1 in.) in the Mobuku Valley on Ruwenzori the author started same period at 4350 m. The annual precipitation in cold rain at an altitude of 4000 m, went into falling on the highest parts has been estimated at wet snow at about 4100-4200 m, and passed a good some 1600 mm (63 in.; cf. ScAETTA, 1934, p. 245). blizzard in the pass at 4400 m (cf. Figs. 6 and 26). There are no reliable dry seasons, but there seem During three weeks spent at 4200 m in Teleki to be on the average better chances for "good Valley on Mt Kenya (July-August 1948) most of weather" in January (-February) and (May-) June the precipitation occurred as hail, a smaller part (cf. SYNGE, 1937, p. 200; HEINZELIN & MoLLARET, as rain. At the level of Lewis Glacier ( 4500-5000 1956, p. 4). m) on the same mountain practically all precipita­ On the Virunga Volcanoes a mean annual pre­ tion comes in the form of snow (TROLL & WIEN, cipitation of about 950 mm (38 in.) has been re­ 1950, p. 262 etc.). corded at the summit of Karisimbi, about 4500 m Yet another type of precipitation occurring at above sea level (cf. SCAETTA, 1934, p. 260; RoBYNS, high level is provided by damp mist (cf. Figs. 7 and 1948 a, p. 9; HEDBERG, 1951, p. 143). At the lower 101). As emphasized above, low clouds and mist limit of the alpine belt on the same mountain are of common occurrence in the alpine belt on (about 3800 m) it has been estimated at about most of the mountains. Often the mist is damp 1250 mm (50 in.; ScAETTA, op. cit., p. 301). Con- enough to deposit droplets of water on the ground

Acta Phytogeog1·. Suec. 49 Macroclimates 15

�.a..,;.. \'-". -.'(. \ "�.

'. {

Fig. 6. vVet snow, about 1 dm deep, among Giant Senecios and Heliclwysum stuhlmannii. Ugan da, Ruwe nzori, l:: etwee n Kitandara and Freshfield Pass, c. 4300 m. Photo A. Holm 10.4.1948. Cf. also Fig. 26. and the vegetation, at the same time as the eva­ of insolation from a fe>v stations in eastern Belgian poration is decreased (cf. KLUTE, 1920, p. 48; Congo. Further notes are provided, e.g., in the TROLL & WIEN, 1950, p. 208; SALT, 1954, p. 380). Bulletin Climatologique Annuel du Congo Belge et du In other parts of the world such mist has proved Ruanda- Urundi. For the alpine belt in East Africa to be a very important source of moisture for the no records are available, however. Because of the vegetation (MOREAU, 1938, p. 485; TROLL, 1956, rarefied and clean atmosphere at high level less of p. 780; ELLENBERG, 1959, p. 56 f.). At the low the incoming radiation is absorbed above the temperatures prevailing in the alpine belt the ground, so that under a clear sky the intensity of amounts of moisture received in this way will sunshine is greater than in the lowlands (HANN­ probably be small, but the ecological significance StiRING, 1939, p. 53; DAUBENMIRE, 1947, p. 225; of that mist may still be considerable, especially GEIGER, 1950, p. 10). Under the same conditions on dry mountains like Kilimanjaro (cf. SALT, the relative amount of ultraviolet light is consider­ loc. cit.). ably higher at high altitudes than at sea level (HANN-SURING, op. cit., p. 61; DAUBENMIRE, op. Insolation and (outward) long wave cit., p. 224). Because of the frequent occurrence of radiation cloud and mist on the upper parts of the mountains As regards solar radiation, ScAETTA (1934, p. 13 f.) (cf. p. 9 above) the intensity of insolation will as has given an account of the amount and distribution a rule be much reduced during at least part of the

.Acta Phytogeogr. Suec. 49 16 Afroalpine environments

Fig. 7. Mist. Tanganyika, Kili­ manjaro, S part of the Saddle be­ tween Kibo and Mawenzi, c. 4400 m. The light-coloured cushion in the foreground is formed by He­ licht-ysum newii f. brevicaulis. Photo A. Holm, June 1948.

day. Although part of the incoming radiation pene­ (op. cit., p. 17-19; HUMPHREYS, 1940, p. 99 etc.). trates the clouds as diffuse light, infrared rays, With increasing altitude above sea level the in­ which account for almost half of the insolation, tensity of reradiation decreases, and hence the ef­ seem to be almost completely absorbed (cf. DAUBEN­ fective long wave radiation from the ground in­ MIRE, 1947, p. 223; HUMPHREYS, 1940, p. 99, etc.). creases (GEIGER, op . cit. , p. 22). Thus on every Therefore the radiation climate of the alpine belt clear night in the alpine belt of the East African js very variable, displaying on most mountains a mountains there is a strong long wave radiation, fairly regular diurnal cycle according to the resulting most often in night frost (cf., e.g., KLUTE, -cloudiness. When the sky is clear about noon the 1920, p. 54; TROLL, 1943b, p. 161). This factor is intensity of the sunshine is quite remarkable, and also very important for the microclimatic differen­ may probably prove deleterious to unprotected tiation within the alpine belt (cf. p. 22 below). plant tissues as well as to human skin. Part of the incoming radiation reaching the Temperature conditions ground is reflected, but much of it is absorbed by the ground and the vegetation, being mainly con­ Lack of sufficient temperature readings preclude verted into heat. Apart from the reflected sunlight the computing of any annual mean temperatures the ground and the plant cover also emit radiation from the alpine belt in East Africa. In view of the themselves, mainly in infrared (GEIGER, 1950, small seasonal variations a mean value for a shorter p. 10 f.). In daytime this long wave radiation is of time may deviate very little from the annual mean, little importance, being much smaller in amount however. At four montane stations in Belgian Congo than the incident short wave radiation, but at at about the same latitude and at altitudes between night the former is by far the more important and 1500 and 2115 m the annual amplitude of mean may lead to a pronounced cooling of the soil sur­ temperatures for ten-day periods was found to face. This long wave radiation from the ground is range between 1.3 and 2.2°C. (ScAETTA, 1934, p. counteracted by reradiation from the atmosphere, 83). Towards higher levels on the mountains the the intensity of which depends upon its temperature seasons of precipitation seem to become less well­ .and its contents of water vapour and carbon diox­ marked (cf. p. 13 above), and hence the annual jde, and especially upon the occurrence of clouds variations in temperature are probably even smal-

Acta Phytogeogt·. Suec. 49 Macr oclimates 17

TABLE 4. Short-time mean temperatures and estimated annual mean temperatures from the afroalpine belt.

An asterisk (*) means that the readings were made at lower level than is usual (about 2-5 dm above the ground) and not in a standard screen. Temperatures given in degrees centigrade.

Methods of computing

Period Mean of Mean nf Mean of Estimated max min Altitude of 24-hour t + t t? + t14 + 2t21 in other Mountain m obseevat.ion means 2 4 ways Souece

Muhavura 4127 June 1929 + 4.2° SCAETTA, 1934, p. 118 Karisimbi 4000 Febr. 1930 + 4.40 Ibid., p. 119 Ruwenzori 4200 12/7-4/8 1932 + 1.90 Ibid., p. 129 Elgon* 35SO 13/5-19/5 1948 + 7.3° + 8.7" + 7.1° Mt Kenya, l* 4200 2/8-9/8 1948 + 3.6° + 4.6° + 3.2° HEDBERG, ined. Mt Kenya, 2* 4200 10/8-17/8 1948 + 3.1° 4.7° + 3.9° } + Kilimanjaro 4160 19/8-12/10 1912 + 3.3° + 1.80 KLUTE , 1920, p. 36-3 7

TABLE 5. Mean diurnal temperature amplitude in the afroalpine belt, in degrees centigrade.

Because of differences in aspect and microclimate these figures are obviously by no means comparable. The value for Muhavura relates to the summit, those from Karisimbi to a saddle a few hundred metres below the summit.

Period Mean diurnal Altitude of No. of temperature Mountain m observation days amplitude Source

Muhavura 4100 9/6 1929 1 c. 14.5° ScAETTA, 1934, p. 117 Karisimbi : Mukuru 4100 17/2-21/2 1930 4 c. 13.6° Ibid., p. 117 Ruwenzori 4200 12/7-4/8 1932 22 5.0° ]bid:, p. 129 Elgon 3580 13/5-19/5 1948 6 10.3° A. HoLM, ined. Mt Kenya, 4200 2/8-9 /8 194� 6 8.2° HEDBERG, ined. Mt Kenya, 2 4200 10/8-17/8 I 948 6 16.4° Ibid. Kilimanjaro 4160 19/8-12/10 1912 46 10.1° KLUTE, 1920, p. 36-37

ler. Even a short-time mean may therefore give a limit of the afroalpine belt should be about +5°- fair approximation to the annual mean. In Table 4 7°C. The annual isotherm for 0°C. will probably I have assembled all short-time mean temperatures be situated somewhere between 4500 and 5000 m, that could be traced from the alpine belt. Since and at the summit of Kibo (Kilimanjaro) at about different methods have been used by different 5900 m the annual mean must evidently be much authors in computing the means I have also noted below the freezing point. which method was employed, and for my own Much more important from a biological point of measurements all the three relevant methods have view than these mean values is of course the mode been employed parallelly in order to get compara­ of temperature variation. This could best be il­ tive figures. Since my ovvn thermograms were not lustrated by a "Thermoisoplethen-Diagramm" as registered in a meteorological screen at standard employed by TROLL (l943 a, 1955, etc.). In the level but much closer to the ground they are not absence of sufficient meteorological data no such .strictly comparable. Judging from the values of diagram can be constructed for an afroalpine cli­ Table 4, the annual mean temperature at the lower mate, but its general appearance would probably

2 - 641464 H edberg Acta Phytogeogr. S7tec. 49 18 Afroalpine environments

· '-+-t; . I -1- :

'- V

10.8. 11 .8. 12.8. 13.8 . 15 .8. 16.8. 1 2 < < a lo)J/ • • •· • 10 < • e 10 10 • • • • 8 10 • • a a • 4 6 8 9 • a • • � 8 9 � s e 10 • s • 4 • 8 • • s ; . M'• XII • 4 " !! !O M'> 4 r.. s )// • e ;o N'& XH '" M' •G_!Ii_ ,; '' M' JOX!i M'9 e ,a!rli 10 M'• �"� r!J4-;Jl t; ./- ;":ct:::; t:::.;· 0_I , -+ 1-- 1 :Ff.-/C::.i::Cft t r!.! t.,.J-nt:-:Icctc:-::L-J:J 'itf'-t=- ·t If ·i -t f:f=i i -1 1 !�f; ;i · i:·f h ! �-,-j ::f t ! c �LTT:� t- �-�r-h:+,Tfffi-t�n_:.:::-t , -· r=t"t:.c:-f=i:L-b�r�'l'!! -· ; .. i- I_, J r--t·· :j. .. ,-- 1 ;:_;T:,-Tj-� f. £-/�.- :i- i- E: '- H· t�l-i ,_,__ r-- ,- -t-·-r· · ' "M'";. ·.; t:::: :- r ..:C·t::· ·-::-J::f:.=_ =f: 1-- = '"1-

,__ t­ :-_:-- t-

-- . \··+-\-\---�- -\··· c L\c-:ti·'_ .;::::l;:t: \ \.. ::\:.\. ;. - 1 Fig. 8. Thermograms from the head of Teleki Valley, Mt Kenya, at about 4200 m altitude (cf. Fig. 103 for view of the locality). The upper thermogram was recorded on the lower part of the valley slope, with the thermograph placed about 5 dm above the ground between a few big boulders. The lower one was obtained on the flat valley bottom, the instrument being placed I dm above the soil surface, in the shadow of Senecio brassica leaves. The horizontal distance between these two stations was only about 50 m, the difference in altitude was certainly less than 5 m (cf. text). resemble something intermediate between those variation will be considered in a following section of Quito and El Misti (cf. TROLL, 1943a, Figs. 14 (p. 23). and 15). Annual variations are almost negligible, One remarkable feature of the afroalpine climate whereas the diurnal ones are very marked. In other is its rapid temperature changes, due largely to the words: in the afroalpine belt there is "summer thin atmosphere with low heat capacity, the in­ every day and winter every night" (HEDBERG, tense insolation and outward radiation, and the 1952a, p. 199; 1957, p. 10; cf. also SALT, 1954, p. changeable cloudiness (cf. above). On a clear 415). In at least the major part of this belt frost morning the temperature increase after sunrise is occurs on many if not most nights of the year, very rapid (cf. Fig. 8), and in daytime violent whereas day temperatures may be quite tolerable. temperature changes (up to 10°C in less than half The thermograph curves reproduced in Fig. 8 may an hour) are often brought about by changes in serve as examples. As regards the diurnal tempera­ wind and cloudiness (cf. VoLKENS, 1897, p. 321; ture amplitude there are large differences between KLUTE, 1920, p. 53; ScAETTA, 1934, p. 117-118; different mountains (cf. Table 5), evidently cor­ SALT, 1954, p. 412). These changes are made even related to the differences in topography, wind more drastic by the accompanying changes in conditions, and cloudiness. As is evident from Fig. 8 radiation. Radiation temperatures are very difficult the size of this diurnal amplitude may sometimes to measure, however. Another important aspect of differ markedly within short distance-this local the afroalpine temperature climate is the high

Acta Phytogeogr. Suec. 49 JJ1acro climates 19

TABLE 6. Number of humid months per year at different stations on the upper parts of Kilimanjaro and Mt Kenya� as determined by de Martonne's aridity index (cf. Lauer, 1952, p. 23).

The primary values were derived from East African Meteorological Department, Summary of Rainfall, for the years 1946- 1958, and from unpublished figures kindly supplied by the Director of the East African Meteorological Department, Nairobi, in a letter of 3 March 1960. The numbers of the raingauges are the same as in Tables 1-3.

Kilimanjaro Mt Kenya

Gauge No. I 2 3 4 5 7 8 9 2 3 4 5 6 7 8

Altitude, m 2200 2850 3800 4250 4850 3050 4000 4750 2450 3050 3650 4250 2450 3050 3650 4250

o. of years with compl. records 6 5 8 9 10 6 6 6 2 2 2 2 2 2 2

Variation in No. of humid months/year 8-12 8-12 5-10 0-4 2-8 6-l l 6-l l 0-3 7-10 9-12 10-12 10-12 5-9 9-l l 8-l l ll

Mean No. of humid months/year 10.0 10.6 7.5 2.0 4.9 7.7 8.0 1.2 8.5 10.5 ll ll 7 10 9.5 ll incidence of days with changes from frost to thaw During the long dry season the moisture seemed and the reverse-it is a "Frostwechsel-Klima" to be lower in the alpine belt as well as at lower level in the sense of TROLL (1943b). These "frost changes" (loc. cit. ). On Mt Kenya MACKINDER (1900, p. 483) are more common at ground level than in the free found the air to be usually dry, "the relative humi­ air (op . cit. , p. 161; cf. also p. 23 below). They are dity falling on more than one occasion to as low as essential for the development of solifluction (cf. 54 and 52 per cent". p. 29 f. below). A hygrometer diagram for about 2 days from Mukuru Saddle at 4000 m on the slope of Karisimbi Relative humidity shows for each day a minimum in relative moisture As regards the variation in relative humidity of the of some 50-55 % at noon, and then another mini­ air in the afroalpine belt the information available mum at night, caused by the dry "Fohn" (ScAETTA, is very fragmentary. The only continuous readings 1934, p. 219-220). A similar nightly minimum for any length of time were made by KLUTE (1920, could certainly be found on the upper slopes of pp. 40, 48 etc.) on Kilimanjaro, where the relative most of the other mountains as well. As regards moisture was measured three times a day (7 a.m., Ruwenzori, which is more of a mountain range than 2 p.m., 9 p.m.) for lf months. On the upper part of the other high mountains concerned (cf. HEDBERG, this mountain the relative humidity was found to 1951, p. 146), and which appears to receive more vary rather regularly according to the temperature, precipitation than any of the others, the relative as long as the daily cycle of cloudiness develops moisture must be expected to be much higher than "normally". When the clouds don't reach the saddle on Kilimanjaro (cf. HAUMAN, 1933a, pp. 707, 710) . plateau at noon the relative humidity reaches a minimum as the temperature attains its maximum Number of humid months about noon. In the localities studied (about 4160 m) the relative moisture was found to vary most A method of evaluating the degree of aridity or often between 60 % and 90 %, reaching the satura­ humidity of a climate that has been much used in tion point in damp mist. Sudden changes in relative recent geographical literature is to compute for humidity parallel the changes in cloudiness, but each locality studied the number of arid, resp. the changes are said to be rarely larger than 30 %. humid months of the year (LAUER, 1952; TROLL,

Acta Phytogeogr. Suec. 49 20 Afroalpine environments

1959, p. 42 etc.). A humid month is one for which Conclusion DE MARTONNE's aridity index is less than 20 (cf. LAUER, op. cit., p. 20-23). The only East African From the data presented above it should be high level stations for which such calculations can be evident that, in spite of their fundamental resem­ made are those mentioned above on Kilimanjaro blance, the alpine climates of the different high and Mt Kenya; the figures obtained from those are East African mountains display substantial dif­ given in Table 6. Of the stations listed, No. 3 ferences. Such differences occur, for instance, in from Kilimanjaro is in the ericaceous belt, No. 8 the amount and seasonal distribution of cloudiness at the lower limit of the alpine belt, and Nos. 4, 5, and precipitation, and in the size of the mean diur­ and 9 in the alpine belt. Nos. 3 and 7 of the Mt nal temperature amplitude (cf. ScAETTA, 1934, p. Kenya stations are at the lower limit of the alpine 116). On each mountain there is of course a gradual belt, and Nos. 4 and 8 in its middle part (cf. also climatic change from the lower limit of the alpine Fig. 3). The values from Kilimanjaro demonstrate belt towards the summit. There may also be con­ a remarkable variability both between different siderable differences in precipitation, insolation, localities and between different years, with quite diurnal temperature amplitude, etc., between dif­ desert-like conditions obtaining at stations 4 and ferent localities at the same altitude. One must 9. That large differences in hygric conditions exist therefore take care not to deduce too much about within a short distance in East Africa has been the climate of one mountain from observations pointed out before by LAUER (op . cit. , p. 39) for made on another, nor to make sweeping generali­ the Kihuiro District in NE Tanganyika, but the zations from the meagre data available. contrasts obtaining on Kilimanjaro seem to be even In order to find a climate of the same type as more striking. On Mt Kenya the number of humid those described above vv e have to look for other months is evidently larger and less variable. high mountains near the Equator (cf. TROLL, 1959, The number of humid months per year in the al­ p. 19 f.). The nearest approach seems to be provided pine belt on the other East African mountains can by part of the Andes of South and Central America at present only be guessed at. Ruwenzori may (cf. e.g. WEBERBAUER, 1905, p. 60 f.; HEILBORN, be expected to have 12 humid months a year in the 1925, p. 153 f.; EsPINosA, 1932, p. 124-134; alpine belt as well as lower down, and the Virunga DIELS, 1937, p. 26 f.; WEBER, 1958, p. 133 f.; and

Volcanoes will probably not fall much short of especially TROLL , 1959, and earlier papers). The this figure, whereas Elgon will probably come in afroalpine temperature climate is also paralleled somewhere between Mt Kenya and Kilimanjaro. to a large extent by the strongly oceanic cold Mt Meru seems to be equally dry in its alpine belt climate of the subantarctic islands (Kerguelen, as Kilimanjaro, and Aberdare seems to match Mt etc.; cf. TROLL, 1943b, p. 170; 1948, p. 49). Kenya.

SNOW AND GLA CIERS

As stated above, most of the precipitation falling unhindered by trees, the winterly snow-cover is on the uppermost parts of the high East Mrican here very uneven. When falling in calm weather mountains (above some 4000-4500 m) occurs as the snow may for some time form an even cover, snow. At higher latitudes the unequal distribution but as a rule it is soon swept away from elevations of the winterly snow-cover in the mountains fur­ and windward slopes, and accumulates in hollows nishes one of the most important ecological factors and on leeward slopes. Thus hilltops and ridges are of the alpine belt, which largely governs the practically free from snow, whereas in hollows the distribution pattern of the vegetation. "As a result snow drift may be several metres thick. Although of snow drifting, which in the alpine belt proceeds the total amount of snow varies considerably from

.Acta PhytogeogT. Suec. 49 Snow and glaciers 21

Fig. 9. Dense vegetation of Senecio adnivalis coll., Helich1·ysum stuhlmannii, Poa ruwenzoriensis, etc., and mosses, close to the snout of a glacier . Uganda, Ruwenzori, S slope of Mt Speke, c. 4400 m. Photo 0. Hedberg 7.4. 1948.

year to year, its distribution pattern is remarkably "nival" zone above those levels vegetation of constant. This uneven snow-cover leads to a num­ vascular plants is absent or very sparse. Such ber of edaphic and micro-climatic divergences temporary snow fields as occur here will probably within a limited area" (HEDBERG, l952 b, p. 51; have little ecological importance, since they are for further details and references see HEDBERG, fairly evenly distributed over large areas and loc. cit.; GJAEREVOLL, 1956, p. 16; DAHL, 1956, mostly of comparatively short duration (cf. MEYER, p. 251 f.; DAUBENMrRE, 1947, p. 101; POLUNrN, 1900, p. 348). 1948, p. 27 etc.; and CosTrN et al., 1959, p. 25 etc.). Glaciers at present exist only on the three highest On the high East African mountains the situa­ mountains of East Africa, viz. Ruwenzori, Mt tion is entirely different. In the absence of strong Kenya, and Kilimanjaro. They descend on Ru­ winds the snow falling at high level is not redeposi­ wenzori to about 4200--4350 m on the West slope ted to any large extent but forms rather an even (MICHOT, 1937, p. 225) and 4550 m towards the cover over the ground (cf. TROLL & WrEN, 1950, East (HEDBERG, 1951, p. 147), on Mt Kenya to p. 270). And because no thermal seasons exist, such about 4500 m in the West and 4700 m in the East

a snow-cover will as a rule melt away within a day (HEDBERG, op . cit., p. 152), and on Kilimanjaro or two in all localities situated below the snow (Kibo) to 4500 m in the Southwest and 5700 m line (cf. Fig. 6). Hence regular seasonal snowdrifts in the Northeast (NILssoN, 1932, p. 28-39; HED­ are absent from the "subnival zone" (MEYER, BERG, op . cit., p. 155). The differences on each 1900, p. 345; TROLL, 1948, p. 48; 1955, p. 720; mountain are probably due to differences in in­ 1959, pp. 28, 40). The altitude of the snow-line solation and precipitation between W- (SW-) and or "firn line" is reported to be about 4750 m on E- (NE-) exposed slopes (cf. pp. 9 and 12 above). Ruwenzori (HErNZELrN, 1953, p. 4), 4730 m on Mt At present all East African glaciers seem to be Kenya (TROLL & WrEN, 1950, p. 265), and 5400 m rapidly diminishing (cf. MEYER, 1900, p. 351 f.; on Kilimanjaro (NrLSSON, 1940, p. 63). In the GrLLMAN, 1923; NrLSSON, 1932, p. 32; SPrNK,

Acta Phytogeogr. Suec. 49 22 Afroalpine environments

1945; BERE, in BERE & HICKS, 1946, p. 84; TROLL & Both on Ruwenzori and on Mt Kenya a fairly WIEN, 1950; JEANNEL, 1950, pp. 26, 84; HEINZELIN, dense vegetation may occur quite close to the side 1953, p. 8). During the pleistocene pluvials, on the of the lower part of a glacier (cf. Fig. 9). Imme­ other hand, they evidently reached much lower down diately below the snout of a glacier the vegetation than at present-to about 2000 m on Ruwenzori, is as a rule very sparse, however. The recent melting 3000 m on Mt Kenya, and 3600-3800 m on Kili­ off of most glaciers has evidently occurred quite manjaro (cf. summaries in HEDBERG, 1951, pp. fast, whereas the colonization of the ground laid 147, 153, and 155). Furthermore, substantial pluvial bare proceeds relatively slowly (cf. HEINZELIN, glaciers existed on Elgon, as well as in Ethiopia 1953). But the proximity to the glacier would not (NILSSON, ·1932, p. 7 f.; 1940, p. 56 f.). seem to bring any marked ecological effect on the vegetation.

MICRO CLIMATES

Superimposed upon the general climatic pattern of The highest temperature observed here by him was the afroalpine belt are a number of microclimatic + 46°0. Considering the crude methods employed variations. Whereas the general character of the in these measurements (loc. cit.) and the difficulties afroalpine vegetation is governed by the macro­ of obtaining the actual surface temperature climate, the detailed vegetation pattern seems to (GEIGER, op. cit., p. 126) these values are probably be determined largely by local microclimatic dif­ too low. Surface temperatures of this magnitude ferences. The climate experienced by a living plant may certainly prove detrimental to many plants, is not as a rule the same as that recorded in an notably seedlings (op. cit., p. 180). At an altitude ordinary meteorological screen. Because of the of 4500 m on the summit of Karisimbi ScAETTA rarefied air and intense radiation the local micro­ (1934, p. 320) recorded in sunny weather at 2.15- climatical differences obtaining at high level are 2.30 p.m. temperatures between + 15° and + 19°0 more extreme than in most low-level biotopes (cf. at the soil surface and in mats of living plants, at the GEIGER, 1950, p. 10, 22). In the afroalpine belt their same time as the air temperature fluctuated be­ ecological importance is further enhanced by the tween + 3.5° and + 5°0. At 4000 m on the same moun­ fact that they recur with about equal intensity tain the temperature of the soil surface was found practically every day throughout the year. Un­ to be in daytime as a rule 1°-7° higher than that of fortunately the microclimatical observations so the air (op . cit., p. 123 f.). At 4200 m altitude on far made here are few and fragmentary, so the fol­ Ruwenzori KuLLENBERG (1962, p. 471) measured lowing account must of necessity be very sketchy. (at noon, in drizzling rain) temperatures of + 15.7° The most important factors influencing afroalpine one mm above the soil surface in a short and poor microclimates are insolation and outward (long grass vegetation, and + 13.5° - 13.9° on bare ground, wave) radiation. In daytime the intense insolation at the same time as the temperature of the air 1.5 m under a clear sky causes a strong heating of the above the ground was +3.8° - 5.6°. In another surface of the ground and the vegetation, creating instance he found (at 10.00 a.m., in sunshine) a a marked temperature gradient above the surface temperature of + 13.5° -14.1 o one mm above a (cf. GEIGER, op. cit., p. ll, 129, etc.; KuLLENBERG, bare ground surface, and +6.0° - 7.8° in the air 1962, p. 471). At an altitude of 4150 m on Kili­ 1.5 m above the ground (loc. cit.). On vegetated manjaro KLUTE (1920, p. 52) measured at noon a ground in the afroalpine belt the temperature of the temperature of 39.5°0. on the rock surface, at the soil surface will probably never reach such extreme same time as the air temperature was +8.3°0. values as recorded by KLUTE from bare rock on

Acta Phytogeogr. Suec. 49 M icroclimates 23

Kilimanjaro, since part of the incoming light is 1948, p. 44). Also in the alpine belt on Mt Elgon absorbed or reflected already above the surface small pools of water become frozen at night (BRYK, (cf. GEIGER, 1950, pp. 179, 280, etc.). Hence the 1927) . Even on the moist Ruwenzori a similar increased temperatures occurring near the ground cooling occurs-on a clear night at Bigo (3400 m in daytime may be expected to be as a rule fa­ altitude, in the ericaceous belt) the author noted vourable for the vegetation. Owing to differences in at 11.30 p.m. a temperature of -2°C at the soil cloudiness between different mountains such "local surface, with some hoar frost, whereas 1 metre heating" may be expected to be less extensive on above the ground the temperature was 0°C. heavily clouded mountains as Ruwenzori than, When camping in the alpine belt an explorer soon for instance, on Mt Kenya and Kilimanjaro. discovers that it pays better to put a sheep skin As described above (p. 9), cloud conditions on below the sleeping bag than on top of it-the cold the East African mountains are very changeable, comes from the surface of the ground. and sunny weather rarely prevails for more than As regards the amount of outward radiation­ part of the day. As soon as the sun is obscured by and hence cooling of the ground-there are evi­ clouds the temperature situation at the soil surface dently large differences within a small distance, is drastically changed. The temperature of the air owing to differences in soil cover. The strongest will as a rule drop very fast (cf. p. 18 above) and radiation and fastest cooling seems to occur on the temperature of the soil surface also seems to bare ground, whereas even a thin vegetation cover decrease rapidly (cf. KLUTE, 1920, p. 53; SALT, seems to be able to diminish appreciably the heat 1954, p. 412). In cloudy weather the temperature losses by insulating the surface (cf. GEIGER, 1950, of the soil surface is about the same as in the air p. 280) . No detailed measurements have been made (KLUTE, lac. cit.), and the plants dwell in a cool and of differences in surface temperatures between bare more or less moist atmosphere. Evidently the vi­ and vegetation-covered ground in the afroalpine gorous macroclimatic changes occurring in the belt at night, but the differences become evident afroalpine climate in daytime are further amplified by the resulting differences in frost heaving (cf. in the microclimate at the soil surface, especially p. 64 below). on sparsely vegetated ground. The low surface temperatures frequently oc­ Whereas in daytime temperature conditions at curring at night on bare ground in the alpine belt the soil surface are regulated mainly by the amount may in themselves-and in combination with of incident insolation, at night the outward long solifluction-be of great ecological significance. wave radiation is the dominant factor (cf. above p. They also have important indirect effects. When the 16, and GEIGER, op. cit., p. 15 f.). Under an over­ soil surface becomes cooler than the air immediately cast night sky so much of the outgoing radiation is above it, this air will lose heat to the soil, hence compensated by reradiation from the clouds that becoming cooler than the higher layers of air. This the temperature of the soil surface does not fall creates an inversion of the normal vertical tem­ much below that of the air (cf. GEIGER, op . cit., perature gradient, the inverted gradient being p. 19). But under a clear sky the effective outward steepest immediately above the surface (cf. GEIGER, radiation is intense in the afroalpine belt, creating op . cit., p. 24 etc.; DAUBENMIRE, 1947, p. 179). a pronounced nightly cooling of the surface of soil Meteorological conditions in the alpine belt on most and�vegetation (cf. p. 16 above). A well known of the relevant mountains seem to be very fa­ example has been provided by KLUTE (op. cit., vourable for the formation of pronounced nightly p. 54), from an altitude of 4050 m on Kilimanjaro. temperature inversions of this type (cf. p. 16 above). Whereas at 9 p.m. the air temperature in the meteo­ Furthermore, to quote from DAUBENMIRE (op. cit., rological screen was +2.4°C., the water contained p. 180): "In mountainous topography the phenome­ in a rubber wash basin placed on the ground had non of temperature inversion is magnified by the frozen solid-4 cm thick (cf. also MEYER, 1900, fact that the cold air of the upper slopes, having p. 158; TROLL, 1943b, p. 161; FRIES & FRIES, greater density than warm air, drains down ravines

Acta Phytogeog1·. S'uec. 49 24 Afroalpine environments

Fig. 10. Tanganyika, Kilimanjaro, SW slope just below the saddle, 4300 m. Vegetation zonation along a small valley trough. In the foreground Helichrysum (citrispinum var. hoehnelii) scrub with much Festuca (pilgeri ssp. supina) tussocks, e tc., on almost flat alluvial soil. The lower part of the slope in the background harbours a zone of almost pure tussock grassland, and higher up occur numerous shrubs of Euryops dacrydioides, mixed with Helichrysum and grass tussocks, and a solitary specimen of Senecio cottonii. This zonation is probably caused partly by cold air drainage along the bottom of the trough, partly by better supplies of soil water in the upper part of the rocky slope than on the porous alluvium below (as indicated by the Giant Senecio growing in the former locality). Photo 0. Hedberg 19.6.1948.

at night and slides under the mass of warm air sloping trough-shaped valleys (cf. GEIGER, op . cit., which has accumulated in the valley during the day. p. 209; TROLL, 1943 b, p. 169). Fig. 8 provides one Owing to this cold air drainage the valley floor example of this from an altitude of 4200 m in at night is occupied by a slow-flowing layer of air Teleki Valley on Mt Kenya. The lower thermo­ which increases in depth as the night progresses gram was recorded from the flat valley bottom and is cooler than the layer immediately above it (cf. Fig. 99, background), the other from the lower which makes contact with the surrounding slopes part of the valley slope less than 50 m away. and ridges". As mentioned above (p. 10), de­ In the former case the thermograph was placed 0.1 sc"ending cold air may occasionally provoke local m above the ground in the shelter of Dendrosenecio strong winds at night. leaves, in the latter it was located 0.5 m above the Such cold air drainage may cause substantial ground between a couple of big boulders. The microclimatic differences between a valley bottom boulders may to some extent have evened out the and the adjacent valley side, especially in gently temperature variations in the latter case. The

Acta Phytogeogr. Suec. 49 M icroclimates 25

thermograms are not contemporaneous but from zori was briefly studied by the present author. consecutive weeks. A direct comparison is therefore Below a steep W -exposed precipice more than 50 m difficult, the more so because the instrument was high was a well-watered talus slope with a rich not at the same level above the ground in both flora, including Carpha eminii, Crassocephal1.trn cases, and because there was a good deal of pre­ ducis-aprutii, and Carduus ruwenzoriensis var. cipitation during the week 2/8-9/8 (upper diagram) lanuriensis, all of which reach here their highest and none in the following week (lower diagram). known locality on the mountain, and furthermore But the nights 4/8-5/8 and 6/8-7/8 were judged Parietaria debilis, Callitriche stagnalis, and Senecio (from camp experience) to be equally cold as most transmarinus var. sycephyllus, which were here near nights during the following week. Consequently to their upper altitudinal limit. The favourable it appears reasonable to deduce that the night local climate evidenced by this flora cannot of minima (of clear nights) fall some 5° or 6°C below course be explained in this case by increased in­ those of the nearby valley side. Proper microclima­ solation due to favourable exposure. Apart from tic studies of sites like this one are obviously highly benefiting from cold air drainage down the slope desirable and may be expected to yield very inter­ such a locality will probably be favourably in­ esting results. fluenced by the decreased outward radiation at The more or less regular cold air drainage through night caused by screening of considerable parts of valleys and ravines can often be traced in the a p­ the sky by the rock wall behind, and by the moun­ pearance of the vegetation, since towards their tains on the other side of the valley (cf. GEIGER, upper altitudinal limit some species that are sensi­ op. cit., p. 20 f.). Edaphical factors may also be tive to frost tend to avoid the valley bottoms, particularly favourable. reaching higher on ridges and valley slopes (of. Many other afroalpine biotopes, such as narrow FRIES & FRIES, 1948, p. 9). This is particularly ravines, stream beds, and caves, certainly also have noticeable for trees and shrubs of Philippia and specialized microclimates of their own, meriting Erica at the upper border of the ericaceous belt future study. One exceptional example of this (HEDBERG, 1951, pp. 181, 182, 186, and Pl. lA, sort was studied at "Maji ya Moto" in the lower

2A). The same factor is probably also largely part of the alpine belt on Elgon. In this locality a responsible for the zonation sometimes observable hot spring creates a small enclave of ameliorated above flat valley bottoms in the alpine belt (cf. local climate, causing 18 species of vascular plants Fig. 10). Cold air drainage-or "valley phenomena" to reach here their uppermost known stations on the has even been held responsible by FRIES & FRIES East African mountains {HEDBERG, 1959). (op. cit., p. 9) for the absence of tree growth on most Finally, each plant individual and patch of vege­ of the high plateau of Aberdare, which caused tation creates to some extent a microclimate of its these authors to join the treeless grassland ex­ own. As emphasized above a plant cover interferes tending over most of that plateau with their both with the incident sunlight and the outward "alpine zone" (loc. cit.). In the latter case edaphic radiation, modifying the temperature gradient factors and fire are probably more decisive for the above the ground and ameliorating the drastic absence of trees, however (HEDBERG, 1951, p. 183). temperature climate at the soil surface (cf. GEIGER, Occasionally the incidence of "subalpine plants" op. cit., p. 179, 280, etc.). The denser a plant-or in microclimatically favoured localities on valley plant community-is growing, the better its slopes in the afroalpine belt may be so marked as possibility to create a microclimate of its own. It is to remind of the "south-bluffs" of the Scandinavian therefore not surprising to find in the afroalpine mountains (cf. ANDERSSON & BIRGER, 1912; SE­ flora numerous examples of compact growth in LANDER, 1950, p. 145; Du RIETZ , 1954; GEIGER, grass tussocks, giant leaf rosettes, mat-forming 1950, p. 223). The vegetation of such a locality at ericoid shrubs, cushion plants, etc. (cf. below). about 3850-3950 m (that is, in the lower part of the Some plants possess particular insulating devices alpine belt) in Mijusi (Mugusu) Valley on Ruwen- such as a mantle of dry leaves around the stem,

Acta Phytogeogr. Suec. 49 26 Afroalpine environments or a water filled leaf rosette round the shoot apex paralleled by those occurring m the paramo and (cf. pp. 50, 59). puna in the Andes of Tropical South America. At The most equalized microclimates of the afro­ an altitude of 3950 m on Pasta bueno in Peru alpine belt are probably found in the dense Den­ WEBERBAUER (1930, p. 345) reports that for most drosenecio forests of the Virunga Volcanoes and of the day the mean temperature of the soil surface Ruwenzori (cf. p. 97 below and Fig. 82). As de­ is 3° - 7°C higher than in the air, whereas rock sur­ monstrated by GEIGER (op. cit., p. 299 f.) and faces are 6° - 13° warmer than the air. Similarly, at others the microclimate inside a forest is much less 3600 m altitude on Cotopaxi in Ecuador the monthly extreme than that on open ground (cf. also, e.g., mean maxima of the soil surface were found to lie MoREAU, 1934, p. 64-65).1 Also dense Helichrysum 7° - 23°C higher than those of the air, whereas and Philippia scrub may be expected to have less monthly mean minima were 0.7° -2.9° lower on a changeable microclimates than open gr_ound. grass surface than in the air (EsPINOSA, 1932, p. The microclimates created by afroalpine plants 124 f.). Occasional measurements of similar kind and plant communities have so far received very have been reported, e.g., by HEILBORN (1925, p. little attention. A few scattered observations on 154) from Pichincha in Ecuador, and by DIELS some of them will be given below in connection (1937, p. 27) from Cotopaxi. A striking example of with a survey of the adaptive trends observed in the the influence of cold air drainage has been reported afroalpine flora. by WEBER (1958, p. 136 f.) from the paramos of The afroalpine microclimates seem to be largely Costa Rica.

SOILS

No less important to a land plant than the climate mountains (Ruwenzori, Elgon, Mt Kenya, and in the surrounding air is the soil on which it grows. Kilimanjaro) the processes of erosion and weathering Having no obvious economical importance, the were largely aggravated by glaciations during the afroalpine soils have, unfortunately, been very Pleistocene pluvials, which created numerous glacial little studied. Since all the relevant mountains valleys and cirques, and deposited huge amounts except Ruwenzori are of volcanic origin their of morainic material (cf. especially NILSSON, 1932, most important bed-rocks are lavas of various pp. 13, 20 f., 29 f., 39 f., PI. I, and PI. Ill). Finally, kinds, basalts, nepheline-syenite, agglomerates, and on Elgon and Mt Meru, at least, many soils seem tuffs, etc. (cf. MEYER, 1900, p. 288 f.; GREGORY, to have been derived from volcanic ashes and dust

1921, p. 144 f.; 0DMAN, 1930; RoBERT, 1946, p. (cf. 0DMAN, op . cit., map. PI. 11, and MrLNE , 1936, 117). Ruwenzori, on the other hand, consists of map). Archaean rocks such as granites, gneisses, diorites, No accurate investigation is known to me of soil amphibolites, quarzites, etc. (MICHOT , 1937, p. 207). texture in the afroalpine belt. From my own field The weathering of the bed-rocks has provided on observations it would seem that the parent material most mountains huge amounts of loose material, of most afroalpine soils-at least on Ruwenzori, which partly remains more or less in situ, partly Elgon, Mt Kenya, Kilimanjaro, and Mt Meru­ occurs in talus slopes, stream terraces and deltas, is of a fairly coarse texture, with high fractions of alluvial fans, etc. At least on four of the highest sand and little fine material. Only along lakes and meandering streams on flat valley bottoms, 1 The moistness of the microclimate in such Dendrosene­ such as the upper part of Teleki Valley on Mt cio forest is evidenced by the thick moss cushions on the Kenya, did I note soils with much fine-textured stems, and by the fact that most of the few vascular plants occurring on the ground grow equally well epiphytically material. The parent material of most afroalpine in moss cushions on Dendrosenecio trunks (cf. p. 99 below) . soils, notably on talus slopes, alluvial fans, and.

.Acta Phytogeogr. Suec. 49 Soils 27

moraines, is therefore comparatively porous with accumulation decreases. On the Shira Plateau of low water-holding capacity, but offering good Kilimanjaro, which lies in the lower part of the drainage. These observations were mainly made in alpine belt, SALT (1954, p. 379) found the soil to be the sparsely vegetated upper part of the alpine dark and peaty to a depth of more than 15 cm, belt, however, and conditions may be different in whereas in the Saddle of the same mountain, the lower part of this belt, where the closed ve­ situated in the upper part of the alpine belt, it getation may have facilitated accumulation and appeared to have little organic content (op. cit., retention of more fine-textured material. p. 400). Local variations in humus content do of The character of the afroalpine soils is, of course, course occur in the afroalpine belt as elsewhere strongly influenced by their humus contents. The according to differences in ground water supply. extent of humus accumulation seems to depend Thus in Carex runssoroensis bogs peat accumula­ largely upon the degree of moisture available, and tions may exceed a depth of 1 meter (cf. HEINZELIN, upon temperature conditions. The moister a 1953, p. 6; HEINZELIN & MoLLARET, 1956, p. 13-15; mountain is, the more humus is accumulated in its HEDBERG, 1955, p. 139 and Pl. 1 B; and p. 121 soils. Within the alpine belt the extent of humus below), whereas on more well-drained sloping accumulation appears to decrease with increasing ground at the same altitude humus accumulation altitude (i.e., decreasing temperature and de­ may be comparatively small. creasing growth rate of plants). Most of the humus Heavy humus accumulation similar to that seems to remain as a rule on top of the mineral described above from the lower part of the alpine soil. Although earthworms are common in some belt on Ruwenzori and Mt Kenya has been reported habitats at least on Ruwenzori (CoGNETTI DE also from the paramos of Tropical America (cf. MARTIIS, 1909, p. 359; HEDBERG, 1955, p. 139), DrELS, 1937, p. 30; FosBERG, 1944, p. 229; JENNY, they appear to be less active as soil mixers than 1948, p. 12; WEBER, 1958, p. 142 and Abb. 8-l l; are the worms of temperate latitudes (cf., e.g., TROLL, 1959, p. 45). The paramo soils sometimes LYON & BucKMAN, 1947, p. 96 f.). This may at contain distinct horizons of volcanic ashes (WEBER, least partly be explained by the soil reaction ( op. loc. cit.). The more desert-like soils on the uppermost cit., p. 97; cf. below). parts of Kilimanjaro and Mt Meru seem to match In the lower part of the alpine belt on Ruwenzori better those of the South American puna (cf. the dark-coloured humus layer of the almost per­ CABRERA, 1958, p. 323). manently moist soil may often be a few dm thick, The colloidal humus substances have a much and is usually rather muddy (cf. HEINZELIN & higher water-holding capacity than mineral soil, MoLLARET, 1956, p. ll f.). Even on steeply sloping and hence humus accumulation greatly increases the "bare" rock a humus layer some cm thick is pro­ water-holding capacity of the soil. The more or less vided by the dense moss-vegetation prevailing in peaty soils predominating on most mountains in such (low-alpine) localities. Because of the slippery the lower part of the alpine belt (and also in the nature of that permanently wet humus layer such ericaceous belt) must therefore evidently have a a moss cover provides a very hazardous foothold much higher water-holding capacity than the less for climbers. In the more sparsely vegetated soils organic soils at higher levels. of the upper part of the alpine be]t on the same One of the most important characteristics of a mountain the humus content may sometimes be soil is found in its reaction, which is correlated to rather insignificant, especially in well-drained soils base saturation, availability of metal ions and on talus slopes, moraines, etc. Similar conditions phosphorus, flocculation, etc. (cf., e.g. DAUBEN­ obtain on Mt Kenya-the soil under the tussock MIRE, 1947, p. 50 f.; LYoN & BucKMAN, 1947, p. grassland in the lower part of the alpine belt (on 300 f.). According to the measurements so far the W slope at least) contains a thick and muddy available the afroalpine soils seem to be on the layer of dark-coloured humus, whereas towards the whole distinctly acid. From the wettest of the upper part of the alpine belt the amount of humus relevant mountains, Ruwenzori, OsMASTON (1958,

Acta Phytogeogr. Suec. 49 28 Afroalpine environments p. 17-18) reports that the moist soils investigated RuNE, 1953; HEDBERG, l952b, p. 53). In the south­ by him had pH values between 4.0 and 5.5, and ernmost parts of Europe this differentiation be­ the peats between 3.5 and 4.5. On the same moun­ comes much less pronounced than in the northern tain FISHLOCK & HANCOCK (1933, p. 216) report and central parts, however (BRAUN-BLANQUET, a pH of about 4.5 for peat and 8 for lake water; op . cit., p. 155). And in the Australian Alps, situa­ HEINZELIN & MOLLARET (1956, p. 11-15) found a ted at a corresponding latitude on the southern pH of 4.5-5 .2 in lakes and running water, and hemisphere, "high mountain soils and vegetation HAUMAN (l933a, p. 614) measured a pH of 3.7 are not strongly differentiated according to rock at the top of a Carex runssoroensis tussock. From type", and "similar communities and soils occur on one of the driest mountains concerned, Kilima­ different rocks" (CosTIN, 1957, p. 175). In the njaro, KLOTZLI (1958, p. 41, and in litt.) reports pH afroalpine belt no evidence was found of such dif­ values of 5.1-5.25 in a number of samples from ferentiation. various plant communities in the alpine belt. These Many soils in at least the upper part of the alpine afroalpine pH values agree very well with those belt in East Africa are obviously young and imma­ reported from paramo soils in the Andes of tropical ture. Below the rim of some present-day glaciers America, which fall as a rule between 3.7 and 4.5 one may find series of such soils of very different (cf. JENNY, 1948, p. 12; WEBER, 1958, p. 143; ages, offering excellent opportunities for the study TROLL, 1959, p. 45). of soil development in a tropical-alpine climate In the mountains of the North temperate zone (HEINZELIN, 1953). Some other soils are very un­ marked differences exist between the flora on stable, partly because of active erosion and ac­ (circumneutral) soils derived from limestone and cumulation (talus slopes, etc.), partly because of other basic rocks, and that on (acid) soils derived solifluction (cf. p. 33 below). On several mountains from granites, quartzites, etc. (cf. e.g., BRAUN­ (at least Elgon, Mt Kenya, Kilimanjaro, and Mt BLANQUET, 1928, p. 145; DAUBENMIRE, 1947, p. Meru) some soils may have been partly degraded 51; FRIES, 1913, p. 228 f.; SELANDER, 1950, p. 168; by burning (cf. p. 36 below and Fig. 19).

SOIL WATER

One of the most important factors influencing the textured with poor drainage, and considerable areas distribution of plant communities is the amount of of gently sloping ground may be more or less re­ soil water available. No detailed study of this factor gularly flushed. Similar conditions obtain along has been made in the afroalpine belt, however, streams on gently sloping ground. On scree slopes and the following account is based on casual field and moraines the coarse-textured soils offer good observations only. drainage, and consequently the amounts of soil On a mountain like Mt Kenya, where precipita­ water available are as a rule limited. Where the tion seems to be fairly equal in different localities bed-rock comes near to the surface in the lower in the alpine belt, the amount of soil water avail­ part of a scree slope the mobile subsoil water may, able will be governed largely by the relief and by the however, be brought within reach of plant roots. depth and texture of the soil. On more or less flat Such localities seem to be especially favourable for ground with shallow soil, as in the bottom of glacial the Giant Senecios (cf. p. 96 and Fig. 10). The cirque valleys, the water table may reach the soil driest habitats consist of projecting eminences with surface, creating a suitable habitat for bog forma­ little or no soil cover. tion. The central part of some cirques is occupied The above observations all concern the middle and by a lake. The soil in such localities is often fine- upper parts of the alpine belt. In its lower part the

Acta PhytogeogT. S�tec. 49 Solifluction 29

Fig. ll. Needle-ice formation on a patch of bare ground in the upper part of the ericaceous belt. Early in the morning the ice-needles in the foreground were about 3-4 cm long. Tanganyika, Kilimanjaro, SE slope near Peter's Hut, 3800 m. Photo A. Holm, June 1948. relief is usually less steep, and the soils seem to con­ Large parts of the "saddle" between Kibo and tain more fine-textured material and humus than Mawenzi are, for instance, quite desert-like with a higher up (cf. p. 27). Consequently the supplies of very scanty vegetation, but in localities where soil water will as a rule be more plentiful than at subsoil water reaches the surface there occur oasis­ higher levels, and flushing seems to be common. like patches of closed vegetation (Figs. 107, 109; Comparison of different mountains reveals a cf. HEDBERG, 1951, p. 193; KLOTZLI, 1958, p. 41). wide variation in the amounts of soil water a­ On the upper parts of Ruwenzori, on the other vailable, the extremes being represented by Kili­ hand, the amount of soil water seems to be every­ manjaro and Ruwenzori. In much of the alpine where sufficient for closed vegetation, and in such belt of the former mountain the annual rainfall is plant communities as Dendrosenecio forest the soil low and erratic (cf. p. 12), and lack of soil water seems to be almost permanently moist (cf. HAUMAN, seems to be a limiting factor over vast areas. 1933a, p. 710, and p. 97 below).

SOLIFLU CTION

As emphasized above (p. 23), in most localities in the of the year. Each frosty night when sufficient alpine belt of the high East African mountains frost amounts of water are available the surface layer occurs at the soil surface on many or most nights will then become frozen to a depth of a few cm,

Acta Phytogeog1·. S�tec. 49 30 Afroalpine environments

Fig. 12. Vigorous solifluction with needle-ice formation on a patch of bare ground in the valley bottom, at about 8.30 a.m. Kenya, Mt Kenya, Teleki Valley, 4200 m. Photo 0. Hedberg 2.8.1948.

Fig. 13. Solifluction structure on the surface of a small soil terrace between tussocks of Festuca pilgeri ssp. supina. Tanganyika, Kilima­ njaro, SE slope above Peter's Hut, 4100 m. Photo 0. Hedberg 17.6. 1948.

Acta PhytogeogL Suec. 49 Solifluction 31

Fig. 14. Small solifluction terrace dammed by Pentaschistis cf. minor. The scale is given by part of a (normal-sized) foot-print at the right margin, demonstrating also the loose structure of the soil sur­ face, maintained by frost heaving. Tanganyika, Kilimanjaro, SE slope of the Saddle above Peter's Hut, 4200 m. Photo 0. Hedberg 15.6. 1948.

thawing up again next day. The alternation be­ temperature changes only, whereas the latter are tween freezing and thawing causes various types of due mainly to seasonal temperature changes (TROLL, movements in the soil, such as the formation of 1944, p. 673; 1947, p. 167; 1959, p. 27). Since the needle-ice (Swedish: "pipkrake", German: "Kamm­ diurnal frosts of equatorial mountains penetrate eis") at the surface, accompanied by heaving of only a few cm below the soil surface (TROLL, 1944, stones, roots, etc.; sorting of the coarser material pp. 562 and 608; about 2 . inches according to into polygons; and downhill soil creep. All these ZEUNER, 1949, p. 21), the frost patterns of these phenomena are embraced by the term solifluction, soils are of much smaller dimensions than those of coined by ANDERSSON (1906) for the "slow flowing arctic and subarctic countries (TROLL, loc. cit.; masses of waste saturated with water", but wi­ ZEUNER, loc. cit.). The occurrence of solifluction dened by TROLL (1944, p. 674; 1947 p. 167) to on equatorial mountains is not dependent upon an comprise all types of soil movements caused by impermeable substratum (TROLL, op . cit., p. 613, (seasonal or diurnal) freezing and thawing. Soils etc.; 1947, p. 167; 1959, p. 27; ZEUNER, 1949, p. 22). showing effects of solifluction have been called The extent of solifluction occurring on the upper frost pattern soils (TROLL, 1944, p. 547). The part of a tropical mountain is of course directly mechanisms of solifluction have been much de­ dependent upon the frequency and intensity of bated (cf. TROLL, op. cit., p. 562 f.) but need not frost temperatures at the soil surface. As mentioned concern us here-it will be sufficient to consider its above (p. 18 and Table 5) there are large differences results. And since the occurrence of solifluction on in mean diurnal temperature amplitude between the East African mountains as well as in other different mountains, and between different loca­ parts of the world has been thoroughly treated by lities on the same mountain. The importance of TROLL (op. cit.) I need not expand on it here. A solifluction may therefore be expected to be much brief review will, however, be necessary for the less on mountains like Ruwenzori, with small tem­ ecological discussion in a later chapter. perature amplitude, than for instance on Mt Kenya The solifluction phenomena of tropical high and Kilimanjaro, and to be less on ridges and slopes mountains differ fundamentally from those of than on flat ground in valley bottoms etc. Two higher latitudes in that they are caused by diurnal other important prerequisites for solifluction are a

Acta Phytogeogr. Suec. 49 32 Afroalp·ine environments

:Fig. 15. Solifluction polygons on level, relatively coarse-textured soil. Most of the polygons were 1-2 dm wide. Uganda, :Elgon, near the highest summit on the SW part of the crater rim, 4300 m. Photo 0. Hedberg 20.5.1948.

Fig. 16. Solifluction polygons on level, relatively coarse-textured soil. Width of the polygons about 1-2 dm. Tanganyika, Kilimanjaro, SE slope of the Saddle above Peter's Hut, 4200 m. Photo 0. Hedberg 15.6.1948.

_Acta Phytogeog1·. S'uec. 49 Soliflu�tion 33 more or less regular supply of soil water, and a substantial fraction of fine-textured material in the soil. The local variation in those features (cf. pp. 26 and 28 above) tends to make solifluction phenomena more important on flat or gently sloping ground than on steeper moraines, in talus slopes, etc. The most conspicuous expression of solifluction occurring on the East African mountains is the formation of needle-ice, that is, bundles of vertical ice-needles formed at the surface of more or less . fine-textured soil. These ice-needles may sometimes (especially on wet and more or less homogeneous loamy soil) be formed on top of the surface and consist of almost pure ice, carrying only a few larger particles on their tops (Fig. 11; cf. also Fig. 62 below). In other instances (especially when the soil surface is drier andfor the soil contains more of Fig. 17. Solifluction polygons on fine-textured wet soil along a stream. Width of polygons c. l dm. Kenya, Mt coarse particles) the ice-needles are less pure and Kenya, Teleki Valley below Tyndall Glacier, 4400 m. are formed below the surface, carrying substantial Photo 0. Hedberg. 29.7.1948. amounts of soil on their tops (Fig. 12). In the latter case they effect a vigorous stirring of the surface soil, heaving plant roots, stones, and similar objects above the surface, and killing any seedlings that fluction is the formation of stone polygons (on may happen to grow on it (Fig. 13). This pheno­ flat ground) and stone stripes (on sloping ground). menon may also cause pronounced erosion of Such structures were described from Kilimanjaro adjacent closed vegetation (cf. TROLL, 1944, p. 577 by KLUTE (1920), FLUCKIGER (1934), and SALT and Abb. 7). Occurrence of such needle-ice has (1954, p. 400 and Photo 9), and from Mt Kenya by been reported by ScAETTA (1933, p. 96) from an TROLL (1944, p. 607 f.), and ZEUNER (1949, p. 20). .altitude of 4100 m at the summit of Muhavura, I have noted them myself both on those mountains .and by TROLL (op. cit., p. 591) from altitudes of (Figs. 16 and 17) and on Elgon (Fig. 15). The poly­ 3900 m and higher on Mt Kenya. I have recorded gons become most obvious on gravelly-sandy soils, it myself on Mt Kenya (Fig. 12) and Kilimanjaro where the coarser fractions are sorted out along {Figs. 11, 13). On the latter mountain vigorous their edges (Figs. 15, 16), but distinct polygons may "pipkrake" was found already at the level of occur also on loamy soils without appreciable ·"Peter's Hut" (about 3800 m). It seems to occur amounts of coarse material (Fig. 17). most regularly on small spots of bare ground in Owing to the frequent and violent movements in tussock grassland, usually on more or less sloping their surface layer most afroalpine frost-pattern ground (cf. TROLL, op . cit., p. 591). Those bare spots soils seem to be difficult or impossible to colonize will often form series of steps or terraces, each for ordinary plant seedlings-they form a type of being dammed at its lower side by grass turf "mobilideserta" (TROLL, 1948, p. 49; cf. also TROLL, (Fig. 14; cf. TROLL, loc. cit., and HEDBERG, 1951, 1944, p. 562). Some adaptations favouring survival p. 193). in this extreme biotope are described below on p. Another conspicuous manifestation of soli- 64 f.

:3 - 641464 Hedberg Acta .Phytogeogr. S�ieC. 49 34 Afroalpine environments

Acta Phytogeogr. S�wc. 49 35

FIRE

One of the most important environment factors special life-forms seem to have developed in re­ acting upon the vegetation of tropical Africa is fire. sponse to this regular burning (cf. CuFODONTIS, op . From time immemorial African shepherds and cit., p. 513). hunters have set fire to the grass during the dry Two things are fundamental for the outbreak of season. This regular burning has influenced pro­ fire-presence of combustible material, and ig­ foundly not only the distribution of various types nition. In most parts of the lowlands the dry grass of vegetation, notably savanna and woodland, but provides plenty of inflammable material during the also their species composition (cf., e.g., PHILLIPS, dry season, and ignition is promptly provided by 1930; TROLL, 1952a; CuFODONTIS, 1955; SHANTZ & the Africans. Consequently most areas of savanna

TuRNER, 1958; KEAY , 1959). In some areas at len.st, and grass steppe are burned every year. On the

Fig. 20. Uganda, Elgon. View to­ wards the Suam Valley from the wagishu track not far W of Maji ya Moto. In the foreground burnt­ over area with dead Euryops shrubs, colonized by luxuriant specimens of Carduus keniensis and Peucedanum kerstenii, etc. Photo 0. Hedberg 17.5.1948 .

upper parts of the high mountains conditions are high level vegetation would rarely be dry enough different, however. According to the evidence to catch fire (cf. p. 14), whereas on the upper parts summarized above (p. ll f.) the dry seaspns are less of Kilimanjaro drought is common. Mountains pronounced on the upper parts of most mountains like Elgon and Mt Kenya seem to be intermediate than round their foot-though large variations in that respect. But occasional fires may occur occur. Thus on a mountain like Ruwenzori the even on Ruwenzori (cf. SYNGE, 1937, p. 44 f.), and

Fig. 18. Uganda, Elgon, vegetation on the inside of the crater rim NW of Maji ya Moto, 3900 m. To the right recently burnt tussock grassland, to the left and in the background unburnt vegetation. Photo 0. Hedberg 16.5.1948. Fig. 19. Recently burnt tussock grassland. Note the charred shrub remains in the foreground and the two dead Dendro­ .senecio stems at intermediate distance to the right. Uganda, Elgon, in the crater, 3800 m. Photo A. Holm, May 1948.

Acta Phytogeogr. Suec. 49 36 Afroalpine environments the matter of ignition may therefore be equally important for the occurrence of fire at high alti­ tudes. According to the evidence available most grass fires are caused by man (cf. MoREAU, 1944, p. 9; CuFODONTIS, op . cit., p. 513). Human habitations are absent from the alpine belt on the high East African mountains, and seem to occur in the eri­ caceous belt only on Elgon. On the latter mountain a number of villages are situated in the lower part of the ericaceous belt, and cattle are regularly kept at altitudes of 3000-3300 m at least. There is furthermore a lively traffic of Africans and cattle along the "wagishu track" across the caldera (HEDBERG, 1951, p. 150). Both the "elgonis" of the Fig. 21. Massive stem base of old Philippia shrub, presum­ high level villages and the travellers passing across ably repeatedly damaged by fire, in the uppermost part the caldera seem to set fire to the grass during the of the ericaceous belt. Tanganyika, Kilimanjaro, SE slope dry season. This may be the main reason why large above Peter's Hut, 4000 m. Photo 0. Hedberg 23.6.1948. parts of the ericaceous belt on this mountain are covered by grassland and not by Philippia scrub shrubs usually seem to be killed by burning, (op. cit., p. 182). Also large tracts in the alpine which therefore promotes the development of belt are burned more or less regularly (op. cit., grassland at the expense of Alchemilla scrub etc. p. 191); cf. Figs. 18-20. Alchemilla and Helichrysum (cf. Figs. 18 and 19). The bare ground between the

Fig. 22. Specimen of Senecio kenio­ dendron damaged by fire, which had burnt off its cylinders of dry leaves around the stems. Accord­ ing to Mr. V. Klarville, Naromoru, this part of the valley had been burnt in 1943, so it had taken 5 years for the Giant Senecio to form the 4-5 dm long cylinders of dry leaves shown in this figure. The Philippia shrubs in the back­ ground to the right had been killed by the fire, whereas the grass (mainly Festuca pilge1·i ssp. pil­ geri) showed luxuriant growth. Kenya, Mt Kenya, on the ridge S of Teleki V alley (along the N aro Moru route), 3900 m. Photo 0. Hedberg 12.8.1948 .

.Acta Phytogeog1". S�tec. 49 Biotic influences 37 tussocks in burnt grassland offers good opportuni­ less often visited by Mricans than Elgon and ties for colonization by certain herbs, such as Kilimanjaro, but nevertheless large areas in its Peucedanurn kerstenii, Oarduus keniensis, Swertia ericaceous and alpine belts have been repeatedly crassiuscula, Ranunculus cryptanthus, etc. (Fig. burned (cf. MACKINDER, 1900, p. 466; MOREAU, 20; cf. HEDBERG, 1957, Plates 5A and llA). 1944, p. 9; HEDBERG, 1951, pp. 185, 192). Those Also the upper parts of Kilimanjaro seem to have recurrent fires may be the main reason for the poor been recurrently . visited by Africans, and large development of the ericaceous belt on this moun­ fires have been reported several times, especially tain, and have probably contributed to the pre­ in the ericaceous belt (HEDBERG, op. cit., p. 187). dominance of grassland in much of the alpine belt Both the occurrence of a grassland (moorland) zone (HEDBERG, loc. cit.). Evidence of earlier burning is in the lower part of the ericaceous belt (HEDBERG, supplied in many places by dead and charred 1951, p. 185) and the stunted growth of Philippia Philippia stems (Fig. 21), and by the absence of and Erica shrubs in its upper part may probably part of the cylinder of dry leaves on stems of be attributed to the influence of burning (cf. Fig. Senecio keniodendron (Fig. 22). 21). In September 1963 the SE slope of Kilima­ Evidence of burning in the ericaceous and alpine njaro suffered from a big fire, believed to have been belts is available also from most of the other high started by honey-collectors, and reaching to the mountains (HEDBERG, 1951, pp. 179, 180, 181, 183, saddle between Kibo and Mawenzi. Also on the SW 187). Obviously fire is a factor of considerable slope around Umbwe Valley large areas of forest ecological significance for the afroalpine flora, as and ericaceous scrub were burned, from about well as for several other high mountain floras in the 2400 m altitude up to the upper limit of the eri­ tropics (op. cit., p. 179; cf. also BRASS, 1941, pp. caceous belt at about 3800 m (S. W. Bie, letter and 330 and 332). The importance of fire at high level personal communication). In the alpine belt no in the Andes of South and Central America has traces of fires were seen by the author on this been testified, int. al., by GoEBEL (1891, p. 5), mountain in 1948-presumably the vegetation FosBERG (1944, p. 228), TROLL (1959, Bild 14), here is too open. and WEBER (1958, p. 158). The upper parts of Mt Kenya may have been

BIOTIC INFLUENCES

Biotic influences are certainly as manifold in the by the introduction of weeds, such as Oxalis afroalpine belt as elsewhere. In the absence of corniculata, Malva verticillata, and Poa annua. detailed investigations, however, I can only give a A similar decimation of Giant Senecios will pro­ brief survey of some more conspicuous examples. bably also occur through visitors to the other high The direct influence of rnanis deemed to be negli­ mountains, because in rainy weather the dry leaf gible in the alpine belt on most of the mountains, cylinders of those plants can always be relied upon particularly on Ruwenzori (cf. HAUMAN, 1933a, to provide dry fuel (cf. p. 50 below). p. 710). On Mt Elgon habitations occur far up in the Grazing and browsing by cattle, buffalo, and other ericaceous belt, however (cf. p. 36), and the alpine large herbivores seem to be rather important on belt is regularly traversed by Africans and cattle. Elgon, less so on the other high mountains. Tramp­ Those travellers certainly influence the alpine ling by those animals may also be important in vegetation to some extent both by collecting fire­ places, for instance near Maji ya Moto on Elgon wood (probably largely Dendrosenecios) and by the (Fig. 23). A list of the mammals occurring on the burning of grassland (cf. p. 36 above), as well as upper parts of Kilimanjaro and Mt Kenya has

.Acta Phytogeogr. Suec. 49 38 Afroalpine environments

Fig. 23. Uganda, Elgon, in the

crater, 3800 m. In the centre of the picture appears a group of Senecio elgonensis, surrounded by Alche­ milla (elgonensis) scrub; in the fore­ ground and in the background heavily grazed grassland. Note the numerous parallel (buffalo or cattle) tracks along the hillside in the background. Photo 0. Hedberg 19.5.1948.

been provided by MoREAU (1944); as regards the and presumably pollinated-by sunbirds (Necta­ latter mountain see also MoREAU 1945. Locally the rinia iohnstonii and allied species). Already MAcKIN­ browsing by small rodents may be very intense, as DER (1900, p. 466) observed that those birds fre­ in the upper part of Teleki Valley on Mt Kenya quent the Giant Lobelias on Mt Kenya, and similar (Fig. 24; cf. MACKINDER, 1900, p. 466). observations have been made several times on Dunging may be important in certain afroalpine various mountains, e.g., by HAUMAN (1933b, p. 16), localities, notably outside Hyrax dwellings, on bird BuRGEON (1937, p. 272 and Fig. 125), MoREAU stones, etc. Outside a Hyrax cave at about 4400 m (1944, p. 24), JEANNEL (1950, p. 215), and HED­ altitude on Kilimanjaro I found a rich and dense BERG (1957, p. 377); cf. also Figs. 26 and 27. vegetation with exuberant specimens of Senecio On Mt Kenya these birds seem to favour parti­ telekii, S. purtschelleri, etc., and similar localities cularly Lobelia keniensis. Flowering specimens of on Mt Kenya were found to harbour uncommonly that plant growing close to a sunbird nest are some­ vigorous specimens of Anthoxanthum nivale, Sedum times visited so often that their bracts become more ruwenzoriense, and Arabis alpina. Bird stones were or less worn out (see photograph in HEDBERG, found to harbour the same type of lichen vegetation 1957, PI. 6B). The importance of the sunbirds for as in Northern Europe (Fig. 25). the pollination of Giant Lobelias has been discus­ The importance of earth worms for soil formation sed by HAUMAN (loc. cit.). Also the Giant Senecios seems to be small on these mountains. Such ani­ seem to be sometimes frequented by sunbirds, mals do occur quite commonly at least on Ruwen­ and may perhaps at times be pollinated by them zori (cf. CoGNETTI DE MARTIIS, 1909; HAUMAN, (cf. GRANVIK, 1923, p. 224; HAUMAN, 1935, p. 27). 1933a, p. 710; HEDBERG, 1955, p. 139), but they To what extent insect pollination occurs in the seem to live largely in wet habitats such as afroalpine belt is unknown. According to HAUMAN bog Sphagnum, Carex runssoroensis tussocks, (1933a, p. 716; 1933 b, p. 16; 1935, p. 27) flying in moss cushions, below lichens, etc. (HEDBERG, insects capable of effecting pollination are practi­ op . cit.). In the sample areas studied by the author cally absent from the alpine belt on Ruwenzori. their importance as soil mixers seemed to be in­ Considerable numbers of small insects must significant (cf. p. 27 above). undoubtedly occur here, however, because the sun­ As regards pollination very few observations are birds seem to depend on them for most of their food available. The best known plants in that respect are (cf. HAUMAN, 1933b, p. 16; 1935, p. 27). Numerous the Giant Lobelias, which are regularly visited- records exist of insects from high levels both on

Acta Phytogeogr. Suec. 49 Biotic influences 39

Fig. 24. Vegetation heavily grazed by rodents (mice). The closely trimmed grass tussock in the centre is formed by Festuca pilgeri ssp. pilgeri, and the left-hand part of its front is covered by a cushion of Myosotis keniensis. In the background a rodent path, in the foreground droppings. Kenya, Mt Kenya, Teleki Valley, 4200 m. Photo 0. Hed­ berg 13.8.1948.

Ruwenzori and the other high East African moun­ and some flower visitors were noted (op. cit., p. tains (see lists of references in HANCOCK & SouNDY 392). On Mt Kenya some butterflies (Lycaenidae) 1931; FISHLOCK & HANCOCK, 1933; DE GRUNNE occur rather frequently up to at least 4200 m et al., 1937; JEANNEL, 1950; SALT, 1954). To what (author's observations). extent some of those insects may effect pollination As regards competition very little can be said at can only be determined by intense field studies by present, eiKcept that it must be more severe in the specialists. lower part of the alpine belt, where the vegetation In the upper part of the alpine belt on Kili­ is as a rule closed, than at higher levels, where it is manjaro the possibilities for insect pollination more or less open. Because of the virtual absence would seem to be largely restricted by the cryptozoic of competition in the uppermost part of the alpine habit adopted by all insects observed (SALT, op . belt the plants tend to occur here largely at random, cit., p. 413). Towards the lower limit of this belt and do not form recognizable associations (cf. the insects seemed to be less sedentary, however, SAVILE, 1960).

Fig. 25. Bird stone with rich vege ­ tation of nitrophilous lichens (Ra­ malina sp., Gyrophora sp., Parmelia sp., etc.). Tanganyika, Kilima­ njaro, In the S part of the Saddle, 4250 m. Photo 0. Hedberg 23.6. 1948 (from colour photograph).

Acta Phytogeogr. Suec. 49 40 Afroalpine envir·onments

Fig. 26. Fresh snow on afroalpine vegetation with Senecio adnivalis coll., Lobelia wollastonii, and Heli­ chrysum stuhlmanii. The small bird perched on one of the upper­ most Senecio leaves to the right is a sunbird, Nectarinia cf. dart­ mouthii. Uganda, Ruwenzori, be­ tween Kitandara and Freshfield Pass, c. 4200 m. Photo A. Holm 10.4.1948.

Fig. 27. (A and B) Fledgling of a sunbird, Nectarinia cf. johnstonii, making one of its first visits to a Lobelia (keniensis} inflorescence in search of food. In Fig. B the bird is stretching its beak towards a flower. Note the old claw-marks of sunbirds in the leaves and bracts. Bird pollination seems to occur regularly in all afroalpine Giant Lobelias. Kenya, Mt Kenya, Teleki Valley, 4200 m. Photo 0. Hedberg 2.8.1 948.

Acta Phytogeogr. S1wc. 49 11. Adaptive trends in the afroalpine flora

INTRO DUCTION

In order to subsist in a given environment a plant on some mountains (notably Ruwenzori); violent species must of course to some extent be adapted temperature changes with rapid heating after to the latter. Adaptation to extreme environments, sunrise and rapid cooling after sunset on most of such as deserts, seashores, arctic tundra, etc., is the others (especially Elgon, Mt Kenya, and Kili­ usually manifest not only in purely physiological manjaro). Cf. p. 18. features but also in gross morphology and anatomy (3) Solifluction. On most of the mountains con­ (cf., e.g., VoLKENS, 1887; ScHIMPER, 1898; BEws, cerned there is intense frost-heaving by needle-ice 1925; DAUBENMIRE, 1947). The amount of such on all bare soil in the alpine belt. Cf. p. 33. easily observable adaptations occurring in a flora (4) Drought. Desert-like conditions prevail in will be related, of course, to its age in the environ­ much of the alpine belt on Kilimanjaro and Mt ment concerned and to the extremeness of the latter. Meru. On the other mountains dry spells of various From the account given above it:should be evi­ length may occur. Drought-like conditions may dent that the afroalpine environment-or group of probably also be ea used even on moist ground by environments-is very peculiar, with no thermal the rapid rise in air temperature in the morning, seasons but pronounced diurnal variations ("sum­ causing strong transpiration while the soil surface mer every day and winter every night"), with may still be frozen, and hence the rate of water strong insolation and outwards radiation, frequent absorption slow (cf. KRAMER, 1949, p. 321; 1956, frost heaving on bare soil all the year round, etc. p. 148). Large differences do of course occur between the The severity of these obstacles is testified by the afroalpine climates of different high mountains in poverty in species of the afroalpine flora-each East Africa-there is a wide gap between the cold mountain harbours only some 60-175 vascular oligothermic and perpetually moist climate of plant species in its alpine belt. Most of those species, Ruwenzori and the desert-like one in the Saddle of incidentally, show no close affinities to the flora of Kilimanjaro, with violent changes in insolation, the adjacent lowlands (HEDBERG, 1961). temperature, and humidity. But the basic pattern In an environment of this kind one would expect is essentially the same. A detailed analysis of the to find a number of conspicuous adaptations. That prime obstacles to plant growth presented by this is indeed the case, as will be shown in the rest of environment is at present out of reach, but ac­ this chapter. It need hardly be stressed that the cording to my field observations and relevant litera­ following account, being based in the main on a ture the following would seem to be some cardinal few days of intense field work on Mt Kenya, to­ factors: gether with occasional observations made in the (1) Radiation. Intense insolation with much course of other work on the same and other moun­ ultraviolet and infrared light (under a clear sky) in tains, is of a highly preliminary nature. Further daytime; intense outward radiation (under a intense ecological field work is necessary to test the clear sky) at night. Cf. p. 15. validity of the conclusions reached. I hope the (2) Temperature conditions. Frequent night frosts present publication may stimulate such work in the all the year round; continuously low temperatures future, both on the high East African mountains

Acta Phytogeogr. Suec. 49 42 Adaptive trends and in other similar environments, as the Andes of below) belong to genera which have displayed Equatorial South America. An ideal approach evolutionary plasticity also by forming numerous would be to make comparative studies in those two vicarious taxa on different mountains (HEDBERG, areas, investigating the parallel ecological adapta­ 1957' p. 376). tion of widely different genera to an equatorial­ The intensity of some of the adverse environment alpine climate. factors shows a gradual increase with increasing One must not expect of course to find equally altitude, e.g. frequency and intensity of night frosts, conspicuous adaptations in all afroalpine plants. intensity of insolation, etc. Investigation of a plant In the first place their histories differ-some belong occurring at a wide range of altitudes may some­ to an old afroalpine or afromontane flora element, times reveal a corresponding clinal variation in while others appear to be comparatively recent some morphological feature deemed to be adapti­ boreal or austral immigrants (HEDBERG, 1961). vely significant, such as leaf size, pubescence, etc., The latter may hold on to the same morphological from lower to higher altitudes, as in Helichrysum life-form as they-or their close relatives-have stuhlmannii on Ruwenzori (HEDBERG, op. cit., in other areas (e.g., Arabis alpina, Arabidopsis p. 344), and in Crassula granvikii (HEDBERG, thaliana, Deschampsia caespitosa and D. flexuosa, op . cit., p. 280; 1959, p. 163). On Muhavura in the Romulea spp., Euryops spp., Subularia monticola, Virunga Volcanoes a similar trend was noted for the etc.), although the physiological make-up may be Giant Senecios, grading from typical Senecio erici­ expected to differ in the afroalpine representatives. rosenii (with thin stems and almost glabrous leaves, Secondly, the rate of evolution appears to differ etc.) in the lower part of the ericaceous belt to more widely between different groups (see HEDBERG, or less typical S. adnivalis var. alticola (with thicker 1957, p. 376, for examples from the afroalpine stems and densely pubescent leaves, etc.) at the flora). Some genera may be more "morphologically summit (HEDBERG, 1957, p. 359). Comparable conservative" than others, although they have trends have been demonstrated by HAUMAN (1935, become sufficiently adapted physiologically to p. 29) for the Dendrosenecios of various other subsist in this environment. It is worth remembering mountains. As a rule those clines are discontinuous, in this connection that the most conspicuously however, because of sharp taxonomic boundaries specialized life-forms of the afroalpine flora, the between the populations predominant at low and Giant Senecios and Giant Lobelias (cf. pp. 49-61 high levels, respectively (cf. p. 43 below).

PROTECTION AGAINST RADIATION

As emphasized above (p. 22), in sunny weather the soil temperatures make water absorption from the insolation is so intense in the afroalpine belt that ground and water transport through root and stem bare rock exposed to the sunshine rapidly becomes comparatively slow (cf. GoEBEL, 1891, p. 11; hot to the touch, and that unprotected human skin HAUMAN, l933 a, p. 714; TROLL, 1959, p. 46). may get severely burned in a short time. And The high intensities of ultraviolet radiation at these sunny weather does occur on all of the high East altitudes may also be deleterious. Any morpholo­ African mountains-even on Ruwenzori. Plant gical trait providing protection against strong leaves and stems exposed to this intense radiation radiation would therefore be expected to confer a will also of course be subjected to strong surface selective advantage in this biotope. Such traits heating-unless they possess some kind of pro­ are in fact common in the afroalpine flora. tection. Such strong heating of the leaves might The most conspicuous example of protection not be so easily relieved by increased transpiration against strong insolation is found in the thick and in this environment, where the predominant low light-coloured indumentum covering one or both

Acta Phytogeog1·. S1tec. 49 Protection against radiation 43

Fig. 28. Flowering bush of Heli­ ch1·ysum newii. The leaves are co­ vered by a thick silvery-white in­ dumentum, and the bracts are shining white-both are liable to reflect most of the incident radia­ tion. Tanganyika, Kilimanjaro, 8 slope above Peter's Hut, 4200 m. Photo 0. Hedberg 15.6. 1948. sides of the leaves (and/or the bracts, peduncles, on (the lower side of) the leaves (viz. S. adnivalis stems, etc.) of several afroalpine species (cf. var. alticola, S . .friesiorum, S. barbatipes, and S. HAUMAN, 1933a, p. 714 etc.; HuMBERT, 1935, cottonii) have the highest altitudinal ranges of p. 838 etc.; SALT, 1954, pp. 385, 412). Species with distribution on their respective mountains, whereas such pubescence are particularly numerous in the only the fifth one (S. brassica) is surpassed by a genera Alchemilla (A. elgonensis, A. argyrophylla non-pubescent species. In some other cases high­ s.lat., A. stuhlmannii, etc.), Helichrysum (H. level specimens of a species with a wide altitudinal .splendidum, H. citrispinum s.lat., H. newii, H. range tend to be more densely pubescent than those brownei, H. amblyphyllum, H. chionoides, H. at lower levels, as in Helichrysum stuhlmannii on gloria-dei, etc.), and Senecio (S. brassica, S frie­ Ruwenzori and in Senecio erici-rosenii-S. adni­ .siorum, S. barbatipes, S. cottonii, S. adnivalis var. valis var. alticola on Muhavura (cf. p. 42 above) . alticola, S. telekii, S. keniophytum, etc.). Some How much of the incident radiation is reflected examples are shmvn on Figs. 28, 39-42, 48, and by a densely pubescent leaf of the type mentioned 60, etc., in this book. Further illustrations may be here can only be decided by experiment-this found, for instance, in FRIES & FRIES, 1922a, might prove an interesting scope for field ecology. Figs. 1 and 8; HAUMAN, 1935, PI. IX etc.; DE That they do reflect much of it is evident from ocu­ GRUNNE et al. 1937, Fig. 34 etc.; JEANNEL, 1950, lar inspection as well as from photographs (cf. PI. XIX, XXXI, etc.; SALT, 1954, Photos 6 and 7; SAr� T, 1954, p. 412). Probably a pubescence of this HEDBERG, 1951, PI. VB, VI B; 1957, Pls. 1 and 12. sort will serve equally well to diminish outward Such densely pubescent plants are commoner on the radiation at night, hence contributing to reduce dry eastern high mountains than on Ruwenzori, the nightly cooling of the vegetation (cf. p. 23). and appear to be more important in the alpine belt Considerable numbers of plants with strongly than lmver down on the mountains. It may be pubescent leaves occur also in the paramos of the worth noting in this connection that four of those South and Central American Andes (cf., e.g., five Dendrosenecios possessing dense pubescence GOEBEL, 1891, p. 16; WEBERBAUER, 1905, p. 77;

Acta Phytogeogr. S�wc. 49 44 Adaptive trends

ESPINOSA, 1932, p. 198; TROLL, 1958, p. 60; 1959, p. 47). The pubescence of the Giant Senecios may deserve some special comment. The fact that dense pubescence occurs as a rule on the lower side of their leaves might at first sight appear peculiar, since the upper side would be expected to be more exposed to the sunshine. This pubescence may, however, help to protect the huge leaf rosettes of those plants against strong outward radiation at night, when the leaves of each rosette are folded together (cf. Fig. 40, and p. 52 below). Pubescence of the sort described here may cer­ tainly be ecologically important also in other res­ pects. A dense indumentum covering the stomata will tend to diminish air circulation immediately above the latter, and hence to reduce transpiration (cf., e.g., VOLKENS, 1887, p. 45; DAUBENMIRE, 1947, p. 161). In the Giant Senecios just mentioned the stomata are located on the lower side of the

Fig. 29. Young specimen of Senecio adnivalis var. adnivalis flowering for the first time. Note the shiny, light-reflecting upper surface of the leaves, and the innovation shoots at the base of the inflorescence. Height of the portion shown in the figure about 1.8 m. Uganda, Ruwenzori, S slope of Mt Speke, 4350 m. Photo 0. Hedberg 7.4.1948.

Fig. 30. Tussock grassland, domi­ nated by Festuca pilged ssp. pilge1·i, on almost level alluvial soil. The small dark bodies on the surface of the soil are loose-lying moss balls (cf. p. 66 and Figs. 63- 65). Kenya, Mt Kenya, Teleki Valley, 4200 m. Photo 0. Hedberg 11.8.1948.

Acta Phytogeogr. S1tec. 49 Temperature insulation 45

leaf, and hence protected by the indumentum (cf. Haplocarpha rueppellii (see HEDBERG, 1957, Pl . HAUMAN, 1935, p. 22; HARE, 1941, p. 366). A high 5B), in some of the Giant Senecios (e.g., S. ad­ light intensity is liable to promote rapid transpira­ nivalis and S. keniodendron-cf. Figs. 29 and 35), tion (cf. DAUBENMIRE, op. cit., p. 238), and as and in a few Giant Lobelias (notably L. keniensis emphasized above (p. 41) water uptake and trans­ and related species-cf. Figs. 52, 53, 57, and 58). port seem to be much retarded by the low tempera­ This shiny leaf surface seems to be often combined ture in this biotope. A dense hair cover may there­ with a thick hypodermis, considered by H UMBERT fore be ecologically significant also in this connec­ (1935, p. 839) to offer good protection against tion. The pubescent leaves of several afroalpine ultraviolet radiation. and paramo plants have usually been interpreted as The shiny involucra! bracts in Helichrysum must xerophytic adaptations (HAUMAN, 1933a, p. 714; certainly also have good capacity for light reflection GoEBEL, 1891, p. 17; cf. also p. 71 below). (cf. Fig. 28, etc.). Also the dense spinescence oc­

Perhaps equally efficient as :1 dense pubescence in curring, e.g., in Carduus keniensis, will certainly reflecting much of the incident radiation may be a serve to reflect much of the incident radiation-and shiny leaf surface. This occnrs, for instance, in to decrease outward radiation at night; cf. Fig. 59.

TEMPE RATURE INSULATI ON

It has been shown above that in the afroalpine belt animals (cf. SALT, 1954, p. 411 f.). Both the re­ seasonal variations in temperature are insignificant current nightly freezing and the rapid warming up as compared to diurnal ones (p. 18). It also appears in the morning would be expected to be deleterious that on most mountains seasonal variations in to many plants, notably to their most vulnerable precipitation are less pronounced in this belt than parts such as shoot apices and flower buds. It was .lower down (p. 13). Apart from spells of excessive therefore but natural to look for structures fa­ drought (which seem to be common in the alpine vouring temperature insulation among afroalpine belt on Kilimanjaro) the most critical period for an vascular plants. Several types of such structures afroalpine vascular plant mus1:; therefore be neither were found, and for some of those a few crude winter nor dry season-but the night. Everywhere temperature measurements were made. In the jn the afroalpine belt night temperatures at or following a review is given of the most conspicuous below freezing are normal, a· d in its upper part types. ·on at least Elgon, Mt Kenya, Kilimanjaro, and Mt Big grass tussocks Meru hard frosts occur on many or most nights ·of the year-although their intensities vary ac­ It was noted already by VoLKENS (1897, pp. 311, cording to variations in topography (cf. p. 23 314; cf. also MEYER, 1890, p. 282) that the grassland above). Shortly after daybreak the intense insola­ on the upper parts of Kilimanjaro is formed by tion will cause as a rule a rapid warming of the sur­ firm tussocks, separated by a network of lower face of the soil and vegetation, and the air tempera­ vegetation or open soil. Similar observations were ture rises sharply. All unprotected plant tissues in made by MACKINDER (1900, p. 466) on Mt Kenya. this biotope will therefore b exposed to strong This rich occurrence of big grass tussocks (Fig. 30), freezing on most nights throughout the year, to be dominating the vegetation over vast areas, is in quickly warmed up again by sunshine in the morn­ fact one of the most characteristic features of the .ing. An environment of this hnd must pose some afroalpine vegetation, especially where burning has ,difficult problems for vascular plants as well as for occurred (Figs. 13, 18, 19, 22, 30, and 69). A dia-

Acta Phytogeogr. Stwc. 49 46 Adaptive trends grammatic drawing of a section through such a tussock is shown in Fig. 31. Throughout its basal part the culms are intermingled with dead and more or less decaying leaves and culm bases, form­ ing a dense felt-like mass. The innovation shoots are formed mainly in the inner part of the tussock, where they are protected against environmental changes in temperature and moisture by a layer of this felt about 5 cm or more thick. In order to get a rough estimate of its insulating capacity I made a couple of temperature measurement� in such a tussock. Shortly before sunrise on a frosty morning after a cold night I measured among the outermost leaves of the tussock (at A in Fig. 31) a temperature of -5°C. Among the innovation shoots in the central part of the tussock (at B in the same figure) the temperature was at the same time + 2.5°C, and in the uppermost part of the rhizosphere (at C in the figure) it was + 3.0°C. A tussock of this kind evidently provides a very good temperature insulation, creating for its most sensitive parts an ameliorated microclimate of its

own. Also moisture variations in the environment Fig. 31. Diagrammatic section through a tussock of Festuca seem to be largely buffered inside the tussock­ pilgeri ssp . pilgeri at 4200 m in Teleki Valley on Mt Kenya, after several days of dry weather the central part drawn approximately to the scale indicated. The living of such a tussock was found to contain a good deal culms and leaves were largely concentrated to the central part of the tussock, whereas the marginal part consisted of free moisture in its interior. Because of this in the main of dead and decaying leaf- and culm bases, good temperature insulation no frost heaving occurs offering good temperature (and moisture) insulation.

Fig. 32. Crescent-shaped centri­ fugally growing tussock of Penta­ schistis cfr. minor on solifluction soil. Tanganyika, Kilimanjaro, in the Saddle above Peter's Hut, 4400 m. Photo 0. Hedberg 19.6. 1948.

Acta Phytogeogr. Suec. 49 Temperature insulation 47

Fig. 33. Senecio keniodendron woodland with undergrowth of Alchemilla argymphylla ssp. argyrophylla. The leaf ro�ettes of the Giant Senecio are in "day position" (cf. the text p. 49). Note the thick bark on the older Senecio stems, with epi­ phytic moss cushions of Tortula cavallii. In the background to the right a dead Senecio tree, from which the bark has fallen off. Kenya, Mt Kenya, Teloki Valley, 4100 m. Photo 0. Hedberg 3.8.1948.

below the tussocks or in their immediate neigh­ Pentaschistis spp., and on Mt Meru Poa leptoclada bourhood (cf. p. 64 below). A detailed investigation and Pentaschistis spp. The tussocks formed by of the diurnal temperature variation in such tus­ Agrostis, Poa, and Pentaschistis are usually smaller socks by means of thermocouples \vould certainly than those of Festuca, and especially the Pen­ be very interesting. At least in Teleki Valley on Mt taschistis tussocks often die off in the centre, forming Kenya their genial microclimate was exploited crescent-shaped structures, sometimes damming by a sun bird, Nectarinia johnstonii, who built its small solifluction terraces (Figs. 14 and 32; cf. covered nest in tussocks of Festuca pilgeri. TROLL, 1944, p. 670 and Fig. 72; PICHI-SERMOLLI, The most important tussock-forming grasses of 1957, p. 762; CABRERA, 1958, Pls. XII: 2 and XIII: the alpine belt appeared to be on Elgon Festuca 2). This phenomenon seems to be most frequent on pilgeri ssp. pilgeri and Andropogon amethystinus, Kilimanjaro. On Ruwenzori tall grass tussocks are on Mt Kenya Festuca pilgeri ssp. pilgeri and rather infrequent, being mainly formed by Deschamp­ Agrostis trachyphylla, on Kilimanjaro Festuca pilgeri sia caespitosa (s.lat.), but the giant tussocks of ssp. supina, F. kilimanjarica, Poa leptoclada, and Carex runssoroensis give much the same impres-

Acta Phytogeogr. Suec. 49 48 Adapt,ive trends

Fig. 34. Leaf rosette of Senecio keniodendron in "night position", densely folded together with re­ volute leaves. Especially on the sun-exposed E side (to the left) the leaves have already begun to unfold. Width of rosette about l m. In the background unfolding leaf rosettes of Lobelia keniensis. Kenya, Mt Kenya, Teleki Valley, 4200 m. Photo 0. Hedberg 27.7. 1948.

sion (Fig. 97; cf. HEDBERG, 1955, Pis. lA and 2A). 1890: "Zwei Drittel der BHitter des Grases sind Similar large Carex tussocks occur also on some of immer trocken und von braunlich-gelber Farbe, the other mountains, notably on Elgon (Fig. lOO) wahrend die neuen blaugrtinen mit jenen dieses and Kilimanjaro (Fig. 105). olivgrtine Kleid erzeugen, welches alle Berge Large and dense grass tussocks of the type de­ uberzieht und pajonal genannt wird" (both quoted scribed above, though formed by other genera, are from TROLL, 1959, p. 19; cf. also DrELS, 1934, pp. physiognomically very important also in the paramo 59-60). The ·widespread occurrence of such big and puna of South and Central America, in suban­ grass tussocks in equatorial-alpine regions and in tarctic countries, and in other areas with similar areas with similar climates (as the subantarctic climejtes (cf., e.g., GoEBEL, 1891, p. 10; CuAT­ islands and the mountains of New Zealand) de­ RECASAS, 1934, p. 105 and Pl. XXII; 1958, Pis. monstrate that this life-form is well adapted to XV, XIX, and end plate; DrELS, 1937, p. 28; their peculiar climate. And the observations re­ FOSBERG, 1944, p. 229; TROLL, 1948, p. 57; 1955, corded above indicate that its adaptive value stems, p. 719; 1958, pp. 58-59, and Figs. 55 and 67; 1959, in part at least, from the good temperature (and pp. 39 , 53, 78, and Figs. 21-22; WEBER, 1958, pp. moisture) insulation provided to the central part of 159, 165, 189, and Pis. 50, 72, 73, 101, and 102; these tussocks. The same morphological peculiarity CABRERA, 1958, Pls. IX, XII, and XIII; SEWELL may obviously possess selective advantage also in et al., 1954, Figs. 2, 5, and 10). Also in the grass other environments. Thus according to DAUBEN­ tussocks of the South American paramo the dead MIRE (1947, p. 159): "The leaves of xerophytic leaves and culm bases long remain, intermingled grasses die back during long drouths, but remain with the living parts. This phenomenon was noted in place and act as a mulch protecting the peren­ already in 1874 by SoDIRO, and WoLF stated in nating buds which lie at the soil surface".

A.c"taPh ytogeogT. S1wc. 49 Temperature insulation 49

Fig. 35. Unbranched young indivi­ dual of Senecio keniodendron, about 1. 7 m high. Part of the leaf rosette has been removed in order to show the compact central cone of young leaves firmly pressed together. Note the imbricate arrangement of the patent old leaves, which makes the rain water run off on the outside of the marcescent leaf cylinder, while its interior is kept constantly dry. Kenya, Mt Kenya, Teleki Valley, 4200 m. Photo 0. Hedberg 5.8.1 948.

Giant leaf rosettes in the alpine belt on Mt Kenya, with notes on the temperature-insulating structures observed in them. Other conspicuous examples of structures providing temperature insulation were found among the SENECIO KENIODENDRON Giant Senecios and Giant Lobelias. These plants Perhaps the most famous of the relevant giant are characterized by thick and unbranched or rosette plants is Senecio keniodendron, some speci­ sparsely branched woody or herbaceous stems, each mens of which are shown in Fig. 33. Its leaf rosettes carrying a continuously growing huge leaf rosette, are about one metre wide, each leaf being some five terminated in due time by an immense inflorescence. decimetres long. In daytime the outermost leaves Because of their continuous growth the shoot are more or less flat and patent, but at night apices of these plants have no dormant stage cor­ practically all leaves are bent inwards towards the responding to the winter buds of Boreal deciduous centre of the leaf rosette with their margins strongly trees, but as will be shown below their large leaf reflexed (Fig. 34). Not even in daytime are all rosettes serve at night as giant "night buds". In leaves patent, however; the young ones are erect the following a short account is given for each of and more or less appressed to the central core of the the Giant Senecios and Giant Lobelias occurring rosette, which is formed by a dense cone of young

4-64146± Hedberg .Acta Phytogeogr. Suec. 49 50 Adaptive trends

revolute yellowish leaves (Figs. 35 and 36). The apex and young leaf primordia will be effectively old and decaying leaves are not shed but remain protected against frost. The same protection is attached to the stem, forming on older stems a given to the young inflorescence, which is formed in dense cylinder of dry and in the end almost mum­ the centre of the leaf rosette (Figs. 37, 38). When it mified leaves (Figs. 35, 37). Especially in night emerges above the protecting leaf rosette the meio­ position a rosette of this kind functions as a huge sis has probably already occurred. The ameliorated bud, offering very good protection for the shoot microclimate inside these leaf rosettes is exploited apex against environmental changes. Its capacity of by some insects-thus a group of Curculionidae, temperature insulation may be illustrated by a feeding on Senecio leaves, are common in this couple of temperature readings. On a frosty morning biotope (cf. BRYK, 1927; HAUMAN, 1935, p. 26). after a clear night I measured inside such a rosette But the interior of the leaf rosette is not the only (at the surface of the dense central cone) a tempera­ part of this plant enjoying good temperature in­ ture of + l.8°C, while the air temperature among sulation-also the stem profits from this. As men­ the outermost leaves was -4.0°C. Inside the central tioned above the dead leaves remain attached to cone the temperature changes of the environment the stem, forming a dry cylinder of more or less will certainly be even better buffered, and the shoot mummified leaf remains (Figs. 35, 37, etc.; cf. MACKINDER, 1900, p. 466; ALLUAUD & JEANNEL, 1914, p. 643; JEANNEL, 1950, p. 210). Its insulating efficiency is indicated by a couple of temperature measurements. One early morning when the tem­ perature of the surface of such a dry leaf cylinder was -5.0°C. I measured in the outermost part of the pith (in a stem cut the moment before) +3.0°C. The ecological significance of this insulating cylinder is not difficult to visualize. Because of the rapid rise in temperature after sunrise in this environment and the strong insolation (cf. pp. 15, 18) the trans­ piration of most plants is likely to be strong already in the early part of the day. An unprotected stem of the same dimensions growing in this biotope would probably get most of its water-conducting tissues frozen almost every night of the year, to be thawed out again fairly late in the day. This would evidently seriously impede the replacement of water lost by transpiration. The matter of main­ taining the water balance may be difficult enough even with an unfrozen stem in this biotope, where the prevailing low temperatures make water ab­ sorption from the soil and its transport through roots and stem comparatively slow (cf. KRAEMER� 1949, p. 231, etc.,). Another ecological peculiarity should also be noted in this connection. As em­ phasized by HARE (1941, p. 365) the Giant Senecios Fig. 36. Leaf rosette of Senecio keniodendron from which all possess uncommonly large amounts of storage mature leaves have been removed to show the central core tissue in pith and cortex. "Both tissues are char­ of young leaves firmly pressed together around the shoot apex. Kenya, Mt Kenya, Teleki Valley, 4200 m. Photo 0. acterized by cells with thin walls, scanty cytoplasm Hedberg 5.8.1948. and a notable absence of any solid cell contents�

Acta Phytogeogr. S�tec.49 Tempemtu1·e insulation 51

Fig. 37. Young tree of Senecio keniodendron in longitudinal sec­ tion. Around the stem with its thick pith (grayish in the photo), thin secondary wood (white in the photo), and thin cortex (gray), there is a dense insulating mantle of marcescent dry leaves. The stem carries a huge leaf rosette, in the centre of which appears a young inflorescence. The latter ie protected by a multitude of densely folded young leaves. In this species the water-conducting tissues of the stem as well as the shoot apex and the young inflo­ rescence are effectively protected against the recurrent night frosts of its biotope. Height of the spe­ cimen 1. 7 m. The dissection had been done 6 days earlier, and the plant was now beginning to wilt. Kenya, Mt Kenya, Teleki Valley, 4200 m. Photo 0. Hedberg 11.8. 1948. starch being quite absent; and though inulin occurs and persistent leaf-bases which cover the stem it is always in small amounts in the stem. Pith doubtless also insulate these aqueous tissues and and cortex, in fact, appear to be aqueous tissues, protect them from the harmful effects of sudden and in view of the rapid fluctuations in humidity fluctuations in temperature" (loc. cit.). and temperature to which the high-altitude Sene­ On old specimens of Senecio keniodendron the cios are subjected, there can be little doubt that lower part of the dry leaf cylinder is usually shed these internal reservoirs of water play a part in because of the increasing girth of the stem (Figs. enabling the trees to maintain a due balance be­ 33, 83; cf. also R. E. FRIES, 1925, Pl. 40). The latter tween water uptake and water loss. The thick bark is then equipped with a thick cork cortex, which

Acta Phytogeogr. Suec. 49 52 Adaptive trends

species most of the leaves are patent in daytime (Fig. 39) but bend inwards towards the centre of the leaf rosette at night (Fig. 40). In S. brassica the lower side of the leaves is covered by a dense and woolly creamy-white indumentum, which makes its leaf rosettes quite conspicuous already at a long distance, particularly when they are in "night position" (Figs. 41, 81; cf. also R. E. FRIES, 1925, Pl. 39). This indumentum must certainly help to reduce the outward radiation from the rosette at night. The capacity of temperature insulation of such a rosette was tested by a couple of tempera­ ture measurements: on a cold morning a tempera­ ture of -4.0°0 was registered among the outermost leaves, whereas the interior of the closed rosette gave + 1.5°. The stem of Senecio brassica is woody in the same

Fig. 38. Senecio keniodendron: longitudinal section through way as in S . keniodendron, but unlike the latter it is the centre of a leaf rosette with a young inflorescence, prostrate (HEDBERG, 1957, pp. 226 and 358, Pl. protected by numerous densely folded young leaves. 12B). The dead leaves seem to remain as a rule (Close-up of the same specimen that is shown in Fig. 37.) attached to the stem, forming an insulating mantle Photo 0. Hedberg 5.8. 1948. of the same type as described for the latter. After it has produced a terminal inflorescence the primary seems to provide equally good temperature in­ leaf rosette dies with the inflorescence, but from sulation. In places that have been frequently some of its upper leaf axils arise innovation shoots visited by man, many specimens have been arti­ forming secondary rosettes in the same way as in ficially denuded of their dry leaves, which provide other Dendrosenecios (cf. Fig. 29). In this way older dry fuel in all types of weather (cf. HAUMAN, specimens may carry a number of leaf rosettes, 1933a, p. 607; 1935, p. 10). And where the vegeta­ below which the branched stem lies concealed (op. tion has been burned most specimens have lost cit., PI. 12B; cf. also Figs. 41 and 81 here). A closer the bulk of their protecting leaf cylinder, which loss investigation of this peculiar life-form is highly seems to be sometimes fatal (Fig. 22; cf. also p. 37 desirable. and Fig�. 18-19). The frost-free biotope provided by the dry leaf OTHER DENDROSENECIOS; COMPARISON WITH cylinders of the Giant Senecios is exploited by a ESPELETIA number of insects and spiders, some of which have A more detailed description of the structure of an­ been found only in this ecological niche (fide Dr. other Giant Senecio, S. friesiorum from Ruwenzori, A. Holm; cf. also ALLUAUD & JEANNEL, loc. cit., has been furnished by HAUMAN (1935, p. 14 f.), and SALT, 1954, p. 396). who investigated also the anatomy of stem and SENECIO BRASSICA leaves.l That species agrees very well with S. keniodendron The second high-level Dendrosenecio of Mt Kenya, in its dense leaf rosette and thick

Senecio brassica, has leaf rosettes of much the same 1 The anatomy of the three Kilimanjaro species was appearance as those described above. Also in that treated, in more detail, by HARE (1941).

Fig. 41. Senecio brassica vegetation on the well-watered lower slope of the valley side, and open Senecio keniodendron woodland on the slightly drier slopes above. Kenya, Mt Kenya, Buguret Valley, 4100 m. Photo 0. Hedberg 16.8.1948.

Acta Phytogeog1·. Suec. 49 Temperature insulation 53

Fig. 39. Leaf rosette of Senecio bmssica in "day position", Fig. 40. Leaf rosette of Senecio b1·assica in "night position", with unfolded leaves. Kenya, Mt Kenya, Teleki Valley, with the leaves densely folded together, exposing the thick 4000 m. Photo 0. Hedberg 25.7.1948, in the afternoon. light-coloured indumentum on their lower side. Kenya, Mt Kenya, Teleki Valley, 4200 m. Photo 0. Hedberg 2.8. 1948, at about 7.40 a.m.

Acta Phytogeog?'. s��ec. 49 54 Adaptive trends

Fig. 42. Open Senecio ba1·batipes woodland with Alchemilla elgonensis as field layer dominant. This Giant Senecio has no insulating cylinders of ma.rcescent dry leaves around its stems, but these are protected by an extremely well-developed cork cortex. The thiclmess of the latter can be deduced by comparison to the dead branches just to the left of the central part of the tree in the foreground, where the bark has fallen off. Uganda, Elgon, in the crater, 4100 m. Photo A. Holm, May 1948. insulating cylinder of dry leaves-cf. HAUMAN, (where the bark is yet thin) protected as a rule by op . cit., Pl. VI, and Fig. 37 above. The description a thick cylinder of dry leaves (cf. JEANNEL, 1950, given above for S. keniodendron does in fact apply p. 209). The sole exception is Senecio barbatipes on quite well also to the other high-level Dendrosene­ Elgon, which is provided with an extremely well­ cios. They all have large and compact leaf rosettes, developed cork cortex (Fig. 42; cf. JEANNEL, and thick and sparsely branched stems, covered loc. cit.). Those Dendrosenecios occurring at lower on their older parts by a thick cork cortex. All levels on the mountains, on the other hand (S. high-level species except one have their young stems erici-rosenii, S. amblyphyllus, S. battiscombei, and

Acta Phytogeog1·. Suec. 49 Temperature insulation 55

S. johnstonii) are as a rule narrow-stemmed with thin bark and caducous leaves (Fig. 43; cf. SYNGE, 1937, p. 76). These differences certainly support the idea that the thick stems with heavy bark and the cylinders of persistent dry leaves occurring in the afroalpine belt represent adaptations to the afro­ alpine climate (cf. HuMBERT, 1935, p. 835 f.; HARE, 1941). That contention is further supported by a comparison with the vegetation of the South American paramos. Here we find in an almost identical climate plants that superficially look almost exactly the same, namely the Espeletias (cf., e.g., GoEBEL, 1891, Pis. 10-ll; MEYER, 1900, p. 92; NEGER, 1913, p. 201-2; MILDBRAED, 1922b, p. 231; HEILBORN, 1925, p. 157; CuATRECASAS, 1934, Pls. XXII-XXV, XXX; 1958, Pls. XV, XVI,

Fig. 43. Senecio er·ici-1·osenii growing in a narrow ravine in the ericaceous belt. Note the loose leaf rosettes and the thin stems without marcescent leaves. Uganda, Kigezi Distr. , Muhavura, W slope, 3400 m. Photo 0. Hedberg 7.10.1948.

Fig. 44. Leaf rosette of Lobelia telekii in "day position". Fig. 45. Leaf rosette of Lobelia telekii in "night position", Most of the leaves are unfolded, but the youngest ones are with tightly folded leaves, photographed immediately after firmly pressed together around the shoot apex. Width of sunrise. On the sun-exposed (left-hand) side of the rosette rosette c. 0.5 m. Kenya, Mt Kenya, Teleki Valley, 4200 m. the leaves had already begun to unfold, although they Photo 0. Hedberg 2.8.1948. appeared quite :;:tiff when touched and were densely co­ vered by hoar frost along the edges; the air temperature was still a few degrees centigrade below freezing. Width of rosette c. 0.4 m. Kenya, Mt Kenya, Teleki Valley, 4200 m. Photo 0. Hedberg 11.8.1948.

Acta Phytogeogr. Suec. 49 56 Adaptive trends

Fig. 46 Flowering individual of Lobelia telekii. The flowers Fig. 4 7. Flowering individuel of Lobelia telekii in longitudinal are hidden between the long hairy bracts, where there is section, showing the arrangement of the flowers and bracts, prob<:Lbly an ameliorated microclimate. From the base of and the soft hollow stem. Height of specimen about 1.2 m. the flowering spike emerge (on the left hand side) two The surrounding vegetation is dominated by Alchemilla innovation shoots. Height of specimen c. 1.2 m. Kenya, elgonensis. Kenya, Elgon, E slope of Koitobbos, 4000 m. Mt Kenya, Teleki Valley, 4000 m. Photo 0. Hedberg Photo 0. Hedberg 10.5.1948. 26.7. Hl48.

XIX, and end plate; SMITH & KocH, 1935; Fos ­ Dendrosenecios of East Africa and the Espeletias BERG, 1944, several figures; TROLL, 1948, Fig. 4; of South America has been repeatedly stressed 1958, Fig. 55; 1959, Fig. 7; JEANNEL, 1950, p. 207; before, e.g., by MEYER (1900, p. 92), MrLDBRAED WEBER, 1956, Figs. 1-3 and 23; 1958, Pls. 97 and (l922a, p. 200), HEILBORN (1925, p. 157), HAUMAN 103-105). From a taxonomic point of view these (1935, pp. ll and 37), CoTTON (1944, p. 162), belong to a different genus, but ecologically they TROLL (1948, p. 58; 1958, p. 60; 1959, pp. 39, 47), appear to belong to exactly the same life-form, with and JEANNEL (1950, p. 210). a sparsely branched or unbranched thick woody stem, surrounded in its upper part by a cylinder of LOBELIA TELEKII insulating dry leaves, and crowned by a huge leaf Also the afroalpine Giant Lobelias present con­ rosette, the leaves of which at night are bent in­ spicuous examples of structures favouring tem­ wards towards the centre of the rosette (fide Dr. perature insulation. In its vegetative state a speci­ S. Nilsson, Uppsala; observations made in 1960 on men of Lobelia telekii, for instance, consists b the paramo de Mucuchies near Merida, Venezuela). main of a large and dense leaf rosette, crowning a The stem anatomy of Espeletia appears to be short erect stem. These rosettes behave in the s::tme l�rgely similar to that of Dendrosenecio (cf. WEBER, way as those of the Dendrosenecios: in daytime 1956). The ecological correspondence between the most of their leaves are more or less patent (Fig.

Acta Phytogeogr. Suec. 49 Temperature insulation 57

44) whereas at night they bend inwards and become firmly appressed to the central part of the rosette, forming a compact cabbage-like head (Fig. 45). Most nights the outermost leaves will be frozen stiff and covered by hoar-frost. Early one morning I measured at the surface of such a closed rosette a temperature of -3.5°C, whereas in its interior (though far outside the shoot apex) was obtained a value of + l.0°C. Also this kind of rosette must evidently possess a good capacity for temperature insulation. One peculiar fe::>.ture is the rapid un­ folding of the rosette in the morning: long before the first rays of suns�ine reached it the leaves on the east-exposed side started to bend outwards­ although the air temperature was well below freezing and the leaves were covered with hoar­ frost and appeared quite stiff. For that reason it

Fig. 48. Inflorescence of Lobelia wollastonii. The hairy bracts in this species are shorter than in Lobelia telekii, and the inflorescence is borne at the top of a leafy stem about 2-5 m high (cf. also Fig. 87). Height of inflorescence c. 0.8 m. Uganda, Ruwenzori, between Lake Bujuku and Stuhl­ mann Pass, 4000 m. Photo 0. Hedberg 23.3.1948.

Fig. 49. Leaf rosette of Lobelia keniensis in "day position" Fig. 50. Leaf rosette of Lobelia keniensis beginning to un­ with unfolded leaves. Note the water in the foreground­ fold from its "night position" shortly after sunrise. Al­ this species only occurs on wet ground. Width of rosette c. though the temperature was still well below freezing in the

0.5 m. Kenya, Mt Kenya, Teleki Valley, 4200 m. Photo surrounding air the leave.:; on the sun-exposed (left-hand} 0. Hedberg 2.8.1948. side were largely unfolded, less so on the shaded side. Diameter of rosette c. 0.4 m. Note the snow cover on the

ground. Kenya, Mt Kenya, Teleki Valley, 4200 m. Photo 0. Hedberg 2.8.1948, about 8 a.m. Cf. also the upper colour photo on the frontispiece.

Acta Phytogeogr. Suec. 49 58 Adaptive trends

Fig. 51. Leaf rosette of Lobelia keniensis in "day position" with patent leaves. Between the central leaves appears the surface of the "water reservoir" described in the text (p. 59)". Diameter of rosette c. 0.5 m. Kenya, Mt Kenya, Teleki Valley, 4200 m. Photo 0. Hed­ berg 8.8.1948. Cf. also the lower colour photo on the frontispiece. proved impossible to get a good photograph of an the de�1sely crowded bracts may be expected to entirely closed rosette-in Fig. 45 several leaves have effect a certain buffering of vigorous temperature already started to unfold. changes in the environment. After the fructifica­ After a vegetative stage of unknown length a tion the whole plant dies, but from its basal part is rosette of Lobelia telekii produces a large terminal often formed one or more innovation shoots (cf. jnflorescence (Fig. 46), in which the flowers are Fig. 46, and Fig. 84 below). A similar pubescent concealed among long, hairy bracts. The whole inflorescence occurs in Lobelia wollastonii, the plant is entirely herbaceous and contains large vicariad of L. telekii on the Virunga Volcanoes amounts of milky sap (Fig. 47). No temperature and Ruwenzori (Fig. 48). mo�:-c.s:uements were made in the inflorescence, but

Fig. 52. Leaf rosette of Lobelia keniensis in longitudinal Fig. 53. Leaf rosette of Lobelia keniensis in "day position", section. An intact rosette carries water up to the tip of the photographed early in the day. Two of the crescent-shaped central cone of young leaves. Width of the rosette c. 0.4 m. pieces of ice found floating in the water between the leaves Kenya, Mt Kenya, Teleki Valley, 4200 m. Photo 0. Hed­ have been placed on top of the rosette. Kenya, Mt Kenya, berg 5.8.1948. Teleki Valley, 4200 m. Ph6to 0. Hedberg 27.7.1948.

Acta Phytogeog1·. Suec. 49 Tempe-rature insulation 59

Fig. 54. Leaf rosette of Lobelia bequaertii in "day po�ition". In its centre appears the surface of the "water reservoir". Width of rosette c. 0.5 m. The shrub in the back­ ground is HelichTysum stuhl­ mannii. Uganda, Ruwenzori, near Bigo, 3400 m (ericaceous belt), at the margin of a Carex runssor-o­ ensis bog. Photo 0. Hedberg 21.3.1948.

LOBELIA KENIENSIS i�1tcrior of the rosette, perhaps from the fleshy leaf bases (Fig. 52). It should be noted that Lobelia The second alpine Giant Lobelia of Mt Kenya, L. keniensis, as well as its geographical vicariads on keniensis, also displays in its vegetative state a huge some of the other mountains (the Lobelia deckenii leaf rosette, behaving at night in the same way as group; cf. HEDBERG, 1957, p. 333) occurs only on those of L. telekii. A full-grown rosette in daytime moist or wet ground (cf. Fig. 49). The lower parts of position is shown in Fig. 49, whereas Fig. 50 shows the rosette leaves are densely imbricate, forming a another one about to expand from its night position series of partly communicating vessels. Towards the -on its sun-exposed side several leaves have al­ central part of the rosette the vmtery fluid is some ready bent out·wards. The changing appearance of 10 cm deep, completely surrounding the shoot apex these rosettes is also sho·wn in the colour photo­ and the young leaf primordia. Every cold night the graphs of the frontispiece. These compact rosettes topmost layer of water will freeze, forming a series may in themselves be expected to provide good of crescent-shaped ice-pieces up to 0.5 cm (or temperature insulation for their inner parts at night. sometimes 1 cm?) thick between the leaves (Fig. But in this plant the protection is made even more 53). But below this surface layer the fluid will never efficient by another specialization. Looking at an freeze, and the most vulnerable parts of the rosette open rosette of Lobelia keniensis from above (Fig. are effectively protected against frost. Since water 51) we find its interior filled with a watery liquid, has its greatest density at + 4 °0 the lower part of the €nclosed between its imbricate leaf bases and form­ rosette with the shoot apex will presumably be kept ing, as it were, a small composite water basin. This a couple of degrees centigrade above freezing. The cannot simply be rain water, because it occurs only temperature reading obtained by me for the quite regularly in rosettes of this kind even after water a few cm below the ice gave + 1.0°0 (early periods of dry weather. Consequently the fluid in the morning, before the air temperature in the must in some way have been secreted from the surroundings had reached above freezing).

Acta Phytogeog1·. S'uec. 49 60 Adaptive trends

cf. also SYNGE, 1937, p. 203). That the rosettes of Lobelia bequaertii on Ruwenzori are filled with water may have attracted little attention, since they occur in an almost permanently wet environment, but also those rosettes seem to follow the same pat­ tern (Fig. 54; cf. SYNGE, 1937, PI. 20). In Lobelia keniensis the vegetative part of the stem is often longer than in L. telekii. On its lower part the dead leaves remain attached to the stem and form an insulating mantle of dry leaves of the same type as in Senecio keniodendron (cf. HEDBERG, 1957, PI. 6A; R. E. FRIES, 1925, Pl. 38). The same thing applies to other m emhers of the Lobelia deckenii group, as L. bequaertii on Ruwenzori (Fig. 98; cf. CHIOVENDA & CoRTESI, 1909, Pis. 41 & 42-sub nom. L. deckenii; FISHLOCK & HAN­ cocK, 1933, Pis. C:1 & G; HAUMAN, 1933b, p. 20 &. PI. VII, lower fig.; SYNGE, 1937, PI. 19); to L. elgonensis on Elgon (Fig. 55); and to L. deckenii itself on Kilimanjaro (cf. ToBLER-WOLFF & ToB­

Fig. 55. Flowering specimen of Lobelia elgonensis. Note the LER, 1914, Pis. 9 & 10). A similar insulating mantle cylinder of marcescent dry leaves around the lower part is found more or less well-developed also in Lobelia of the stem. Kenya, Elgon, E slope of Koitobbos, 4000 tn". wollastonii (cf. CHIOVENDA & CoRTESI, op . cit., Photo 0. Hedberg 10.5.1948. Pl. 44; FISHLOCK & HANCOCK, op . cit., Pl. I:2; HAUMAN, op. cit., Pls. I & II; SYNGE, op . cit., PI . 62; DE GRUNNE et al., 1937, Figs. 38-39)- The peculiar water-holding rosettes of Lobelia In its earliest stages the young inflorescence of keniensis and its vicariads on other mountains (cf. Lobelia keniensis will presumably profit from the HEDBERG, 1957, pp. 187 f. and 333) are rather temperature insulation of the "rosette reservoir". conspicuous-any explorer happening to kick a As it grows taller it will soon reach above the water rosette of this kind will, for. instance, get his leg level, but for some time yet it may profit from the thoroughly soaked. A few scattered references to protection of the uppermost stem leaves, which them occur in the literature. Thus BRYK (1927) bend inwards towards it every night (Figs. 56 and classified the young Lobelia ( elgonensis) shoots on 57). The densely spaced bracts themselves will Elgon as "Insektenfallen", because drowned in­ probably also effect a certain amount of tempera­ sects were often found in the fluid accumulated ture insulation, although they are shorter than in among their leaf bases; he believed the fluid to be Lobelia telekii and glabrous (Fig. 58). The latter rain water. HANCOCK & SouNDY (1931, p. 172) assumption may derive some support from an oc­ reported about Lobelia elgonensis that "The young casional observation: A specimen of L. keniensis plants were full of a viscid liquid in which pieces of carrying a young inflorescence that had been cut in ice were floating". Also EDWARDS & TAYLOR half (cf. HEDBERG, 1957, Pl. 6A) proved next (1935, pp. 181-184) noted that the young plants of morning to have become frozen stiff during the that species "hold a quantity of watery fluid be­ night, whereas the intact specimens in the vicinity tween the bases of their leaves"-a picture of such were soft and unfrozen. The freezing may of course a rosette is shown in their Fig. 17. The latter authors have been due in part to the results of the wounding found the fluid to be very viscous, being "evidently itself, and not only to the decreased facilities for largely a secretion of the plant" (op. cit., p. 183; temperature insulation. The members of the

Acta Phytogeog1·. Suec. 49 Temperature ins�tlation 61

Fig. 56. Lobelia keniensis. Young flowering individual in Fig . 57. Lobelia lceniensis. Young flowering plant in "day position", with patent leaves. Height of specimen c. "night position", with the leaves tightly folded around the

{).8 m. Kenya, Mt Kenya, Teleki Valley, 4200 m. Photo base of the inflorescence. Height of specimen c. 0.7 m. 0. Hedberg 2.8.1948. Kenya, Mt Kenya, Teleki Valley, 4200 m. Photo 0. Hedberg 11.8.1948.

Lobelia deckenii group are all entirely herbaceous, CARDUUS KENIENSIS and seem to be regularly propagated by innovation At least one more genus is represented in the afro­ shoots from the base of the stem as well as by seeds alpine flora by a species possessing a very large (HEDBERG, 1957, PI. 6A). The marginal part of and dense leaf rosette, viz. Carduus, with the species their rosettes are often inhabited by some Cur­ C. keniensis s.lat. (Fig. 59; cf. HEDBERG, 1957, p. culionidae (Xestorhinus spp.; cf. BRYK, 1927). 249 and Pl. 11). Its leaves are fiercely spinescent. Also the Giant Lobelias have their physiognomic No temperature measurements were made inside -counterparts in the vegetation of the South Ameri- such rosettes, but presumably they do effect good -can paramo and puna, namely in the genera Lu- temperature insulation for the shoot apex. Carduus pinus (L. alopecuroides etc.) and Puya (P. raimondii keniensis is notable also for its compact agglomera­ etc.); cf., e.g., HEILBORN (1925, p. 157 and Fig. 3), tion of spinescent capitula, often measuring more CuATRECASAS (1934, Pls. XXXI and XXXII-1; than 10 cm across (Fig. 59; cf. R. E. FRIES, 1923, 1958, Pls. XIV-1, XVII, and XX-1), CoTTON Figs. 1-3; HEDBERG, op . cit., Pl. 11). Closely related {1944, p. 160), FosBERG (1944, p. 231), TROLL (1948, Carduus species occurring at lower levels on the p. 59 and Fig. 8; 1952a, p. 40; 1958, p. 61 & Figs. mountains have less compact leaf rosettes and less 58-61; 1959, p. 47), and WEBER (1958, Figs. 31-34 & closely aggregated capitula (R. E. FRIES, op . cit.). 98). The denseness of both in C. keniensis must pre-

Acta Phytogeogr. Suec. 49 62 Adaptive trends

nicht besonders hervorgehoben werden" (op. cit., P- 18). HAUMAN (1933 a, p. 715) suggested that some of the morphological peculiarities discussed above might be interpreted as adaptations favouring temperature insulation: "Quelques-unes des mor­ phoses enumerees au paragraphe precedente peu­ vent, cependant, s'interpreter comme des protec­ tions contre les variations brusques de tempera­ ture: ainsi les feuilles mortes persistantes au long des troncs, les subers epais ou feuilletes; mais on peut aussi bien ne voir la, comme dans certaines xeromorphoses, que des accidents anatomiques accompagnant les qualites sine qua non a la vie sur ces sommets-" (cf. also HAUMAN, 1935, p. 11; HuMBERT, 1935, p. 840; and JEANNEL, 1950, p. 210). HARE (1941, p. 365) expected the thick bark and persistent leaf bases of the Giant Senecios to Fig. 58. Lobelia keniensis. Longitudinal section through insulate the aqueous storage tissues of the stem and young inflorescence, showing the sheltered position of the ''protect them from the harmful effects of sudden young flower buds between the bracts, and the hollow fluctuations in temperature" (cf. also p. 94 below). herbaceous stem. Kenya, Mt Kenya, Teleki Valley, 4200 m. Photo 0. Hedberg 5.8. 1948. CoTTON (1944) dealt with the "megaphytic habit» of Giant Senecios and Giant Lobelias in some detail. He believed the origin of "megaphytism" sumably represent an adaptation favouring tem­ in these plants to be connected with the intensity of perature insulation. soJar radiation at high level in the tropics (op. cit., CONCLUSION ON ��GIANT ROSETTE PLANTS" p. 166). SALT (1954, p. 412) stated about the dead The striking appearance of the Giant Senecios and leaves around the trunk in Senecio cottonii and the Giant Lobelias, as well as of their physiognomic large bracts hiding the flowers and peduncle in counterparts in South America, the Espeletias, Lobelia deckenii that: "These envelopes protect Puyas, etc., have often attracted the interest of the main stem from direct radiation at the same travellers and botanists. The ecological significance time as they produce an air cushion which can of their huge specialized leaf rosettes, dry leaf absorb and retain heat and ameliorate the changes of cylinders or thick bark, etc., has never been fully temperature". appreciated, however (cf., e.g., DIELS, 1934, p. 59 f.; The only suggestion I have found that the huge CoTTON, 1944, p. 166; JEANNEL, 1950, p. 210-211). leaf rosettes of Giant Senecios, Giant Lobelias, and Below I shall attempt to summarize the sugges­ Espeletias provide an ameliorated microclimate for tions earlier advanced in this connection. the shoot apex was made by DIELS (1934, p. 68): Already 70 years ago GoEBEL (1891, p. 17) re­ "Die Knospen liegen bei ihnen umhiillt und ge­ corded that (because of their pubescence) the leaves borgen hinter dicken Lagen grosser Blatter, ge­ of Espeletia and Culcitium provide good tempera­ wissermassen in einem selbstgeschaffenen System ture insulation, being therefore used "Zum Zu­ hoherer Warme ...". Without special reference to decken gegen die erstarrende Kalte der Paramos''. the giant rosette plants concerned here, DAUBEN­ As regards the importance of this feature for the MIRE (1947, p. 186) stated that "With massive plants themselves he only noted: "Die Bedeutung organs of plants, and others covered with hair or dieser Einrichtung fiir die Transpirationsvermin­ cork, the temperature changes of the inner tissues derung, den Schutz gegen raschen Temperatur­ lag behind those of the environment, the lag often wechsel, austrocknende Wirkung der Winde, brauch being sufficient to prevent injury from short pe-

Acta Phytogeogr. Suec. 49 Temperature insulation 63

Fig. 59. Oarduus keniensis. Fullgrown leaf rosette, c. 0.6 m wide, with emerging inflorescence. These dense and extremely spinescent rosettes presumably provide good protection for their central portions not only against grazing animals but also against violent changes of temperature. Uganda, Elgon, in the crater near Maji ya Moto, 3700 m. Photo 0. Hedberg 20.5. 1948. riods of extreme temperatures". The first-and growing shoot-apex gets on better when protected only-reference I have found to the nightly move� against recurrent freezing and thawing needs no ments of the leaves in the giant rosette plants comment. And "frozen shoots of course conduct was made by MACKINDER (1900, p. 483): "The no water" (DAUBENMIRE, op . cit., p. 205). A frozen Lobelias closed their heads of leaves at night�time shoot of relevant dimensions would certainly just like daisies." require a few hours for thawing almost every The observations recorded above indicate that a morning-in those morning hours when insolation number of the morphological peculiarities of the is usually strong and transpiration intense. The "giant rosette plants", notably their huge leaf ecological significance of frost protection to the rosettes forming "night buds", their cylinders of stem is therefore also easily visualized (cf. also p. marcescent leaves, or thick bark, and their very 93 below). compact inflorescences, afford good temperature The importance of good temperature insulation insulation for the most vulnerable parts of the for the Giant Senecios and Giant Lobelias may be plant, and for its water-conducting tissues. That a accentuated by the fact that most of those plants

Acta Phytogeogr. Suec. 49 Adaptive trends require large amounts of soil water. Especially on perience of the afroalpine (or paramo, etc.) climate. the drier mountains (with more violent temperature Anybody who has camped for a few weeks in their changes) they will therefore occur mainly in loca­ biotope, where the air temperature inside the tent lities with good water supply, as in valley bot­ in the morning falls regularly somewhere between toms, along streams, etc. Because of cold air -5°C and 0°, will be prepared to appreciate the drainage at night such localities are subject to importance of any structures providing temperature heavier night frosts than the adjacent slopes (cf. insulation-be it sleeping bags or mantles of dry p. 24 above), and hence these more or less hygrophy­ Dendrosenecio leaves. tic plants will have to withstand more severe frosts With regard to the facilities for temperature than the more xerophytic flora on drier slopes. insulation-as in many other respects-the afro­ The reason why almost all earlier botanists alpine flora evidently forms a marked contrast to dealing with these floras have-in my opinion­ the alpine and Arctic floras of lower latitudes, underestimated the ecological importance of struc­ where, according to S0RENSEN (1941 p. 14) . .. "no tures favouring temperature insulation in afro­ external morphological features capable of pro­ alpine plants (or those of paramo and puna) may tecting the living tissue against the cold is known to perhaps be that few botanists have personal ex- us."

FROST HEAVI NG RESISTANCE

One of the greatest hazards for plant life in much rule seem to be absent or insignificant, evidently -of the afroalpine belt must lie in the actions of because of the insulating capacity of the vegetation, ·solifluction, notably by frost heaving. As described which creates an ameliorated microclimate of its above (p. 33) this factor causes vigorous and fre­ own (cf. pp. 23 and 25 above). In spite of root com­ quent movements in the surface layer of bare soil petition etc. seedlings must therefore stand a much .at high level, especially in the upper part of the better chance of survival and establishment in .alpine belt. Such phenomena are particularly closed vegetation than on open soil. As a conse­ .apparent on Mts Kenya, Kilimanjaro, and Elgon quence of this the vegetation in the upper part of (loc. cit.). Frost heaving seems to be particularly the alpine belt on Mt Kenya and Kilimanjaro (at detrimental to seedlings. The large surfaces of open least) is definitely patchy-dense grass tussocks or ·soil occurring at this level might otherwise appear Helichrysum bushes, etc., with associated smaller to be favourable for the germination of seeds, but plants, alternate with areas of bare ground (Figs . . all seedlings occurring on them are evidently 60, 61). At the margin of each vegetation segment rapidly eliminated by frost heaving (cf. Figs. 11- there seems to be something of a dynamic equili­ 13). The needle-ice formation may also penetrate brium between growth and erosion. into the margin of a grass sod (etc.) and cause a Because of the short duration of each night frost, gradual erosion (cf. TROLL, 1944, pp. 577, 587, 591, freezing does not as a rule penetrate deeper than .and Fig. 7). The powerful rhizomes and root systems about 3-5 cm (cf. p. 31 above). Consequently larger -of many afroalpine herbs (as Ranunculus oreophytus, stones are not affected by solifluction but remain Haplosciadium abyssinicum, Haplocarpha rueppel­ as calm "islands" in the restless surface of the lii, Nannoseris schimperi, and Carex monostachya) "mobilidesertum" (cf. TROLL, op. cit., p. 606). may represent adaptations favouring resistance to This implies that a thin soil layer adjacent to each frost heaving. such stone is less affected by solifluction than the Under closed vegetation solifluction would as a open soil-possibly the greater heat conductibility

Acta Phytogeogr. Suec. 49 Frost heaving resistance 65

Fig. 60. Patchy vegetation with denee bushes of Helichrysum citrispinum var. hoehnelii, mixed with stray specimens of ·Oar-duus lceniensis (in the lower left corner), Senecio meyeri-johannis (at the front of the big bush in the foreground), ..Festuca tussocks, etc., alternating with large areae of open soil with traces of solifluction polygons. The height of the bush in the foreground is c. 0.3 m. The light-coloured strip across the picture is a path. Tanganyika, Kilimanjaro, S slGpe -of the Saddle between Kibo and Mawenzi, 4250 m. Photo 0. Hedberg 23.6.1948.

-of stone as compared to soil may also contribute in the Saddle of Kilimanjaro is largely concentrated to this. Hence the immediate surroundings of to such stone "edafids" (Fig. 61; cf. VoLKENS, larger stones and rocks form refugia where oc­ 1897, p. 316; SALT, 1954, pp. 395, 402, and Photo 9). ·casional seedlings may become established. And Similar conditions obtain in part of the alpine ·once a grass tussock, Helichrysurn shrub, etc., belt on Mt Kenya. has gained a foothold it creates to some extent a In spite of the difficulties imposed by soli­ local climate of its own, enabling the establishment fluction some plants manage to subsist even on soil of other plants in the way outlined above. The exposed to regular needle-ice formation. One :sparse vegetation (as well as the invertebrate life) example is afforded by Subularia rnonticola, which

.5-641464 Hedberg Acta Phytogeogr. Suec. 49 66 Adaptive trends

Fig. 61. Old specimen of Haplo­ sciadium abyssinicum with richly branched caudex, forming a cush­ ion with an ameliorated micro­ climate. In this cushion a few small grass tussocks have gained a foothold; others grow close to the stones in the background, which also provide refuge from the vig­ orous solifluction on the surround­ ing bare ground. Note the soli­ fluction polygons on the open soil. Tanganyika, Kilimanjaro, in the

S part of the Saddle between Kibo � and Mawenzi, 4400 m. Photo 0. Hedberg 25.6.1948. forms open communities along the edge of small stagnalis, can also a void the perils of solifluction by streams and lakes. At least in the upper part of growing below the water surface in tarns and streams, the alpine belt on Mt Kenya such localities are where the nightly ice is rarely thicker than one or subjected to regular needle-ice formation. But two cm. because the soil in those localities is loamy and Another remarkable way of withstanding the permanently moist the ice-needles are formed on effects of solifluction is to ride on top of it, without the surface only, leaving the soil intact. Further­ any fixed attachment to the soil. This way of life more, for some reason ice-needle formation does is found among mosses, lichens, and algae. Thus in not occur within a mm or two of the bases of the several localities on Mt Kenya I encountered Subularia plants, and consequently the latter are loose-lying moss balls with diameters between l and left intact in small cavities in the ice (Fig. 62). 7 cm (Figs. 30, 63, 64). As a rule these were found This species, like Limosella africana and Oallitriche on open soil with more or less distinct signs of solifluction. The central part of each ball was found on dissection to contain a core of soil, from which the moss stems radiated (Fig. 65). Presumably

the balls had originated by erosion, the margin of a moss carpet being broken up by needle-ice forma­ tion. Afterwards the pieces had been lying on top of the restless soil surface, being lifted by needle­

ice at night, to subside again in daytime in a slightly different position. By these perpetual mo­ Fig. 62. Diagram illustrating the mode of needle-ice for­ vements they were kept shifting around, so that mation in a Subularia monticola-community on wet the moss could grow about equally well on all sides . loamy soil in the alpine belt. The ice-needles (B) are formed In this way some of them had become almost per­ on top of the soil (0), lifting only a few coarse plant frag­ fectly globular (Fig. 63), while others, which had ments etc. (A) on their top, and leaving a little tunnel around each Subularia rosette (cf. text above). Kenya, Mt happened to turn the same side upwards for a Kenya, Teleki Valley, 4200 m_ longer period, were flattened with weaker moss

Acta Phytogeogr. Suec. 49 Frost heaving resistance 67

Fig. 63. Loose-lying spherical moss ball on active solifluction ground. Kenya, Mt Kenya, Teleki Valley, 4200 m. Photo 0. Hedberg 7.8. 1948 (from colour photograph).

Fig. 64. Heap of loose-lying moss balls on active solifluction soil. Each ball is about l-3 cm in dia­ meter. Kenya, Mt Kenya, on the ridge S of Hohnel Valley, 4250 m. Photo 0. Hedberg 14.8.1948.

Fig. 65. Loose-lying moss ball cut in half to show the central core of soil, from which moss stems radiate in all directions. Locality: Kenya, Elgon, on the summit of the highest peak on the SW part of the crater rim, 4300 m, on open solifluction soil, Hedberg lOll, UPS. Photo U.-B. Sahlstrom 1962. Magnification x 2.

Acta Phytogeogr.Suec. 49 68 Adaptive trends

growth on the lower side. Only on balls of the latter type were sporophytes to be discovered. Such moss balls were found in several localities on Mt Kenya and-though less frequently-on Kilimanjaro and Elgon. Their greater frequency

on Mt Kenya than on Kilimanjaro may be related to the moister climate of the former mountain, frost heaving occurring only when the soil surface is moist (cf. p. 33 above). Elgon, on the other hand, is barely high enough to get strong solifluction. The moss balls collected by me were formed by Grimmia campestris BuRCH. (No. lOll from Elgon) and G. ovata W. & M. (Nos. 1711, 1753, and 1909 from Mt Kenya, and presumably No. 1342 from Kilimanjaro). Ball-forming specimens of the latter species were named by PO'l'IER DE LA VARDE (1955, p. 147) "forma glacialis". Similar loose-lying moss balls, formed by species of A ndreaea, Dicranoweisia, Grimmia, Kiaeria, and Rhacomitrium, have been described by LID (1938) from different localities in the alpine belt in Nor-

I

Fig. 67. Loose-lying cake of Nostoc commune (VAUCH.) BoRN. & FLAH. (det. H. Skuja), about 2 dm across, on small terrace of moist solifluc­ tion soil along a streamlet. Kenya, Mt Kenya, Teleki Valley, 4200 m. Photo 0. Hedberg 11.8. 1948.

Acta Phytogeogr. S1rcc. 49 Frost heaving resistance 69

Fig. 68. Two specimens of Hap­ losciadium abyssinicum on small terrace of active solifluction soil. Tanganyika, Kilimanjaro, S slope above Peter's Hut, c. 4100 m. Photo 0. Hedberg 23.6.1948.

way, and from Jan Mayen. Those were also provided (1939, p. CXXII, Fig. 2). The lichens described by with a central core of soil (op. cit., Fig. 2) and him were classified by MATTICK (1951, p. 405) as showed about the same range of size. They were "aerolichenes" or "Wanderflechten". believed by LID to have been formed through rolling A third example of loose-lying (macroscopic) by the wind, but since they were found on bare soil cryptogams is afforded by an alga, N ostoc commune in areas subjected to strong solifluction I consider (VAUCH.) BoRN. & FLAH. (det. H. Skuja). Large it more probable that those balls were formed by cakes of this species, often more than a dm across, the action of solifluction in the way suggested were found on more or less moist solifluction soil above. Some other examples of "unattached mos­ at about 4200 m altitude on Mt Kenya (Fig. 67). ses" have been related by MARTIN (1952) from They were evidently lifted by needle-ice solifluction Great Britain and New Zealand. The latter cases almost every night. The life-forms of the crypto­ must evidently have a different explanation, gams mentioned here are further discussed on p. however. 92 below. The same mode of life has been adopted by at The only vascular plant, seedlings of which ap­ least one fruticous lichen, a Parmelia species near pear to be able occasionally to become established P. vagans (Fig. 66; det. R. Santesson), collected by on open solifluction soil, is Haplosciadium abys­ me on Mt Kenya and on Kilimanjaro (Nos. 1705 sinicum. That species is equipped with a thick and 1354, resp.). The thalli were lying quite loose and deep tap-root, and if a seed of it germinates on the soil surface. Probably the same species is during a period without pronounced solifluction its shown on Fig. 38 in TROLL 1944, where the thalli root may evidently sometimes manage to pene­ appear to have been sorted out by solifluction to trate below the layer of mobile surface soil and get the edges of the polygons. Also TROLL found them rooted below it. Once it has become firmly an­ to be quite unattached (op. cit., p. 562). Similar chored it may evidently withstand solifluction observations of "licheni vaganti" along the edges quite well (Fig. 68). By branching of the caudex of polygons on solifluction soil were also made at older specimens may form small mats which modify high level in northern Ethiopia by PICHI-SERMOLLI the microclimate at the soil surface, reducing soli-

Acta Phytogeog1'. S�tec. 49 70 Adaptive trends

Fig. 69. Festuca pilgeri grassland on sloping ground, where the small s::>il terraces dammed by the graf's tussocks are covered by mats of Alchemilla johnstonii (cf. text, p. 70). Kenya, Mt Kenya, Teleki Valley, 4200 m. Photo 0. Hed­ berg 7.8.1948. fluction so that also seedlings of other plants may prevented by needle-ice solifluction, but the per­ become established (Fig. 61; note the strip of soil petual heaving and sinking will not damage the without frost-pattern adjacent to the Haploscia­ stems. From the initial refugium the roots can dium colony). easily spread below the restless top layer of the A more indirect way of colonizing open soli­ soil, and by continuous growth and branching the fluction soil is practised by Alchemilla johnstonii. stems may sometimes manage to form a dense mat Seedlings of that species seem to be at high level on across a solifluction terrace (Fig. 69). Such a mat Mt Kenya confined to the edge of grass tussocks will evidently decrease outward radiation at night or other closed vegetation. Once the plant has be­ so much as to decrease or inhibit freezing of the come established its long decumbent or prostrate soil surface at night and hence solifluction. In stems may easily spread over the solifluction soil. this way the spot may be colonized by other plants Formation of adventitious roots will be effectively as well.

XEROMO RPHY

A classification of land plants according to their supply." Such structures that appear to enable water relationships into hydrophytes, rnesophytes, plants to restrict transpiration are commonly called and xerophytes, as suggested by WARMING (1895) xeromorphic characters-e.g. small, leathery leaves has been widely used by plant ecologists. Plants with much sclerenchyme; a thick cuticle; small and belonging to each of these groups display one or numerous stomata. The stomata are often sunken, more morphological, anatomical, and physiological or hidden behind a thick pubescence or by the peculiarities believed to adapt them to wet, in­ leaves being folded or rolled (cf., e.g., DAUBENMIRE, termediate, and dry conditions, resp. Xerophytes 1947, p. 155; WALTER, 1950, p. 239; STOCKER, are, according to WARMING (1909, p. 101) "Those 1956,p. 728).Earlier itwas supposed that xerophytes plants that are adapted to meet the conditions of would transpire much less than mesophytes, but it strongest transpiration and most precarious water- is now realized that when water is freely available

Acta PhytogeogT. S�t ec. 49 Xeromorphy 71 they do in fact often transpire faster than the latter they attain to greater altitude on the mountain (cf. ScHIMPER-v. FABER, 1935, p. 12; WALTER, side ...''. A good example of diminishing leaf size 1950, pp. 247, 250). When the water becomes towards increasing altitude is provided by the scarce, however, xerophytes are able to reduce hydrophytic plant Orassula granvikii-cf. HED­ drastically their transpiration, hence being capable BERG, 1957, p. 280 & Fig. 24; 1959, p. 162 & Fig. l. of surviving severe drought (op. cit.). Similar tendencies were noted by HAUMAN (l933a It was early recognized that plants which display p. 714) in the afroalpine flora of Ruwenzori. It xeromorphic features are by no means restricted would appear, therefore, that the intensity of the to dry localities-they occur also, for instance, in environment factors that cause xeromorphy in­ bogs and in Arctic and high mountain vegetation creases towards higher level. (cf., e.g., GoEBEL, 1891, p. 16 etc.; WARMING, Also in the case of xeromorphy we find parallels 1895, pp. 147, 181 etc.; ScHIMPER, 1898, p. ll; to the afroalpine flora in the paramos of South NEGER, 1913, p. 195 f.; S0RENSEN, 1941, p. 13 etc.). America. Numerous xerophytic features were thus That xeromorphic habit is predominant in the described from the vegetation of Venezolan para­ afroalpine flora of Kilimanjaro is easy to under­ mos by GoEBEL (1891, pp. ll, 17, etc.), who as­ stand, considering the desert-like conditions pre­ cribed them to the difficulty for the roots to ab­ vailing in much of the alpine belt on that moun­ sorb water from the permanently cold soil. The tain (cf. p. 13 above; VoLKENS, 1897, p. 322; most important organizations that reduce trans­ and SALT, 1954, p. 412, etc.). But even in the almost piration in the Oompositae of this paramo flora permanently wet vegetation on Ruwenzori xero­ were stated to be: (1) dense, woolly pubescence; phytic features are very prominent in the alpine belt, (2) leathery leaves; (3) folded or rolled leaves; (4) as emphasized by HAUMAN (1933 a, p. 714): "Un small leaf surface; (5) condensed growth (cushions); autre point fondamentale est l'evidente xero­ (6) absence of aerial stems. All of these features are morphie, en ce climat si humide, d'un grand nombre in fact prominent also in the afroalpine flora; the

d'especes : a signaler les feuilles lineaires enroulees first four of them will be exemplified in this sec­ et plus ou moins couvertes de poils d'Helichrysum tion, while the remaining two will be treated in the Stuhlmannii et de sa variete latifolium surtout, next (p. 73). As regards other xeromorphic features d'un type tout a fait mediterraneen, les feuilles such as root development, density and size of filiformes ou rigides de nombreuses Graminees, stomata, thickness of epidermis, etc., no observa­ celles assez charnues, a cuticule superieure epaisse tions were made by the author. et a poils plus ou moins abondants sur la face l. Dense woolly pubescence. A thick indumentum

inferieure des Senecio arborescents ..." may evidently serve not only to reduce transpira­ The amount of xeromorphy occurring in the tion but also to protect the plant tissues against afroalpine flora a pp ears to increase towards higher strong insolation (cf. p. 42). Numerous species levels. In Helichrysum stuhlmannii on Ruwenzori with such dense pubescence occur in the afroalpine there is, for instance, a gradual transition from lax flora; some of the most important are listed on p. specimens with patent and almost flat leaves 43. A few examples are shown in Figs. 28 (Heli­ without pubescence on the upper side, occurring chrysum newii), 39-41 (Senecio brassica), and 91 in the ericaceous belt, to robust bushes with more (Helichrysum citrispinum var.). or less erect and revolute leaves covered by a dense 2. Thick leathery leaves occur, for instance, in all pubescence, occurring in the upper part of the afroalpine species of Dendrosenecio and Lobelia­ alpine belt (HEDBERG, 1957, p. 344). Similar cf. Figs. 29 (Senecio adn·ivalis), 34-37 (Senecio trends may be found in the pubescence of the keniodendron), and 49-53 (Lobelia keniensis). Also Giant Senecios (cf. p. 42 above). For the Kili­ in most other afroalpines with comparatively large manjaro species of Dendrosenecio it was concluded leaves the latter are thick and more or less coria­ by HARE (1941, p. 369) "that they fall naturally ceous, e.g. in Carduus keniensis (Fig. 59), and into a series exhibiting increasing xeromorphy as Haplocarpha rueppellii (Fig. 77).

Acta Phytogeogr. Suec. 49 72 Adaptive trends

Fig. 70. Euryops dacrydioides, top of flowering shrub. Note the scale­ like xeromorphic leaves. Tangany­ ika, Kilimanjaro, S slope of the Saddle above Peter's Hut, 4200 m. Photo 0. Hedberg 15.6.1948.

3. Folded or convolute leaves, or at least revolute some Senecio species (S. snowdenii, S. telekii, etc.), leaf margins, are very common in the afroalpine in Bartsia lc ilimandscharica and B. macrocalyx, etc. flora. Several grasses have filiform leaves, notably 4. Examples of small leaf surface are also nu­ in the genera Agrostis (e.g. A. gracilifolia, A. merous in the afroalpine flora, coinciding in many leptophylla, A. trachyphylla) and Festuca (e.g. F. cases with revolute leaf margins (Ericaceae; abyssinica, F. kilimanjarica, F. pilgeri)-cf. Figs. Helichrysum spp., etc.) or with a dense pubescence 13, 22, 69. The peculiar subulate, furrowed leaves in (Helichrysum citrispinum etc.; Alchemilla argy­ Romulea (R. keniensis etc.) should also be brought rophylla and related species, cf. Fig. 94). Small to this group. Strongly revolute leaf margins occur scale-like leaves occur, for instance, in Euryops in all afroalpine Ericaceae and in several species dacrydioides (Fig. 70). An example of diminishing of Helichrysum (H. amblyphyllum, H. brownei, leaf size with increasing altitude in Orassula H. citrispinum-Fig. 91, and H. newii-Fig. 28, granvikii was quoted above p. 71. Small leaves etc.). Slightly revolute leaf margins occur in some are evidently most common in frutescent species, more species of Helichrysum (H. cymosum ssp. forming more or less dense shrubs (cf. Figs. 28, fruticosum, H. chionoides, H. stuhlmannii, etc.), in 70, 91, and 94).

Acta Phytogeog·r. S11ec. 4D 73

REDUCED LENGTH OF INT ERNODES

One of the most striking features in the afroalpine The "acaulescent rosette plants" mentioned flora is an emphatic tendency towards reduction of above all have unbranched or sparsely branched the stem-notably the length of the internodes stems. A similar inhibition of internode elongation (cf. HEDBERG, 1957, p. 273). This tendency is may occur also in plants with more or less richly manifested not only in the Giant Senecios and Giant branched stems. One example is rendered by Lobelias described above but also in numerous Haplocarpha rueppellii, which under certain con­ herbs of more conventional size. In several species ditions may become copiously branched, forming the overground part of the plant consists of a big cushions (Fig. 77). Similar fairly dense cushions dense leaf rosette only, in the centre of which are formed by Swertia subnivalis, the flowers of eventually appears a sessile (or almost sessile) which are also almost sessile. Large and very dense flower or inflorescence. The best examples of such cushions are formed by Agrostis sclerophylla "acaulescent rosette plants" are Ranunculus cryp­ (Fig. 78) and Sagina afroalpina (Fig. 79); smaller tanthus (cf. HEDBERG, 1957, Plate 5A) , Oreophyton . ones are formed by Myosotis keniensis (Fig. 80). falcatum (op. cit., Plate 4A), Haplosciadium abys­ These five are the only cushion plants observed in sinicum (op. cit., Plate 4B; cf. Figs. 61 and 75 the afroalpine flora (cf. p. 88). Physiognomically here), Wahlenbergia pusilla (op. cit., p. 185, Fig. such plants play a minor role in this flora, although 11), Oarduus platyphyllus and 0. nanus (cf. R. E. they may become locally dominant in moist places FRIES, 1923, Taf. 2), and Nannoseris schimperi (Fig. 81). (Figs. 71, 73). In all these the acaulescent habit The ecological background for the widespread appears to be genetically fixed so that it remains tendency towards reduced internode length in high even if the plants occur occasionally at lower mountain plants is not yet completely understood. level, or are kept in a greenhouse (cf. Figs. 73, 75). That such dwarfing is common in Arctic and high In other cases the dwarf habit appears to be more mountain plants has long been known, as testified, easily modified by changes in the environment, e.g., by ScHIMPER (1898, p. 740 etc.), ScHROTER since specimens from lower levels or those kept in a (1908, p. 643), WARMING (1909, p. 253), NEGER greenhouse develop appreciably longer scapes, e.g. (1913, p. 37 etc.), RAUH (1939, p. 480 etc.), and in Ranunculus oreophytus (cf. HEDBERG, 1957, p. DAUBENMIRE (1943, p. 333). From the paramos of 271 regarding caulescent specimens, however), northern South America numerous examples are Oonyza subscaposa (Figs. 72, 74; cf. also RoBYNS, known of "acaulescent rosette plants" and cu­ 1947, Plate XLIV), and Haplocarpha rueppellii shion plants (cf. GoEBEL, 1891, pp. 36, 44; ScHIM­ (cf. HEDBERG, 1957, Plate 5B). PER, 1898, pp. 742-3, 781; HEILBORN, 1925, p. The same tendency towards reduction of the 157; DIELS, 1937, p. 29; etc.), and in the drier puna stem is also visible in the afroalpine race of Arabi­ further south in the Andes their importance seems dopsis thaliana which even under greenhouse con­ to be even larger (TROLL , 1958, p. 68 etc.). ditions is low and copiously branched from the It was shown long ago by BoNNIER (1890, etc.) base (Fig. 76). In some other afroalpines the stems that ramets of lowland plants transplanted at are prostrate and the peduncles much reduced, high level in the mountains become much smaller making the flowers appear sessile at the nodes, e.g. than those maintained in their ordinary milieu, in Trifolium acaule and T. elgonense (cf. GILLETT, with shortening of leaves and internodes, etc. (cf. 1952, p. 392, Fig. 4:9). In Ootula cryptocephala the also ScHROTER, 1908, p. 628). Some of the con­ vegetative stems are prostrate and the peduncles clusions reached by BoNNIER may evidently be much reduced (cf. HEDBERG, 1957, Fig. 16 p. 221). unreliable (cf. TURRILL, 1940, p. 55), but similar

Acta Phytogeogr. Suec. 49 74 Adaptive trends

Fig. 71. Nannoseris schimperi. Flowering specimen, collec­ Fig. 72. Conyza subscaposa. Flowering specimen collected ted in nature. Drawn after Hedberg 1183 (Tanganyika, in nature. Drawn after Hedberg 1774 (Kenya, Mt Kenya, Kilimanjaro, S part of Saddle between Kibo and Mawenzi, Teleki Valley, 4400 m; UPS). Del. M. Induss. 4400 m; UPS). About natural size. Del. M. Induss. results have been obtained by several other (SAUBERER & H.ARTEL, loc. cit.) and ultraviolet botanists, including CLAUSEN, KECK & HIESEY (cf. DAUBENMIRE, 1947, p. 223; WEBER, 1958, (1940, p. 312, etc.). That high-level specimens p. 174). But there are evidently considerable tend as a rule to be smaller and more dwarfed than differences between different plant species in their low-level ones of the same species is a well-known reaction to light of various quality (SAUBERER fact, illustrated for instance by VAN STEENIS & H.ARTEL, loc. cit.), and WEBERBAUER (1931, p. (1948, p. XXVII), and WEBER (1958, p. 174). 33) considered that: "die Gestaltung der Hoch­ Similar observations were recorded above (p. 73) gebirgspflanzen weit weniger durch die Licht­ for Ranunculus oreophytus, Conyza subscaposa, verhaJtnisse bestimmt werden, als man gewohnlich and Haplocarpha rueppellii in the afroalpine flora. annimmt.'' The dense growth of some cushion plants may also The low temperatures, notably at night, are also occasionally become looser under unusual en­ believed to retard shoot elongation (cf. ScHIMPER, vironment conditions (cf. ScHROTER, 1908, p. 1898, p. 750; SCHROTER, 1908, p. 643). In some 644; HAuRI & ScHROTER, 1914, p. 623; Du RIETZ, cases such mountain plants that tend to lose their 1931, p. 70; RAUH, 1939, p. 481). There can be no dwarfed habit when transplanted in the lowlands doubt therefore that the weak internode elongation may be kept low by exposing them to low tempe­ of high mountain plants in some cases may re­ rature during the night (ScHROTER, loc. cit.). For present largely modifications induced by extreme mountain plants at low latitudes these low night environment conditions (cf. ScHROTER, 1908, p. temperatures recur throughout the year (cf. p. 18 644; RAUH, 1939, p. 481). above). The importance of nightly frosts for the As regards the physiological background for this origin of cushion growth has been particularly modificatory dwarfing numerous suggestions have emphasized by TROLL (1948, p. 56; 1958, p. 71, etc.). been made. Strong light is known to have a re­ A third factor that may promote dwarf habit is tarding effect on shoot elongation (cf. ScHIMPER, the lack of water (cf. WARMING, 1909, p. 130; 1898, p. 749; 8CHROTER, 1908, p. 643; 8AUBERER & DAUBENMIRE, 1947, p. 143). H.ARTEL, 1959, p. 232, etc.), especially blue light Obviously none of these explanations may be

Acta Phytogeogr. Suec. 49 Reduced length of internodes 75

Fig. 73. Nannoseris schimperi. Specimen grown in a green­ Fig. 74. Conyza subscaposa. Specimen grown in a green­ house in Uppsala, maintaining under greenhouse condi­ house in Uppsala. Under greenhouse conditions this plant tions its acaulescent habit. Ori gin of collection: Kenya, developed a fairly long scape, whereas herbarium speci­ Mt Kenya, Teleki Valley, 4200 m, 4.8.1958, G. Nelson s.n. mens from the same locality were acaulescent. Origin of Photo 0. Hedberg 1959. The markings below the name on collection: Kenya, Mt Kenya, Teleki Valley, 4200 m, 4.8. the label are 1 cm apart. 1958, G. Nelson s.n. Photo 0. Hedberg 1959.

expected to cover all cases of "induced nanism" 1961). In these species the stunted habit appears to in Arctic and high mountain plants-each case be genetically fixed so that it remains under all may call for its own explanation, and much more environment conditions tested for the plants.! experimental work is required. Discussions of To explain how this feature came to be genetically these matters are found, for instance, in ScHROTER fixed is certainly not an easy matter (cf. ScHIMPER, (1908, p. 643) and RAUH (1939, p. 476). 1898, p. 751; ScHROTER, 1908, p. 644; RAUH, 1939, It has been shown above that the dwarfed habit p. 482). One conclusion lies close at hand, how­ of some high mountain plants may be modified ever. Since reduced internode length (usually as­ under ameliorated environment conditions. But on sociated with rosette formation, cushion growth, the other hand it is equally evident that many such etc.) is such a common feature in Arctic and high plants retain their stunted growth also under low­ mountain plants and has become genetically fixed land conditions-even when they are kept in a in many species of diverse relationships, this or­ greenhouse (cf. ScHROTER, 1908, p. 644; RAUH, ganization must evidently confer some selective 1939, p. 481; DAUBENMIRE, 1943, p. 333). This advantage in their milieu. appears to be the case with most cushion plants As regards the nature of this selective advantage (RAUH, loc. cit.). Some examples from the afroalpine opinions vary (cf. especially RAUH, 1939, p. 476). flora have been given above (p. 73; Figs. 73 and Thus it has been pointed out that sessile rosette 75); to these may be added Sagina afroalpina, plants and cushion plants may be favoured by the which has remained low and compact also under greenhouse conditions in Uppsala. To this group belong some of the most exclusive afroalpines, such 1 The same is probably true for the high level Giant Senecios and Gi�n t Lo belias-thusLobelia keniensis has been as the endemic and monotypic genera Oreophyton, kept in cultivation in a greenhouse for four years at Upp­ Haplosciadium, and Nannoseris, which appear to sala without losing its compact rosette growth with ex­ represent an old orophyte element (cf. HEDBERG, tremely short internodes.

Acta Phytogeogr. Suec. 49 76 Adaptive trends

Fig. 75. Haplosciadiurn abyssinicurn. Specimen grown in a greenhouse in Uppsala, maintaining under greenhouse condi­ tions its acaulescent habit (the partial umbels emerge from the centre of the leaf rosette). Cf. Fig. 68. Origin of collection� Kenya, Mt Kenya, Teleki Valley, 1959, M.Coe s.n.

temperature conditions at the soil surface, where 1939, p. 476). But HEILBORN (1925, p. 163) and _the temperature in daytime is usually a few WEBERBAUER (1931, p. 32) have objected that the degrees higher than in the air above (ScHROTER, high cushions of the Andean vegetation are more 1908, p. 647; WARMING, 1909, p. 130; NEGER, 1913, exposed to the wind than the smaller plants p. 35; WEBERBAUER, 1930, p. 346-7). But as growing among them. Other objections are given pointed out above (p. 23) temperature conditions by RAUH, lac. cit. As regards the afroalpine flora at the soil surface at night are by no means parti­ the scarcity of high winds in the afroalpine belt cularly favourable in the afroalpine belt, so this speaks against this explanation. explanation appears hardly convincing. Possibly Furthermore it has been maintained that cu­ the foliage of acaulescent rosette plants and shion growth may be an adaptation to meet cushion plants may become less cooled by outward conditions of excessive drought (WARMING, 1909,. radiation at night than the bare ground (cf. p. 23), p. 38; ScHROTER, 1908, p. 646; cf. also RAUH� and hence these plants may help to ameliorate 1939, p. 477). This can hardly apply to the afro­ their own microclimates (cf. p. 25 above). In this alpine cushion plants and sessile rosette plants, way they might perhaps become less exposed to the however, since they occur mainly in moist or wet nightly frosts than less compact plants. localities. It has also been suggested that cushion shape or Evidently none of the explanations mentioned acaulescent growth may afford some protection above is entirely satisfactory. The author would against the wind (ScHIMPER, 1898, p. 751; WARMING, therefore like to suggest another explanation, 1909, pp. 38, 130; NEGER, 1913, p. 39; EsPINOSA, which, it must be stressed, refers only to such 1932, p. 195; etc.; for further references see RAUH, acaulescent rosette plants and cushion plants that

Acta Phytogeogr. Sttec. 49 Reduced length of internodes 77 occur at alpine levels on high mountains in the Tropics, or in ecologically equivalent habitats (cf. 'TROLL, 1948). Let us start from the observation that both the acaulescent rosette p]ants and the cushion plants discussed above display several xeromorphic fea­ tures-as evidenced for the former by WARMING (1909, p. 130) and HEILBORN (1925, p. 157; cf. also GoEBEL, 1891, p. 44), and for the latter by GOEBEL (1891, p. 40), SCHROTER (1908, p. 575), WARMING (lac. cit.), HAURI & ScHROTER (1914, p. ·622), 8KOTTSBERG (1916, p. 136), HEILBORN (1925, p. 163), and EsPINOSA (1932, p. 193). Since several species of cushion plants occur in moist or wet localities this xerophytic habit has appeared peculiar to some authors; thus HEILBORN (1925, p. 162) declared: "It seems absurd to regard the -cushion shape as a xerophilous adaptation in order to store water, when the latter actually grow in Fig. 76. Arabidopsis thaliana. Grown from seed in a green­ water." About the "xerophilous" characteristics house in Uppsala. Under greenhouse conditions this strain observed in cushion plants he remarked: "These kept its peculiar tufted habit. Origin of collection: Kenya, characteristics have, however, probably been Mt Kenya, Teleki Valley, 4000 m, Coe & Karika 29(} developed earlier in dry stations, the cushion- (UPS). Photo 0. Hedberg 1959.

Fig. 77. Dense cushion of Haplo­ carpha rueppellii, about 8 dm wide, growing in a trickle of water. Kenya, Mt Kenya, Buguret Val­ ley, 4100 m. Photo 0. Hedberg 16.8.1948.

Acta Phytogeogr. Suec. 49 78 Adaptive trends

part of the afroalpine belt ground frost occurs on most nights of the year, so that at high level the soil surface is frozen down to a depth of a fevv cm during most of the night and part of the morning. Furthermore, according to KRAMER (1956a, p. 150): "Low temperature decreases the absorption of water in a number of different ways. The vis­ cosity of water itself is twice as great at zero as at 25°C and this alone reduces the rate of water move­ ment through the membranes about 50 %. The viscosity of the protoplasm must also be greatly increased, further decreasing the permeability of living tissue. The result is a decrease of water move­ ment through the roots at 5°C to about 25 % of the rate at 25°C under the same gradient." (Cf. also GoEBEL, 1891, p. 11; VoLKENS, 1897, p. 322; ScHIMPER, 1898, p. 750; HAUMAN, 1933a, p. 714; ScHIMPER-v. FABER, 1935, p. 8; TRoLL, 1959, p. 46.) In the afroalpine belt the temperature of the soil and of unprotected plant tissues will presumably fall well below +5°C during at least the night and the early part of the day, and consequently water absorption and conduction will then be even more retarded. When we recall that at the same time the transpiration is favoured in this biotope by intense insolation, low air pressure, and rapidly increasing air temperature in the morning (cf. pp. 8, 15, and 18 above) it may not be difficult to

Fig. 78. AgTostis scleTophylla, part of a dense cushion, visualize the selective advantage of a xerophytic collected in nature (Hedberg 1724, UPS; Kenya, Mt habit even on moist ground. Kenya, Teleki Valley, 4350 m) . Photo U.-B. Sahlstrom. Let us then return to the reduced length of the Magnification x 2. internodes occurring in cushion plants and acaules­ cent rosette plants. It was shown above that both water absorption by the roots and water movement through the plant is much retarded by low tem­ shape being of secondary origin .'' Similar views peratures-a frozen stem will of course conduct no were expounded by WEBERBAUER (1931, p. 32). water at all. Any thin and unprotected stem pro­ These two authors have evidently both drawn truding above the ground may become frozen the conclusion that when cushion plants occur on during the night in this environment, and hence the moist soil the water must be easily available to particular habits of acaulescent rosette plants them. That does not necessarily follow, however. (with a very short stem, more or less buried in the Another interesting observation concerning cu­ ground) and cushion plants (where the stems are shion plants is th8Jt (at least in the high mountains protected below the surface of the cushion) must of the Tropics and in Subantarctic countries) they be expected to confer a decided selective advantage occur mainly in regions with recurrent night frosts in this biotope-in the same way as the insulating all the year round (TROLL, 1948, p. 56; 1958, p. 71). cylinders of dry leaves around the stems of the Thus it was shown above (p. 18) that in the major afroalpine Dendrosenecios (cf. p. 50 above).

Acta Phytogeog1·. Suec. 49 Reduced length of internodes 79

Fig. 79. Big tussock of Sagina afroalpina in longitudinal section, showing central soil core. On moist soil with locally in­ tense solifluction along a streamlet. Kenya, Mt Kenya, Teleki Valley, 4200 m. Photo 0. Hedberg 27.7. 1948.

Fig. 80. Cushion of .Myosotis

keniensis at the side of a small soli­ fluction terrace in open Alchemilla johnstonii community. Kenya, Mt Kenya, Teleki Valley, 4250 m. Photo 0. Hedberg 7.8.1948.

Acta Phytogeogr. S�teo. 49 Ill. Afroalpine life-forms

INT RODUCTION

In the preceding paragraphs a number of peculiar It must be remembered, of course, that not all of morphological features, interpreted as adaptations the morphological pecularities observed in the flora to the vicissitudes of the afroalpine environment of an extreme milieu need to have developed in were described. Some of the plants displaying these response to conditions resembling the actual ones­ features may be classified into special life-forms, some may just have been inherited from ancestors -such as the giant rosette plants. Others present who acquired them under entirely different con­ -only general adaptive trends, recognizable in plants ditions. Several afroalpines do in fact occur also in -of otherwise diverse life-forms, for instance the areas outside Tropical Africa, and seem to thrive dense pubescence on the leaves of many species. A equally well in widely different environments ( c£. large number of afroalpine vascular plants display HEDBERG, 1961, p. 916). But a critical scrutiny of no visible morphological adaptation to the afroalpine the afroalpine flora reveals a number of such milieu, however, but look much the same as they­ striking convergent adaptations, paralleled only ·Or their close relatives-do in areas with entirely in the flora at high level on other tropical moun­ -different types of climate, e.g. Arabis alpina and tains, that their connection with an equatorial­ Subularia monticola. All members of the afroalpine alpine climate appears entirely convincing. flora must evidently be to some extent physiologi­ For reasons explained below (p. 91) it is hardly .cally adapted to this environment, since they do practicable to distribute all afroalpine vascular thrive here. Nature finds many different ways of sol­ plants to definite life-forms. Before entering upon ving an ecological problem. And consequently "Die this matter we should, however, devote some at­ Mannigfaltigkeit der Organbildung ist ... grosser tention to the meaning of the term life-form, and .als die M<:mnigfaltigkeit der Lebensbedingungen" to some of the life-form systems proposed in the 'GOEBEL, quot:)d from STOCKER, 1956, p. 730). past.

CURRENT LIFE-FO RM SYSTElVIS

Since long ago botanists have attempted to classify HuMBOLDT (1806, 1808; cf. Du RIETZ, 1931). plant species not only after similarities in their To HuMBOLDT these physiognomic types ("Haupt­ reproductive parts and assumed relationship into formen") were mainly tools for the description of a "natural system" but also according to similari­ vegetation. Physiognomic life-form systems of ties in general physiognomy, or in form and func­ a similar type were used by many later authors tion of their vegetative parts, into "life-forms" such as KERNER (1863), GRISEBACH (1872), and -or "growth-forms". The first elaborate system of HULT (1881). After the upheaval in biological such "physiognomic types" is ascribed to A. v . thinking caused by the penetration of the evolu-

.Acta Phytogeogr. Suec. 49 Current life-form systems 81

Fig. 81. Vegetation along a small stream in the alpine belt, largely dominated by cushion plants (Sagina afroalpina, Agrostis sclerophylla, and 11 1yosotis keniensis), together with Lobelia lceniensis, Senecio bmssica, Festuca pilgeri, Agt·ostis trachyphylla, etc. Kenya, Mt Kenya, Head of Teleki Valley, 4200 m. Photo 0. Hedberg 27.7.1948.-This community reminds of the "Hartpolstermoor" described by TROLL (1958, p. 69, and Abb. 65; 1959, p. 53, and Bild 16, etc.) from the South American puna. tionary theories of DARWIN (1859) and his con­ adaptation of the plant to survive the unfavourable temporaries, there developed a tendency to eva­ season, having special regard to the protection of luate physiognomic characters according to their the surviving buds or shoot-apices" (RAUNKIJER, "biological significance", that is, their supposed 1934, p. 112), and has been widely used for statisti­ relationship to the environment. The first system of cal comparison of different floras and their relation this type is due to WARMING (1884), who based his to the prevailing climates; it has also recently been classification on the duration of life, the power employed in an attempt to assess the relative age of vegetative reproduction, the duration of the of various plant groups (M0LHOLM-HANSEN, 1956). various shoots, the position of the shoots in relation RAUNKIJER's system and its statistical application to the soil surface, the duration of the leaves, etc. have been criticized, int. al., by WARMING (1908), A less complicated system, which soon became very SKOTTSBERG (1912, 1913, and 1914), BEWS (1925, popular, was created by RAUNKIJER (1907, 1908, p. 14), NORDHAGEN (1928, p. 594 f.), Du RIETZ 1934, etc.). His system was based upon "the (1931, p. 34 f.), ALLAN (1938), TROLL (1956, p.

6-641464 Hedberg Acta Phytogeog1-. St£ec. 49 82 Afroalpine life·forms

752; 1958, pp. 7 and 20); and PETTERSSON (1958). of the systems outlined by Du RIETZ. But even Its applicability to the afroalpine flora is further with such a set of definite life-form systems it will discussed below on p. 91. be very difficult to cover all the different types of Revised or new life-form systems have later been vegetation. Such difficulties become very obvious supplied by, e.g., DRUDE (1913, 1914), LINDMAN when one attempts to classify the afroalpine flora (1914), WARMING (in WARMING-GRAEBNER, 1914- into definite life-forms. Because of the peculiar 1918; 1923), GAMS (1918), CLEMENTS (1920), rhythm of the tropical high-mountain climate there Du RIETZ (1921), LINKOLA (1922), BRAUN-BLAN­ are no thermal seasons, and at high level the dif­ QUET (1928), and SKOTTSBERG (1929). In most of ferences between rainy and dry seasons are not as these systems the main emphasis was again laid a rule so pronounced as in the lowlands. Thus in at on the general physiognomy of the plants. A least the majority of afroalpine vascular plants comprehensive account of these and earlier systems there are no definite flowering seasons (HEDBERG, has been provided by Du RIETZ (1931). Mter his 1957, pp. 11 and 21; cf. TROLL, 1948, p. 47). The historical review of "Life-form Theories and Life­ "unfavourable season" is therefore neither winter form Systems in Botanical Literature" Du RIETZ nor dry season but the night. This evidently in­ (op. cit., p. 42) concludes that " ... there is still very validates here three of the six life-form systems little agreement between ecologists regarding the outlined by Du RIETZ (Nos 3, 4, and 5 in the above general theory of life-forms as well as their prac­ list), all of which refer to the appearance of the tical delimitation and classification". Because of plant during different seasons. Of the three re­ the many conflicting life-form systems in existence maining sets numhers 1 and 2 would seem to be he infers that "the only way out of these difficul­ most useful for our purpose. But even the classi­ ties appears to be that of admitting not one single fication according to "set No 1" into simple life-form system but several parallel ones, based physiognomic types (Du RIETZ, op . cit., p. 45) offers upon different points of view ..." (op. cit., p. 43). several difficulties. Thus trees are said to be In the following (p. 44) he enumerates six different "Holoxyles with a distinct main trunk remaining classes of life-forms, namely: (1) "Main life-forms unbranched in its lower part". To this group must ... based upon the general physiognomy of the evidently be brought not only such plants as plant during the height of their annual vegetation Senecio keniodendron with stout and erect woody period." (2) "Growth-forms . .. based primarily upon stem but also S. brassica, which has a plagiotropic shoot architecture". (3) "Periodicity life-forms . .. woody stem, in older specimens branching sym­ based primarily on the vegetative periodicity of the podially in the same way as S. keniodendron (cf. plant, i.e. the difference between their physiognomy p. 52 above, and HEDBERG, 1957, p. 226 and Pl. in the different seasons, and the seasonal distribu­ 12B). Both physiognomically and ecologically these tion of their vegetation- and resting periods." plants differ, however, so much from trees of (4) "Bud-height life-forms ... based upon the height ordinary type as to make their assignment to the above (or below) the ground-level of the uppermost same life-form quite unsatisfactory (cf. p. 86 buds perduring the most unfavourable seasons" below) . The distinction between shrubs (or dwarf­ (i.e., life-forms in the sense of RAUNKIAER). shrubs) and half-shrubs appears difficult to main­

(5) "Bud-typ e life-forms . .. based entirely upon the tain in the afroalpine flora. In half-shrubs is, structure of the buds perduring the most un­ according to Du RIETZ, loc. cit., "only the lower favourable season." (6) "Leaf life-forms ... based part of the ( epigeous) stem lignified and perennial, upon the character "(form, size, duration, structure, the upper part annual and herbaceous". Now, etc.) of the leaves." in a climate like this, it is difficult if not impossible In this way different life-form systems may be to distinguish between annual and perennial­ chosen according to the area studied and the pur­ even the distinction between monocarpic (hapax­ pose of the study. It will not as a rule be neces­ anthic) and polycarpic (pollacanthic) is often im­ sary-or desirable-to use more than one or two possible to make (cf. WEBERBAUER, 1905, p. 81).

Acta Phytogeogr. Suec. 49 Current life-form systems 83

Many afroalpine species are frutescent, but the p. 55). Only a few of these species (viz., Lobelia younger shoots are, of course, as a rule herbaceous giberroa, L. lanuriensis, and L. bambuseti), which (cf., e.g., HEDBERG, 1957, Pl. 7B), though they get possess distinct secondary xylem, can, however, woody with age-irrespective of seasons. Even the possibly be classified as "trees". All the other spe­ distinction between herbs and dwarf-shrubs (or cies, including the Lobelia deckenii group, L. dwarf half-shrubs) may be quite obscure. Thus in telekii, L. wollastonii, L. aberdarica, etc., are en­ Veronica glandulosa young specimens may be tirely herbaceous without noticeable secondary entirely herbaceous, with terminal inflorescence. wood (cf. HAUMAN, 1933b, Pl. II and VII; HED­ After the first period of flowering the top shoot BERG, 1957, Pl. 6A). Biologically, however, all grows through the inflorescence (as in Callistemon), Giant Lobelias from high level seem to belong to the developing into a vegetative shoot, which in due same general "growth form" in spite of the dif­ time produces at its apex a new inflorescence. In ferences in stem anatomy (cf. p. 84 below). the meantime the basal part of the stem becomes In some instances, at least, Du RrETZ seems to increasingly frutescent. A young specimen of this have pushed the subdivision of his types a little too species flowering for the first time would no far, as in the distinction between" Aeroxylic semi­ doubt be classified as a "low herbaceous plant", prostrate sedentary dwarf-shrubs" (op. cit., p. 61) but an older one might sooner be regarded as a and "Semi-sedentary (creeping) dwarf-shrubs" (op. dwarf half-shrub-although it does not fit the cit., p. 65). The distinction between these groups definition given by Du RrETZ, loc. cit. I have pro­ lies in the occurrence of adventitious roots in the visionally denoted plants of this type "suffrutes­ latter. Theoretically, this difference may perhaps cent herbs". Similar cases of "through-grown in­ be of great significance, but in practice it must florescences" occur also, int. al., in most afroalpine often depend upon the age of the actual specimen species of Bartsia, which because of their distinctly and the nature of its habitat whether an inherent woody (and brittle) stem, unbranched in the lower capacity of forming adventitious roots is realized part, would be classified in Du RIETZ's system as or not. Calluna vulgaris will certainly not be the dwarf trees or pygmy trees. The absence of a clear only species pertaining to both groups (op. cit., demarcation between herbs and shrubs in Central pp. 61, 65). The strict application of subdivisions Africa has been testified before by WHITE (1962, of this kind must meet with many difficulties. p. XI). On the whole, it appears to be rather a hopeless Such difficulties of classification also become task to create a manageable system of growth­ evident when one attempts to apply the available forms covering in detail all the various types of part of the more elaborate life-form system No 2 plants on the earth, corresponding to a multitude of Du RIETZ, or "Growth-forms on the Basis of of combinations of different thermic and hygric Stem-types and Stem-type Combinations" (op. climates and a variety of edaphic conditions (cf. cit., p. 52 f.). Since unfortunately only part of this IVERSEN, 1936, p. 9; TROLL, 1955; 1958, p. 9). system has been published, covering the holoxyles There will always be a number of border-line cases­ and part of the hemixyles, it is difficult to judge its not even the distinction between deciduous and potential usefulness and impossible to use it con­ evergreen trees is always sharp (HOLTTUM, 1953, sistently. Two small corrections are required as p. 165). BEws (1925, p. 83) also testifies: "There regards the Giant Lobelias of the East African can be no definite lines drawn between different mountains. Firstly, all high-level species of this classes of growth forms any more than between group are unbranched in their upper parts, in­ many genera and families. It is indeed obviously cluding Lobelia bambuseti, quoted under "Branched much more difficult to do so in the former case rosette-trees" on p. 56 (reference: "R. E. and Th. than in the latter." Furthermore, no single bo­ C. E. Fries 1922b fig. 4"). Secondly, all the un­ tanist can hope to master the vegetation of all parts branched Giant Lobelias were classified as "Simple of the globe. And if he could, the resulting system (unbranched) rosette trees" (Du RrETZ, op cit., would be bound to comprise such a large number of

Acta Phytogeog1·. Suec. 49 84 Afroalpine life-forms

subdivisions, exceptions, and intermediate cases as TANSLEY & CHIPP (1926, p. 20) considered that to become all but unmanageable. The complicated "the independent student of evolution will do well, system proposed by GrnzBERGER (1939) may serve however, to make his own classification of life­ as an example (cf. TROLL, 1958, p. 12). I therefore forms of the communities he actually studies". agree with SKOTTSBERG ( 1929, p. 565) that ''it That none of the current life-form systems is ap­ might be more convenient to put off the making of plicable to the flora of the tropics has been testi­ such all-comprising classifications and to start fied before by TROLL (1956, p. 752). with separate systems for separate regions, to use, In the remaining part of this chapter short wherever possible, well-established and generally descriptions will be given of those life-forms that accepted terms which are understood far and wide, are believed to be particularly adapted to the afro­ and to make special types for special needs, as alpine climate, and to similar climates in other few as possible and as many as necessary". Also parts of the world.

GIANT ROSETTE PLANTS

To this life-form belong in the first place the Giant stage by the leaf rosette, and later by numerous Senecios and Giant Lobelias of the afroalpine belt; leafy bracts, which in some species are densely another representative is Carduus keniensis (s.lat.). pubescent. Some specific examples are given above on pp. 49- From an ecological point of view this life-form 64, where especially the temperature-insulating may be regarded as a product of particular adapta­ structures are described. tions to the peculiar afroalpine . (tropical-alpine) The giant rosette plants are characterized by climate with nightly frosts and rapid temperature thick and unbranched or sparsely branched stems fluctuations. The huge leaf rosettes folding up at which (in exposed situations at least) are usually night do evidently provide excellent temperature covered either by a dense mantle of marcescent insulation for their central part, so that the shoot dry leaves or (in older Dendrosenecio stems) by a apex and the young leaves and inflorescences are thick and cork-like bark. The vegetative part of effectively protected against frost. The mantles of the stem may be erect and unbranched, as in the marcescent leaves and the cork-like bark furnish Giant Lobelias1 and young specimens of most equally good temperature insulation for the stems, Giant Senecios; erect and branched, as in older enabling the water transport through the latter to specimens of most Giant Senecios; procumbent, as proceed without interruption. Some of the giant in (Dendra- )Senecio brassica and S. brassiciformis; rosette plants display further ecological specializa­ or extremely short, as in Carduus keniensis. Each tions along the same lines, such as the water storage stem (branch) is terminated by a huge dense leaf tissues in the Giant Senecios, buffering variations rosette, which folds together at night, forming a in transpiration intensity (cf. p. 50 above); the compact giant "night bud", and unfolds in the heat-retaining water-basins in the Giant Lobelias morning. Its centre contains a dense kernel of of the L. deckenii group (cf. p. 59 above); the young leaves and leaf primordia, protecting the densely clustered capitula in Carduus keniensis shoot apex against extreme temperatures. The (Fig. 59; cf. HEDBERG, 1957, Pl . 11); and the dense inflorescences are (in the East African represen­ pubescence of the lower side of the leaves in, for tatives) terminal, being protected in the young instance, Senecio brassica, S. barbatipes, and S. friesiorum. 1 Some of the Giant Lobelias, at least, have innovation shoots from the base, however, cf. Figs. 46, 84, and 104, It is particularly noteworthy that what appears and HEDBERG, 1957, Plate 6A. to be exactly the same life-form has arisen in other

Acta Phytogeogr·. Suec. 49 Giant rosette plants 85 parts of the world with a similar type of climate­ buschgeholze" (DRUDE, 1913, p. 47 etc.-these in­ though in different plant genera. A remarkable cluded also, for instance, Teophrasta, Carica, and ecological parallel to the afroalpine Giant Senecios Aralia). is found in (some of) the Espeletia species in the The term "Megaphyte", introduced by CoTTON Andean paramos of northern South America. (1944, pp. 158, 161) is also rather a wide concept, These have the same huge "night-bud-forming" including ... "certain perennials of non-arborescent leaf rosettes and the same insulating cylinders of genera (i.e. not true trees) which develop a 'lar­ marcescent leaves around the stem (cf. p. 55 above, geness of habit' (often in contrast to other members with references). Also the remarkable Pringlea anti­ of the genus) and exhibited in the vegetative por­ scorbutica from Kerguelen Island appears to match tion of the plant by unusual bulk or by exceptional very well the afroalpine Giant Senecios in its height. The plants in question are characterized life-form (TROLL, 1948, p. 59, and Abb. 7). by the possession of one, or at most only a limited Ecological parallels to the afroalpine Giant number of, active growing points, and are, there­ Lobelias seem to be furnished by the Andean genus fore, unbranched or only very slightly branched, Puya, and by some species of Lupinus (cf. p. 61 and their peculiar habit is definitely related to the above), as well as by the famous "Silver Sword" few outlets available for expansion" (loc. cit.). of Hawaii, Argyroxiphium sandwicense (cf. TROLL, Apart from the Giant Senecios and Giant Lobelias 1958, p. 61, and Abb. 63). To what extent these the M egaphytes comprise also such plants as Aloe, last-mentioned plants agree also in forming "night Agave, many Palms, Cacti, etc. Their 'largeness of buds" remains to be established. habit' must evidently have arisen in a number of As regards their life-form the giant rosette plants, quite diverse environments, however, and in spite as defined here, have been variously treated in the of some common morphological features these plants past. In the famous life-form system of RAUNKIJER must be considered to form from an ecological ( 1907, 1934) most of them would be classified as point of view rather a heterogeneous assemblage. phanerophytes, notably herbaceous phanerophytes Ecological viewpoints concerning the giant rosette (most Giant Lobelias), and evergreen micro- and plants were brought to the fore by TROLL (1948, nano-phanerophytes without bud-scales (some Giant 1950, 1952a, 1958). The Giant Senecios and Lo­

Lobelias, most Giant Senecios). Senecio brassica belias were denoted by him ( 1948, p. 58) as '' Stamm­ and S. brassiciformis would rather pertain to the Schopfblattgewachse", and he stressed the ecological chamaephytes, however, whereas Carduus keniensis connections between this life-form and the peculiar would be a hemicryptophyte. But as demonstrated climate pertaining to the alpine belt of Equatorial above the most important thing in this cas� is not high mountains and to the subantarctic islands the height above the ground of the surviving (op. cit.). Those Giant Lobelias which have densely shoot apex but rather the type of leaf rosette and pubescent bracts were called '' Wollkerzenkrauter'' of water economy (cf. also p. 93 below), and there­ (TROLL, 1952a, p. 40) or "Wollkerzengewachse" fore the life-form system of RAUNKIJER cannot be (TROLL, 1958, p. 60). Such Giant Senecios, Es­ applied with profit to these plants. peletias, etc., that have densely pubescent leaves In the "growth-form system" of Du RIETZ were denoted by TROLL (1958, p. 60) "Woll­ (1931, p. 52), "based primarily on shoot architec­ schopfgewachse". As demonstrated above, how­ ture", the Giant Lobelias and Giant Senecios were ever, the pubescence possessed by the last-men­ brought to the "Rosette-trees" (op. cit., p. 55; cf. tioned types is but onea of the many adaptations also p. 83 above). Other denominations used for favouring temperature insulation (radiation in­ (most of) the relevant plants are "Schopfbaume" sulation, etc.), and it seems hardly desirable to (DRUDE, 1913, p. 25, etc.; cf. also TROLL, 1958, p. subdivide the group of the giant rosette plants 61) or "Tuft-trees" (WARMING, 1909, p. 10-these after occurrence or non-occurrence of dense pu­ included, among others, tree ferns, palms, cycads, bescence. and arborescent Liliaceae), and "Dikotyle Feder- From the above discussion it should be evident

Acta PhytogeogT. Suec. 49 86 Afroalpine life-forms

that in spite of their woody stem the Giant Sene­ be homologized with the "tree limit" or "timber cios show very little resemblance to "trees" in the line" of Boreal countries. On some of the East ordinary sense-they represent a life-form pe­ Mrican mountains the Giant Senecios do in fact culiar to high levels on equatorial mountains. Their reach the upper limit of phanerogamic plants upper altitudinal limit can therefore by no means (HEDBERG, 1951, p. 164).

TUSSOCK GRASSES (AND SE DGES)

To this life-form belong the majority of the afro­ grasses are distinctly xeromorphic, usually den­ alpine grasses, notably those occurring in more or sely folded or convolute, appearing more or less less exposed situations on the drier mountains filiform. (Elgon, Aberdare, Mt Kenya, Kilimanjaro, and Mt From an ecological point of view this life-form Meru). Some species occurring in less exposed may be regarded as well adapted to the afroalpine localities are not tussock-forming, however, and in climate. The innovation shoots in the tussocks the almost permanently moist afroalpine belt on the are evidently effectively protected against damage Virunga Volcanoes and Ruwenzori tussock-forming by frost, and the good temperature insulation species are less prominent than on the other provided by the tussock will also prevent freezing mountains. To distinguish between true tussock­ of the soil below it, hence counteracting soli­ forming species and those that are only loosely fluction and frost heaving in its rhizosphere, and tufted is by no means easy-in some cases high­ enabling the absorption of water by the roots to level and low-level specimens of the same species proceed without interruption (cf. Fig. 31, and may behave differently in this respect, e.g., in pages 23, 46, and 94). The felt-like mass of decaying Festuca abyssinica and Poa leptoclada. The most matter in the marginal parts of the tussock would important tussock-forming grasses are listed above also seem to act as a mulch, assisting to retard on p. 47; to these may be added at least: Agrostis evaporation from the tussock and the soil below gracilifolia and A. volkensii, Festuca abyssinica, it. As mentioned above (p. 46) the central parts of F. kilimanjarica, and Koeleria gracilis (s.lat.). Big such tussocks may therefore contain considerable and dense cushions of similar type are formed also amounts of free water even after several days of dry by a few other afroalpine monocotyledons, notably weather. Carex runssoroensis (Fig. 97; cf. HEDBERG, 1955, Exactly the same type of big grass tussocks with Pl. 2A) and C. monostachya (Fig. 101). long remaining dead leaves and culms, though The dense and firm tussocks of the afroalpine formed by other species, is known to occur in other tussock grasses consist not only of living shoots areas with similar types of climate, especially in but contain also large amounts of dead and de­ the paramo and puna of the South American Andes, caying culms and leaves, particularly in their and in subantarctic countries (cf. p. 48 above). The marginal parts, which give them a dull brownish­ same life-form may evidently also be adapted to green colour (cf. p. 46 above and the background of different types of climate, however (loc. cit.). the upper photo in the colour plate). Innovation The denomination tussock grasses ("Tussock­ shoots seem to be formed mainly in the central parts graser") for the relevant life-form has been re­ of the tussock, where they are protected against peatedly used before, e.g. by CocKAYNE (1921, p. the ' temperature (and moisture) changes of the VI etc.), BEWS (1925, p. ll0), VAN STEENIS (1939, environment by a dense felt-like mass of decaying p. 448, etc.), TROLL (1948, p. 57; 1959, p. 39, etc.), leaf- and culm-bases. The leaves of these tussock- SEWELL et al. (1954), and others. Other terms em-

Acta Phytogeogr. Suec. 49 Acaulescent rosette plants 87 ployed are bunch-grasses (BEws, 1925, p. 109; form system of RAUNKIJER (1907) the tussock FosBERG, 1944, p. 229), "Bulten-Gramineen" grasses concerned here are classified as hemicrypto­ (DrELS, 1937, p. 28), and "Biischelgraser" (TROLL, phytes, since their innovation shoots are produced 1948, pp. 52, 57; 1952 b, p. 131; 1955, p. 719; 1958, at the soil surface, being protected by withered p. 58; 1959, p. 39, etc.). In the conventional life- leaves and culms.

ACAULESCENT ROSETTE PLANTS

The afroalpines belonging to this life-form have such solifluction soils where a firm anchorage is been enumerated on p. 73 above. They are charac­ essential for survival (cf. pp. 33 and 64 above). terized by the more or less complete absence of an Plants possessing this life-form are particularly overground stem, the leaf rosette arising at or prominent on flat or gently sloping moist ground below the general level of the soil surface (cf. along streams etc. (cf. p. 33, and Figs. 68, 102, 104). HEDBERG, 1957, Plates 4-5). The subterranean part It may be worth noting in this connection that of the stem (caudex) is usually stout, as shown in species of acaulescent rosette plants are more Figs. 72 (Conyza subscaposa) and 71 (Nannoseris numerous and physiognomically more important schimperi); cf. also, for instance, R. E. FRIES, 1923, on the comparatively dry mountains along the Plate 2 (Carduus platyphyllus, etc.). It may some­ Eastern Rift Valley (Elgon, Aberdare, Mt Kenya, times become repeatedly branched in older spe­ and Kilimanjaro), where solifluction is very im­ cimens, so that a number of acaulescent leaf ro­ portant in the alpine belt, than on those along the settes form a more or less compact cushion, as Western Rift (Virunga Volcanoes, Ruwenzori), shown in Figs. 61 (Haplosciadium abyssinicum) where the alpine climate is moister and solifluction and 77 (Haplocarpha rueppellii). Some of these phenomena less important. Of the ten species of plants are equipped with a robust tap-r.o..o_t, notably acaulescent rosette plants recorded on p. 73, only 3 Oreophyton falcatum, Haplosciadium abyssinicum, are known to occur on the Virunga Volcanoes and 2 and Nannoseris schimperi (Fig. 71). In others the on Ruwenzori, whereas 9 occur on Elgon, 6 on side roots are very sturdy, as in Ranunculus Aberdare, 8 on Mt Kenya, and 7 on Kilimanjaro. cryptanthus and R. oreophytus, Carduus platyphyllus Similarly, in the South American Andes this life­ and C. nanus (cf. R. E. FRIES, loc. cit.), Haplocarpha form seems to be more important in the puna, with rueppellii, and Conyza subscaposa (Fig. 72). In its violent temperature changes and pronounced most of the species the leaves are large and only frost heaving, than in the more oligothermic and moderately xeromorphic (cf. p. 94 below). The moist paramo (cf. EsPINOSA, 1932, p. 193; TROLL, flowers are borne on more or less rudimentary 1959, p. 53). scapes (cf. Figs. 71-75). In conformity with the two life-forms treated From an ecological point of view the acaulescent above (giant rosette plants, and tussock grasses) rosette plants seem to combine two different modes the acaulescent rosette plants are well represented of adaptation. The acaulescent habit with a very also in the paramo and puna of the South American short overground stem may be viewed as an adap­ Andes (cf. GOEBEL, 1891, p. 42-44; WEBERBAUER, tation to resist the strain imposed on the water 1905, pp. 71, 73-74; 1930, p. 346; HEILBORN, 1925, balance by the afroalpine climate (cf. p. 94 below). p. 157; ESPINOSA, 1932, p. 196; DIELS, 1937, p. 29; The stout rhizome (caudex) and robust tap root, TROLL, 1959, pp. 47, 53). Their peculiar habit has or side roots, on the other hand, may be regarded been deemed advantageous for the plants by keeping as adaptations enabling the plants to subsist on the leaves away from the strong winds (GoEBEL,

Acta PhytogeogT.Suec. 49 88 Afroalpine life-forms

1891, p. 44) and by enabling them to exploit the employed for these plants are "rosette plants" genial microclimate of the uppermost layers of the (HEILBORN, 1925, p. 157), and the German equi­ soil, where day temperatures are higher than in the valent "Rosettenpflanzen" (WEBERBAUER, 1930, air (GOEBEL, loc. cit.; WEBERBAUER, 1905, p. 74; p. 347; EsPINOSA, 1932, p. 196; TRoLL, 1958, p. 71; 1930, p. 346). [As demonstrated above on p. 23, 1959, p. 53, etc.), or "Rosettenkrauter" (WEBER­ etc., the microclimate of the soil surface at night BAUER, loc. cit.). The rosette plants of the Peruvian may be far from genial, however.] HEILBORN (1925, puna were referred by TROLL ( 1959, p. 53) to the p. 157) regarded the acaulescent rosette plants as "Nototriche- Typ der Puna-Flora".-In the life­ "xerophilous types", whereas EsPINOSA (1932, p. form system of RAUNKI.lER (1907) the acaulescent 196) could find no particular xeromorphic adapta­ rosette plants would be called "rosette plants" and tions in them. classified under the hemicryptophytes. The commonest life-form denomination earlier

CUSHION PLANTS

The number of cushion plants occurring in the afro­ solifluction, as on flat or gently sloping ground alpine flora is small-only four species seem to be along streams and ponds. They are usually rather regularly cushion-forming, viz. Agrostis sclerophylla sparse but may occasionally become locally do­ (Fig. 78), Sagina afroalpina (Fig. 79), Swertia minant on wet valley bottoms at high altitude subnivalis, and Myosotis keniensis (Fig. 80). One (Fig. 81), forming plant communities reminding of more species may occasionally form cushions, the "Hartpolstermoor" described by TROLL (1948, Haplocarpha rueppellii (Fig. 77). But these five Abb. 3; 1958, p. 69). Like the acaulescent rosette seem to be the only true cushion plants of the afro­ plants, cushion plants are more important on the alpine flora (cf. p. 73). They are characterized by a mountains along the Eastern Rift Valley (which more or less profusely branche.d_stem with very harbour five species) than on those along the short internodes. The branches are densely crowded Western Rift (with one species only); cf. p. 87 (cf. Figs. 78 and 79), and each branch is terminated above. This tallies with their relative frequency by a dense leaf rosette and, in due time, a flower or in the puna and paramo of the South American inflorescence. The spaces between the branches are Andes; cf. below. filled with decaying leaf remains and soil (Fig. 79). Whereas cushion plants are so scarce in the afro­ One more afroalpine plant that approaches this alpine flora as to have escaped attention in the life-form is H elichrysum newii, which at high level past (cf. RAUH, 1939, p. 464), they are very im­ on Kilimanjaro occurs in a prostrate and more or portant in the South American mountain vegeta­ less cushion-forming modification, "forma brevi­ tion; cf., for instance, GoEBEL, 1891, p. 41; HAURI & caulis ENGL." (cf. Fig. 7; HEDBERG, 1957, Plate 1 A; ScHROTER, 1914; SKoTTSBERG, 1916, p. 125 f.; and SALT, 1954, Phot. 7). Similar dense mat-like HEILBORN, 1925, p. 157 f.; WEBERBAUER, 1930, p. growth may be assumed by high-level specimens of 347; ESPINOSA, 1932, p. 193; DIELS, 1937, p. 29; Alchemilla subnivalis on Ruwenzori. But neither of RAuH, 1939, p. 463; FosBERG, 1944, p. 233; TROLL, these becomes compact enough to be classified 1948, pp. 52-54, 56; 1958, pp. 64-71; 1959, pp. as a real cushion plant. 46, 53; CuATRECASAS, 1958, Lam. XVIII and XXII. The ecological importance of this life-form in the They are also very prominent, int. al., in the moun­ afroalpine environment is discussed above on p. tains of New Zealand (cf. GoEBEL, loc. cit.; CocK­

73 f., and on p. 94. The plants belonging to it occur AYNE, 1921, p. 195 etc; Du RrETZ , 1931, p. 69 etc.; mainly in moist localities, subjected to pronounced RAuH, loc. cit.; TROLL, 1948, p. 52), and form a

Acta Phytogeogr. Suec. 49 Sclerophyllous shrubs 8!)

conspicuous part of the vegetation on Kerguelen it seems to be particularly common in the flora of and other subantarctic islands (cf. TROLL, 1948, the Andes and in some other South-hemispheric p. 52). In South America they seem to be more areas, notably New Zealand. Presumably such abundant in the dry puna, with its recurrent hard plants have existed in those countries for a long night frosts, than in the moister and more oli­ period. The Andes would also seem to have provi­ gothermic paramo (TROLL, 1958, p. 68; 1959, p. 53; ded a convenient pathway for the dispersal of cf. also ESPINOSA, 1932, p. 193). mountain plants, whereas the high East Mrican The paucity of cushion plants (and acaulescent mountains appear to have stood isolated from each rosette plants) in the afroalpine flora, as compared other since their origin (HEDBERG, 1961, p. 918). with the floras of paramo and puna in South An analogous case is found in the isolated mountain America, is very interesting from a phytogeo­ flora of Malesia, where cushion plants are also very graphical point of view. Because of the present rare (VAN STEENIS, 1935, p. 346; 1939, p. 448), climatical parallelism between the areas concerned corresponding to the long distance from other high (cf. p. 20) the reasons for this difference must evi­ mountain areas (cf. also VAN STEENIS, 1962).

dently be historical. As demonstrated by RAUH In the life-form system of RAUNKIJER (1907 � (1939, p. 483 f.) the first prerequisite for cushion etc.) the cushion plants were brought as a sub­ growth is the capacity for a particular type of division of the chamaephytes. A number of different branching, leading to a vigorous ramification with sub-types of cushion plants were distinguished by branches of comparable dimensions. This type of HAURI & ScHROTER (1914); other systems were branching may appear in plants of diverse rela­ used by Du RIETZ (1931), and by RAUH (1939). tionships, but does not at all occur in all genera, and

SCLEROPHYL LOUS SHRUBS (AN D DWARF-S HRUBS }I

To this life-form belong about fifty afroalpine polycarpic plants, in the afroalpine flora (see p. 82 taxa, including Protea kilimandscharica, Thesium above, and HEDBERG, 1957, p. 11; cf. also WEBER­ kilimandscharicum, five species of Alchemilla, BAUER, 1905, p. 81, concerning the Andean moun­ Adenocarpus mannii, three Hypericum species, tain flora). In the absence of pronounced seasons all afroalpine Ericaceae (12 taxa), two Satureja such phenomena are but natural. For the same species, H ebenstretia dentata, four species of Bartsia, reason all of the sclerophyllous shrubs concerned Anthospermum usambarense, ten species of Heli­ are of course ''evergreen''. chrysum, Stoebe kilimandscharica, four species of The size of the afroalpine sclerophyllous shrubs Senecio, and three of Euryops. The pteridophyte (and dwarf-shrubs) varies from 1-2 dm in Blaeria. Lycopodium saururus might also be compared to filago and Satureja biflora up to some 10 m in tall these. specimens of Hypericum keniense. The stem in To distinguish between this life-form and the suf­ some of them may be almost unbranched, as in frutescent herbs is by no means easy-there is a Bartsia longiflora and B. macrophylla, but in most gradual transition, for instance, from the weakly of the relevant species it is profusely branched. frutescent Helichrysum nandense over the more or The branches are thin and distinctly woody with less strongly frutescent H. argyranthum to the thin, sometimes shiny bark. The leaves are rigid distinctly lignified H. brownei, H. amblyphyllum, and H. newii, etc . In the same way it is often dif­ 1 In the following account dwarf-shrubs will be treated ficult or impossible to distinguish between annuals together with the shrubs-no rational limit can be drawn and perennials, or even between monocarpic and between them in this flora .

Acta Phytogeogr. 8-uec. 49 90 Afroalpine life-forms and ± coriaceous, usually quite small, and often species may gain a foothold (cf. p. 25 above, folded or revolute. Their surface is often covered and Fig. 60). by a dense whitish pubescence, often mixed with Sclerophyllous shrubs occur also in the paramo gland hairs, or sometimes by gland hairs only (cf. and puna of the South American Andes (cf. GoEBEL, p. 42 above). Representative examples are shown 1891, p. 29; WEBERBAUER, 1905, p. 72; EsPINOSA, in Figs. 28 (Helichrysum newii), 70 (Euryops 1932, p. 199; CuATRECASAS, 1934, p. llO and Lam. dacrydioides), 93 (Alchemilla johnstonii), and 94 XXIX; 1958, p. 250-251 and end plate; DIELS, (Alchemilla elgonensis). 1937, p. 29; WEBER, 1958, p. 159; TROLL, 1959, The ecological significance of this life-form has p. 47), and in other areas with comparable climate, been discussed above on p. 71 f., and on p. 94. The as on the subantarctic islands (cf. TROLL, 1948, plants possessing it are evidently adapted to pp. 56-57; 1955, p. 719). The same life-form is, of withstand the vicissitudes of the afroalpine climate course, common in several other areas as well, for by their xeromorphic leaves, which should enable instance in Mediterranean and South African ve­ them to reduce drastically their transpiration getation, and in temperate mountain floras. during periods of precarious water balance, whether In the life-form system of RAUNKI1ER (1907, these be caused by slow uptake of water from the etc.) most of the sclerophyllous shrubs discussed cold soil during the night and the early part of the here would be classified as "evergreen nanophane­ day, failure of water transport through frozen rophytes without bud-cover", others (notably the stems during the same periods, or high evaporative smallest ones) as "sulfruticose chamaephytes". power of the air during the warmest part of the day Such a classification is, of course, refuted by the (cf. p. 93 below). Several of these plants tend to fact that in the absence of definite seasons the grow in dense mats or bushes (cf. Figs. 93 and 94), growth of these plants proceeds more or less which presumably create ameliorated microclima­ without interruption, and no particular "surviving tes of their own, in which some other less hardy buds" are formed.

OTHER PHANEROGAMIC LIFE-FO RMS

The five phanerogamic life-forms described above cific life-forms for them. An attempt to fit them all (giant rosette plants, tussock grasses, acaulescent into a special life-form system designed to comprise rosette plants, cushion plants, and sclerophyllous the whole afroalpine flora would rather blur the .shrubs) were all distinguished after their morpholo­ clear ecological picture provided by the five gical peculiarities, but as emphasized above these examples mentioned above. peculiarities apparently represent clear adapta­ Several of the taxa constituting the remaining tions to particular factors of the afroalpine en­ 55 % of the afroalpine flora may, of course, be vironment. To those life-forms belong about 120 fitted into the life-form system of RAUNKilER taxa-or some 45 % of the true afroalpine phanero­ (1907, etc.). There are, for instance, a number of gamic flora (cf. HEDBERG, 1961, p. 916). Physiog­ geophytes, particularly in the families Cyperaceae, nomically, these species form by far the most im­ Juncaceae, Liliaceae, and Iridaceae. Some of these portant part of the afroalpine flora, comprising all appear well adapted to grow on solifluction soils, the dominants of the main vegetation types (cf. e.g., Carex monostachya and Romulea keniensis Chapter V below). The remaining 55 % of the afro­ (cf. Figs. 102 and 104). The hemicryptophytes are alpine vascular plants show less conspicuous adap­ also rather an important group, comprising, for tations to this environment, and consequently I in�tance, many grasses, a few Alchemilla species, have not found it desirable to design further spe- and several small herbs. A few chamaephytes and

.Acta Phytogeogr. Suec. 49 Other phanerogamic life-forms 91 therophytes may also be distinguished, as well as a species (and other acaulescent rosette plants) couple of hydrophytes. But, as repeatedly em­ should be classified in RAUNKilER's system as a phasized above, a very substantial part of the afro­ hemicryptophyte is a moot point. The same type of alpine flora cannot be fitted into the life-form uninterrupted growth is displayed in a more con­ system of RAUNKilER. The reasons for this will be spicuous way by the Giant Senecios and other giant summarized below (cf. also pp. 81-84 above). rosette plants. As described above on pp. 10-18, thermic In plants where innovation shoots are formed from seasons are absent in the afroalpine belt, and subterranean stems, as in Pimpinella kilimand­ hygric seasons are on most mountains absent or scharica and Satureja kilimandschari, it is very at least less well-marked than in the lowlands. The difficult to guess whether the development of these growth of the afroalpine plants is therefore not shoots would have been arrested for a while at the restricted to definite vegetation seasons but pro­ soil surface, had the plants been exposed to a ceeds more or less uninterruptedly. Consequently it regular unfavourable season (cf. RAUNKilER, is in many cases impossible to distinguish between 1907; 1935, Figs. 21, 45, and 59). It is therefore annuals and perennials-or between monocarpic often impossible to distinguish between geophytes and polycarpic plants (HEDBERG, 1957, p. ll; cf. and hemicryptophytes in the afroalpine flora. In­ WEBERBAUER, 1905, p. 80, and p. 82 above). To cidentally, this may be difficult enough even in the take an example: Young specimens of Silene Scandinavian flora-see for instance RosvALL & burchellii flowering for the first time have a slender PETTERSSON, 1951, p. 104-105, for examples among main root and thin herbaceous stems, and from the Scandina vian orchids. them the species would easily be classified as a Yet another life-form of RAUNKilER's system therophyte. But after the first anthesis the basal that becomes difficult to distinguish in the afro­ parts of the plant continue their growth, while the alpine flora is that of the phanerophytic epiphytes. fruiting branches wither away. The specimen then This flora contains but one species that is regularly develops into a chamaephyte with a thick tuberous epiphytic, viz. Polypodium rigescens (cf. HEDBERG, main root supporting one to several richly bran­ 1957, p. 30). But in the permanently moist micro­ ched stems, suffrutescent at the base. Cf. also climate of dense Dendrosenecio forests on Ruwen­ the account of Veronica glandulosa on p. 83 zori and some of the Virunga Volcanoes several of above. the few vascular plant species normally occurring The absence of a resting period also implies that in the undergrowth may occur also epiphytically in in most species no defined "surviving buds" are moss cushions on the Senecio stems, for instance formed. The acaulescent rosette plant Nannoseris Poa schimperiana, Deschampsia flexuosa, Carda­ schimperi, for instance, consists of a thick tap-root mine obliqua, and Arabis alpina (cf. p. 97 below). merging into a very short stem carrying a dense Difficulties of the sort mentioned above make it leaf rosette (cf. Figs. 71 and 73). The latter grows all but impossible to squeeze all members of the quite continuously until its growing point is con­ afroalpine flora into the life-form system of RAUN­ sumed by an inflorescence-then the growth is KilER-or into any other similar system. Since gradually taken over by one or two lateral innova­ furthermore the life history and shoot architecture tion shoots. This may eventually lead to some of several species are insufficiently known it has branching of the caudex (cf. Figs. 61 and 75). But not been attempted to assign all afroalpine vas­ no resting buds are formed, and whether this cular plants to definite life-forms.

Acta Phytogeogr. S1tec. 49 92 Afroalpine life-forms

SOLIFLUCTION FLOATERS

To this life-form I refer a few cryptogamic plants of the vegetation period, in climates with distinct "macroscopic" size, the relatives of which are most seasons. According to sporadic notes in the litera­ often firmly attached to their substrate. The cases ture it seems to occur also, at least, at high level in observed by me comprise two mosses (Grimmia northern Ethiopia, in the alpine belt in Norway. campestris BuRCH. and G. ovata W. & M.), one and on Jan Mayen (cf. p. 68 above). fruticose lichen (Parmelia sp . near P. vagans), and Among the life-form designations earlier used for one alga [Nostoc commune (VAUCH.) BoRN. & similar loose-lying cryptogams may be mentioned FLAH.]. A description of these is provided on p. "Wanderflechten" (cf. especially MATTICK, 1951, with 66-69 above; cf. also Figs. 63-67. A detailed study quotations of earlier authors), "Aegagropiliden" would presumably disclose several more species (SERNANDER, 1927, p. 57); "licheni vaganti" (PICHI­ belonging to the same life-form. SERMOLLI, 1939, p. cxxn and Fig. 2); and The common characteristic of these plants is that "Aerolichenes" (MATTICK, loc. cit.). Most of those they occur in unattached specimens or colonies, cases will, of course, have a different ecological lying loose on top of mobile solifluction soil, where background. It should also be pointed out in this they are lifted by needle-ice formation almost connection that none of the species enumerated every night when enough moisture is available at above (with the possible exception of the Parmelia the soil surface (cf. Figs. ll-13), to subside again species) occurs exclusively as "solifluction floater". when the ice-needles melt away in the morning. Especially the mosses mentioned occur more often That way of life would seem more natural in some growing attached to the soil or to stones. microscopic algae, etc., but appears rather strange It may perhaps appear odd to append this life­ in organisms of larger size. form to the present chapter, since no other crypto­ This life-form must evidently be adapted to a gamic life-forms are considered here. But in view biotope with much open soil, and a climate with of its peculiar ecology and previously neglected frequent frost changes during the year-or during existence a short presentation seemed desirable.

Acta Phytogeog1·. Suec. 49 IV. Water economy of the afroalpine flora

Looking back at the adaptive trends described in since most roots penetrate below this layer. On the Chapter II we find that several of them may be other hand these deeper layers of soil, where most fitted into a common pattern-they represent of the roots occur, will presumably never be different ways of coping with the problems of warmed up to any considerable extent, and hence water economy in the peculiar afroalpine environ­ root absorption must be expected to be slow ment described in Chapter I. At least on most of throughout the day. the high East African mountains the most critical 2. The rate of water movement through the roots environment factors would seem to be the recur­ will evidently also be slow, particularly during the rent nightly frosts in combination with the rapid night and the early part of the day. In some plants temperature increase after sunrise. As emphasized the conduction of water may be temporarily stop­ above (p. 78), this must cause considerable diffi­ ped in some of the roots at least by their upper culties for the plants to absorb enough water from parts being frozen (cf. Fig. 109). the soil to meet the demands of transpiration, 3. The rate of water movement through the stem particularly in the early morning hours when the will depend very largely upon the dimensions and soil surface may still be frozen while the sun shines exposure of the latter. In plants with unprotected warm and the evaporative power of the air is high . thin and elongated stems the transport of water These difficulties may evidently explain the oc­ will presumably often be effectively stopped during currence of numerous xeromorphic features in the the night and the early part of the day by the stems afroalpine flora, even in the almost permanently being frozen, whereas it may be expected to proceed moist climate of Ruwenzori (cf. p. 71). But we relatively fast during the latter part of the day find in this flora not only xeromorphic features of when they have been warmed up. In stems provided "traditional" type but also some rather remarkable with some kind of temperature insulation the water variants, the significance of which seems to have transport will evidently not be brought to a stand­ been earlier disregarded. still by freezing, but on the other hand its rate will The problems of water economy in the afroal­ be rather uniformly slow. pine flora may conveniently be discussed under 4. The transpiration may obviously be intense four different headings, viz.: (1) Water absorption during the warmer part of the day in this environ­ from the soil; (2) Water movement through the ment, under the combined influence of low air roots; (3) Water movement through the stem; (4) pressure, intense sunshine, and mountain upwinds Transpiration and the maintenance of the water (cf. p. 78 above). In view of the difficulties con­ balance. nected with water absorption and water transport 1. In view of the low temperature prevailing in in this environment the maintenance of the water the upper layers of the soil during the night and balance will therefore necessitate reductions in the early part of the day the rate of water absorption transpiration in most plants during the warmest from the soil will of necessity be slow during at least part of the day. As described above most afroalpine the early part of the day (cf. p. 78 above). It will plants display various xeromorphic features enab­ probably not be seriously impeded by the frequent ling them to reduce their transpiration. Some of freezing of the few topmost centimetres of soil, them furthermore have particular water storage

Acta Phytogeogr. Suec. 49 94 Water economy tissues, enabling them to buffer large variations probably more exposed to evaporation than the in transpiration, notably the Giant Senecios (HARE, latter. The leaves of the few afroalpine cushion 1943, p. 365; cf. p. 50 above). The peculiar tem­ plants are either more or less coriaceous (Haplo­ perature-insulating water reservoirs in the spe­ carpha rueppellii, Swertia subnivalis), densely cies of the Lobelia deckenii group may possibly pubescent (Myosotis keniensis), or subulate (Agros­ also have a function of this sort. tis sclerophylla, Sagina afroalpina). The slow rates of water absorption by the roots In the giant rosette plants the stems are thick and and of water transport through them will pre­ as a rule very well insulated by a mantle of dry sumably be largely responsible for the widespread leaves or-in older stems of Dendrosenecio-by a occurrence of xeromorphic features in the afroal­ thick cork cortex (cf. p. 54 above). The water­ pine flora. The difficulties connected with water conducting tissues will therefore presumably never conduction through the stem have been met by become frozen, and a slow water transport through diverse adaptations in different afroalpine life­ the stem may occur without interruption. This forms, and in the same way the maintenance of the fact, in combination with the water-storage tissues water balance may be variously affected in dif­ mentioned above (pp. 50 and 62) may explain the ferent types of plants, as will be exemplified below. comparatively weak xeromorphic character of the In the acaulescent rosette plants the stem is ex­ huge leaves in the Giant Senecios. Also in the Giant tremely short and comparatively thick, being Lobelias the leaves are fairly large and do not ap­ insulated by the leaf bases and by being immersed pear to be particularly xeromorphic. in the upper part of the soil (cf. Figs. 71-75). In at least the larger tussock grasses the basal Consequently the distance across which the water parts of culms and leaves, as well as the soil below has to move is very short, the capacity of the organs with the roots, are well insulated against the frost for water transport comparatively large, and the (cf. p. 46 above), and the absorption and conduc­ water transport will presumably not be impeded tion of water may presumably proceed uninterrup­ by freezing. Furthermore since the evaporative tedly, though slowly. Although temperature con­ power of the air is lower at the soil surface than ditions are less rigorous a couple of decimetres higher up (cf. DAUBENMIRE, 1947, p. 93; GEIGER, above the ground than at the surface (cf. p. 22 1950, p. 89) , the maintenance of the water balance above) the upper parts of culms and leaves will in these plants should not be particularly difficult. evidently become frozen during cold nights. The It therefore stands to reason that some of the a­ leaves are as a rule distinctly xeromorphic. caulescent rosette plants, notably Oreophyton falca­ In some of the sclerophyllous shrubs a certain tum, Conyza subscaposa, and N annoseris schimperi, amount of temperature insulation may be obtained have relatively large and thin leaves witJwut for the thin stems by the formation of thick mats conspicuous xeromorphic features (Figs. 71-74; or bushes which insulate the soil surface, hence cf. GoEBEL, 1891, p. 44, footnote; and EsPINOSA, decreasing the amount of nightly cooling (cf. Figs. 1932, p. 196). 10, 60). But in most of these plants the thin woody In the cushion plants the stems are insulated stems will stand a small chance of a voiding be­ below the surface of the compact cushions (cf. coming frozen during most nights of the year, to be Figs. 77, 79, 80). Since the surface of the cushions thawed out :::ome time after sunrise. In these the appears to lose less heat by outward radiation at transport of water through the stem must therefore night than the surface of open soil (cf. p. 25), the be expected to be drastically interrupted during the stems inside it will probably be fairly well protected early part of the day-at the same time as the tem­ against freezing. Hence the water transport perature of the air rises quickly and its evaporative through the stems may be expected to proceed power must be high. It is consequently but natural more or less uninterruptedly, albeit slowly. The to find in these plants the most pronounced stems are, however, longer than in the acaulescent xeromorphic features occurring in the afroalpine rosette plants, and at least the larger cushions are flora, with small, often convolute or revolute leaves,

Acta Phytogeogr. Suec. 49 Water economy 95 dense pubescence, etc. (cf. for instance Figs. 28, 70, above life-forms, must also be adapted to the 94). afroalpine climate, since they occur in the same These examples should suffice to exemplify the flora. Their physiological adaptations may be various types of adaptation to a precarious water equally important as the morphologically discernible economy found in the afroalpine flora. It need ones described above, though they are more diffi­ hardly be stressed that most of the above discus­ cult to observe. But the fact remains that all the sion is of a highly hypothetical nature, and that main dominants of the afroalpine vegetation belong many more ecological observations and experi­ to the five life-forms described in Chapter Ill, ments are required in this field. It should also be which indicates that these may represent the most emphasized that many plants without conspicuous efficient types of adaptation to this peculiar en­ morphological adaptations, not classified to the vironment.

Acta Phytogeogr. Suec. 49 V. Principal afroalpine plant communities

INT RODUCTION

While the vascular flora of the afroalpine belt is Helichrysurn scrub, (3) Alchernilla scrub, (4) now becoming reasonably well known, its vegeta­ Tussock grassland, and (5) Carex bogs and related tion has been very little studied. A few notes on the communities (cf. HAUMAN, op . cit.; FRIES & FRIES, main features of the vegetation on the various op. cit.: HEDBERG, loc. cit. ). The relative importance mountains were published by ENGLER (1925, pp. of these communities varies considerably on dif­ ·235, 238, 265, 269, 298, 307, 317), but his descrip­ ferent mountains according to edaphic and climatic tions are merely compilations from various incom­ divergences, etc. In the following part of this chap­ mensurable sources and give no detailed accounts ter short descriptions will be given of the com­ -of the vegetation. A brief description of the vege­ munities mentioned and of their appearance on tation on the Virunga Volcanoes and Ruwenzori those East African mountains reaching into the was supplied by RoBYNS (1948a). Some more alpine belt (Virunga Volcanoes, Ruwenzori, Elgon, ·detailed contributions have been rendered by Aberdare, Mt Kenya, Kilimanjaro, and Mt Meru). HAUMAN (1933a) from Ruwenzori, SALT (1954) and In order to save space no abbreviations of au­ KLOTZLI (1958) from Kilimanjaro , FRIES & FRIES thor's names are annexed to the latin names cited (1948) from Aberdare and Mt Kenya, etc. Con­ in the vegetation descriptions below.-The nomen­ -densed accounts of the afroalpine vegetation on the clature for vascular plants follows HEDBERG, 1957, various mountains concerned were given by HED­ for mosses PoTIER DE LA V ARDE, 1955, and for BERG (1951, p. 187 f.). The five most important liverworts ARNELL, 1956. Voucher specimens for all types of plant communities represented on these bryophyte localities mentioned are listed in the last­ mountains are: (1) Dendrosenecio woodland, (2) mentioned two papers.

DENDROSENECIO WOODLANDS

'The Dendrosenecio woodlands form the most unique rule only as scattered specimens or in small groves ..and characteristic plant communities of the afro­ (Figs. 85, 86), and on Mt Meru no species of Den­ ..alpine belt. The subgenus Dendrosenecio is re­ drosenecio is known from the alpine belt. :presented on different mountains by different, In their edaphic requirements the Dendrosenecio though closely interrelated, vicarious species, and woodlands of different mountains display some the Senecio woodlands of each mountain have their fundamental resemblances: they seem to need own characteristic features, as shown in the fol- comparatively deep soil with a good supply of more lowing descriptions and accompanying pictures. or less mobile soil water. The Senecio "trees", which as a rule are only 4-6 m tall, may form patches of fairly dense forest, as on VIRUNGA VOLCANOES the Virunga Volcanoes and Ruwenzori (Fig. 82), The dominant Tree Senecio at the lower limit of the or comparatively open woodland, as on Elgon, alpine belt on Muhavura is Senecio erici-rosenii, Aberdare, and Mt Kenya (Figs. 83, 89, and 100, and towards the summit it is S. adnivalis var. -etc.). On Kilimanjaro Tree Senecios occur as a alticola (cf. SYNGE, 1937, Pl. 62). At intermediate

Acta Phytogeogr. Suec. 49 Dendrosenecio woodlands 97 levels a large fraction of the Giant Senecios seem Dendrosenecio forests of similar type evidently to be (probably hybridogeneous) intermediates occur also on the three highest of the other Virunga between these two (HEDBERG, 1957, p. 359). The Volcanoes, viz. Mikeno, Karisimbi, and Vissoke Senecios often form quite dense groves; sometimes (ScA:ETTA, 1934, p. 301 and PI. XIX; BuRTT, they are intermingled with stray specimens or 1934, pp. 155, 157, 158; ROBYNS, 1948a, p. 29). groups of Lobelia wollastonii and L. lanuriensis (cf. Fig. 87). In sheltered places one often finds RUWENZORI Galium ruwenzoriense, climbing several metres high On Ruwenzori Dendrosenecio groves occur mainly into the foliage of the Senecios by means of its on distinctly sloping ground along small ravines, recurved leaf prickles. on well-watered talus slopes, below rocky outcrops, The field layer is usually dominated by more or etc. (cf. Fig. 99). The ground is often encumbered less dense Alchemilla johnstonii scrub (Fig. 92; cf. with big boulders, interspersed with decaying also p. llO), and among its commonest subordinate fallen Senecio stems, and the local climate seems to species may be mentioned Asplenium cf. kassneri, be extremely moist and oligothermic (cf. HAUMAN, Festuca abyssinica, Poa schimperiana, Deschampsia 1933a, p. 606). This permanently moist micro­ flexuosa, Cerastium octandrum var. adnivale, Carda­ climate enables a number of small herbs and gras­ mine obliqua, Arabis alpina,1 and Peucedanum ses to thrive equally well epiphytically in moss kerstenii. cushions on the Senecio stems as on the ground (see The commonest ground mosses seem to be below). Similar conditions obtain on Muhavura Breutelia subgnaphalea and B. stuhlmannii, Brachy­ (cf. above) but not on the other mountains. thecium ramicola, Campylopus stramineus, Drepano­ The main altitudinal distribution of Senecio wood­ cladus uncinatus, and Leptodontium squarrosum. lands on the W side of the mountain was stated by Other bryophytes observed in the bottom layer are HAUMAN (loc. cit.) to be between 3800 and 4300 m. Brachythecium implicatum and B. spectabile, Breid­ On the E side they seem to be most important be­ leria africana, Lophocolea cuspidata and L. new­ tween 3900 and 4350 m (HEDBERG, 1951, p. 190). tonii, Metzgeria hamata and M. hedbergii, Mylia Accounts of the Dendrosenecio forests on the W hedbergii, and Plagiochila haumanii. slope were given by HAUMAN (1933a, p. 605 f.; Those parts of the Dendrosenecio stems from 1935, p. 4 f. and Plates); cf. also some of the plates which the leaves have fallen are often covered by in DE GRUNNE, 1937 (Figs. 28, 34, 42-44). The fol­ thick moss cushions, formed above all by Anti­ lowing description applies to groves studied by the trichia cf. curtipendula, Brachythecium vellereum, author on the E slope of the mountain, at about Campylopus stramineus, Hypnum cupressiforme, 3900-4100 m altitude in Bujuku, Bukuyungu, and Leptodontiopsis fragilifolia, and Tortula cavallii. Mugusu Valleys. On these cushions often occur large specimens of The tree layer is usually fairly dense, generally

Peltigera canina (L.) WILLD. , and stray specimens formed by Senecio adnivalis (s.lat.), tall specimens of a few vascular plants, notably Asplenium cf. of which may attain some 7-8 m in height (cf. kassneri, Poa schimperiana, Deschampsia flexuosa, Figs. 82 and 99). In the denser groves most spe­ Cardamine obliqua, and Arabis alpina.1 cimens have rather poorly developed cylinders of dry leaves (Fig. 82), and their defoliated stems are more or less covered by thick moss cushions. 1 The report of Oreophyton fal catum from this corn· munit.y on Karisimbi by MuscliLER in MILDBRAED, 1910- Lobelia wollastonii occurs fairly regularly in stray 1914, p. 219 (reiterated by ROBYNS, 1948 b, p. 221; RoBYNS specimens or small groups intermingled with the & BOUTIQUE, 1951, p. 543; and FRANCHETTI, 1958, p. Senecios (Figs. 82 and 99; cf. also HEDBERG, 1952a, 203) is certainly erroneous, being a misidentification of Fig. 4:59). Arabis alpina. Like the latter Oreophyton was reported by The field layer is mostly rather open and dis­ MusCHLER to grow both on the ground and epiphytically on Senecio branches. Concerning the general unreliability continuous; its main constituents on this mountain of MuscHLER's work see VoLKENS et al. 1915. are Festuca abyssinica, Poa schimperiana, De-

7 - 64 1 464 Hedberg Acta Phytogeogr. Suec. 49 98 Principal afroalpine plant communities

Acta Phytogeog·r. Suec. 49 Dendrosenecio woodlands 99 schampsia flexuosa, Agrostis gracilifolia, Luzula Field layer: open and discontinuous, formed by johnstonii, Cerastium octandrum var. adnivale, Festuca abyssinica, Poa schimperiana, Agrostis Ranunculus oreophytus, Cardamine obliqua, Arabis gracilifolia, Cerastium octandrum var. adnivale, alpina, Alchemilla microbetula, A. subnivalis, A. Cardamine obliqua, Arabis alpina, Alchemilla micro­ argyrophylla ssp. argyrophylloides, Peucedanum betula, and A. subnivalis. kerstenii, Galium ruwenzoriense (climbing up to 5 m Bottom layer: formed by a thick and soft moss high or more on the Giant Senecios), Helichrysum carpet, covering also the big boulders and fallen stuhlmannii, and Senecio transmarinus var. syce­ Senecio stems scattered over the ground. The phyllus . dominant species were Breutelia subgnaphalea and The bottom layer is formed by a fairly continuous, Brachythecium spectabile; other constituents col­ thick and soft carpet of mosses, covering not only lected were Breutelia stuhlmannii (mainly on boul­ the ground but also the boulders and decaying ders), Drepanocladus uncinatus, Lophocolea moelleri, Senecio- and Lobelia-stems lying on it. The most and Metzgeria hamata. On stones with thinner moss important species of this moss carpet seem to be cover occurred also Pleuropus sericeus. Brachytheciumspectabile and Breutelia subgnaphalea; The stems of the Dendrosenecios carried nu­ apart from the species enumerated below for the merous thick epiphytic moss cushions, from which sample plot were also recorded Hylocomium splen­ the following species were collected: Antitrichia dens, Polytrichum keniae, Tortula cavallii, and Metz­ curtipendula, Bartramia afro-ityp hylla, Brachythe­ geria hamata. cium ramicola, Breidleria africana, Bryum capillare, On older Dendrosenecio stems thick moss cush­ Campylopus stramineus, Drepanocladus uncinatus, ions are common (Fig. 82), whereas lichens are Grimmia perichaetialis, Leptodontiopsis fragilifolia, comparatively rare (cf. HEDBERG, 1952a, Fig. Leptodontium sublaevifolium, Zygodon intermedius 4:59). The most important epiphytic bryophyte var. strictifolius, Lophocolea muricata, Lophozia seems to be Antitrichia curtipendula, but numerous ruwenzoriensis, and Radula meyeri. On some of the others occur; in addition to those enumerated moss cushions occurred Peltigera canina (L.) WILLD. below from the sample plot the following were obtained: Hypnum cupressiforme, Anastrophyllum ELGON gambaragarae, Lophocolea moelleri, Plagiochila colo­ The Dendrosenecio woodlands in the alpine belt on rans, and Radula boryana. The only epiphytic Elgon occur as a rule on sloping and more or less lichen identified from this habitat is Peltigera well-drained soil. They are much more open than canina (L.) WILLD. In the moss cushions also those on the Virunga Volcanoes and Ruwenzori. grow a few epiphytic vascular plants, notably Below an altitude of some 3900 m they are usually Polypodium rigescens, Poa schimperiana, Cardamine formed by Senecio elgonensis (Figs. 23, 89), whereas obliqua, Arabis alpina, and Senecio sp . (juv.). above this altitude another species takes prece­ An example of this community is afforded by the dence, Senecio barbatipes (Figs. 42, 100). At inter­ following analysis of a square of 10 x 10 m, from a mediate levels these two are about equally com­ Dendrosenecio forest at about 4000 m altitude in mon. The first-mentioned species seems to be more Mugusu Valley, not far from the lake. The photo­ dependent upon abundant soil moisture than the graph in Fig. 82 was taken in the same locality. latter (HEDBERG, 1957, pp. 228-230). Tree layer: fairly dense, eo-dominated by Senecio The undergrowth in these Senecio woodlands is adnivalis var. adnivalis and S. adnivalis var. most often dominated by Alchemilla elgonensis petiolatus, with scattered specimens of the tall (Figs. 23, 42, 100), or sometimes by Alchemilla (herbaceous) Lobelia wollastonii. johnstonii (Fig. 93). A few analyses of corresponding

Fig. 82. View from the Dendrosenecio forest analyzed on p. 99. The dominant plant is Senecio adnivalis (s.lat.), and the narrow-leaved plants in the foreground are Lobelia wollastonii. In such dense forests the dead Senecio leaves are often shed at an early stage-possibly the permanently moist air makes them susceptible to fungus attacks. Uganda, Ruwen­ zori, Mugusu Valley not far from the lake, 4000 m. Photo 0. Hedberg 28.3.1948.

Acta Phytogeogr. Suec. 49 100 Principal afroalpine plant communities

Alchemilla communities are given below on p. 111. Because of the peculiar life-form of Senecio Other constituents of the undergrowth recorded in brassica, with its prostrate woody stem (cf. HED­ Senecio elgonensis woodland were Cardamine ob­ BERG, 1957, p. 226), this plant can hardly be called liqua, Sagina cf. abyssinica, Peucedanum kerstenii, a tree, nor can its communities be called woodlands. and Marchantia polymorpha. In a fairly dense Sene­ A general view of the Senecio brassica community cio barbatipes stand were recorded Luzula johnstonii, is shown in Figs. 41 and 81; cf. also FRIES & FRIES, Parietaria debilis, Arabis alpina, Anthriscus sylves­ 1922a, Fig. 8; SYNGE, 1937, PI. 68; JEANNEL, tris, Heracleum elgonense, Myosotis vestergrenii, and 1950, PI. XX; HEDBERG, 1951, PI. III B, etc. Senecio sotikensis; of bryophytes were collected I made no analysis of this community, but a brief Antitrichia curtipendula, Hypnum cupressiforme, description of the "Senecio brassica-Festuca As­ Neckera hedbergii, Orthotrichum rupestre, Pleuropus sociation" has been given by FRIES & FRIES (1948, sericeus, Radula meyeri, and Rhytidium rugosum. p. 46). Fairly common constituents of Senecio barbatipes The communities dominated by Senecio kenioden­ communities were also Euryops elgonensis and dron, on the other hand, certainly merit to be called Lobelia telekii. woodlands (Figs. 33 and 83). The Senecio stems may reach at least 6 m in height and 1.85 m in girth at ABERDARE the base. They are usually well spaced (cf. Fig. 83); On Aberdare two Giant Senecios occur in the alpine the woodland pictured in Fig. 33 was said to be one belt, Senecio battiscombei (Syn. S. aberdaricus) and of the three densest Senecio forests on Mt Kenya S. brassiciformis. The first of these forms groves (fide V. Klarville, Naro Moru). The microclimate of along ravines and precipices in the ericaceous and a Senecio keniodendron grove is therefore presum­ alpine belts, whereas the latter occurs in quantity ably not very different from that of surrounding on gently sloping moist ground at corresponding "tree" -less communities. altitudes. I have no observations on their com­ The field layer in Senecio keniodendron woodland munities. The record provided by FRIES & FRIES is usually formed by Alchemilla scrub, dominated (1948, p. 51) for the "Senecio brassicaeformis in most cases by A. argyrophylla ssp. argyrophylla, Association" was obviously obtained from a loca­ or on moister patches sometimes by A. johnstonii. lity situated well below the alpine belt, whereas In drier or recently burnt localities the under­ the notes on "The Senecio aberdaricus Association" growth may occasionally be dominated by large (op. cit., p. 52) may have been made in the lower­ tussocks of Festuca pilgeri (Fig. 22; cf. R. E. FRIES, most part of the alpine belt. 1925, Taf. 40). The most conspicuous herbs of these woodlands are Lobelia telekii and H eracleum MT KENYA elgonense; other constituents of the undergrowth are The two Giant Senecios regularly occurring in the Festuca pilgeri, Poa schimperiana, Descharnpsia alpine belt on Mt Kenya, Senecio brassica and S. flexuosa, Anthoxanthum nivale, Cerastiurn octandrurn keniodendron, have partly dissimilar ecological var. adnivale, Ranunculus oreophytus, Cardarnine preferences. The former occurs mainly on gently obliqua, Myosotis keniensis, Galiurn glaciale, Valeri­ sloping ground where the percolating soil water ana kilirnandscharica ssp. lcilirnandscharica, and reaches more or less to the soil surface; it is most Senecio keniophytum. The Alchernilla scrub of this abundant at altitudes between 3500 and 4100- woodland evidently deviates very little from that on

4200 m (HEDBERG, 1957, p. 227). The latter occurs open ground, which is more fully described below most often on more steeply sloping and better (p. 112). drained soil, where the percolating soil water is The moss layer on the ground in the Senecio well below the surface; it is most prominent at keniodendron woodland is rather discontinuous. Its altitudes between 4000 and 4500 m. At intermediate most important species appeared to be Tortula levels, where both occur, their communities usually cavallii, Hylocornium sp lendens, and Hypnum display a distinct zonation (Fig. 41). cupressiforrne. Some other representative crypto-

Acta Phytogeogr. Suec. 49 Dendrosenecio woodlands 101

Fig. 83. View of the Senecio keniodendron woodland described below. The dominant plant of the tree layer is Senecio keniodendron, of the field layer Alchemilla argy1·ophylla ssp. aTgyrophylla. Note the wilted old inflorescences hanging down from the Senecios in the foreground. Kenya, Mt Kenya , Teleki Valley, 4100 m. Photo 0. Hedberg 2.8.1948. Cf. also Figs. 33 and 84. gams are listed below from the sample square; tween 2 m and 5 m; most of them were 3.5-4 m tall. to these may be added Brachythecium atrotheca The girth of the thickest stem 0.3 m above the and Bryum ellipsifolium. ground was 1.85 m; its bark ridges were up to 12 The epiphytic vegetation on the Senecio stems cm hjgh. That specimen carried 11 leaf rosettes was completely dominated by Tortula cavallii (cf. and showed three storeys of branching, indicating Fig. 33). No epiphytic vascular plants were recorded three past periods of flowering (cf. HAUMAN, 1935, in this community. p. 9; HEDBERG, 1957, p. 355). An example of the composition of this Senecio Field layer: the most conspicuous plant of this keniodendron woodland is furnished by the fol­ was Lobelia telekii, represented within the square of lowing analysis of a square of 10 x 10 m, from a one fruiting specimen and two leaf rosettes (cf. grove situated on the S slope of Teleki Valley at Fig. 84 from another stand). The dominant species an altitude of about 4100 m. The inclination of the of this layer were Alchemilla argyrophylla ssp. ground was about 15°-20° towards NE. The argyrophylla and A. johnstonii. These do not as photographs reproduced in Figs. 33 and 83 were a rule grow mixed together but form separate taken in the same grove; cf. also Fig. 84 from segments, the latter species prevailing on patches another stand. with slightly moister ground. The only other field Tree layer: formed by Senecio keniodendron, layer species found within the square were stray represented within the square by 18 living stems specimens of Poa schimperiana, Agrostis sp. (ster.), and one dead. The height of the stems varied be- Cardamine obliqua, and Galium glaciale.

Acta PhytogeogT. Suec. 49 102 Principal afroalpine plant communities

a

Fig. 84. Line transect through a patch of Senecio lceniodendron woodland on the S slope of Teleki Valley at about 4150 m altitude. The inclination is drawn approximately to scale. The symbols denote: (a) Senecio lceniodendron; (b) Alchemilla argyrophylla ssp. argyrophylla; (c) Lobelia telelcii; (d) decaying fallen trunk of Senecio lceniodendron; (e) stones at the soil surface; {f) Poa schimperiana; (g) Myosotis lceniensis.

Bottom layer: consisted of a rather discontinuous The epiphytic vegetation on the denudated parts moss carpet, dominated around the Senecio stems of the Senecio stems consisted of some relatively by Tortula cavallii, and between the stems by thin moss cushions, formed by Tortula cavallii. Hylocomium splendens and Hypnum cupressiforme. Another, slightly more open Senecio keniodendron On slightly more open ground occurred Breutelia grove was studied by means of a line transect; the madagassa var. kivuensis, and on pieces of decaying result is shown in Fig. 84. wood, stem bases of Alchemilla, etc., grew scattered Earlier annotations from Senecio keniodendron cushions of a few other mosses, notably Bryum communities have been published by FRIES & argenteum, Campylopus stramineus, Ceratodon pur­ FRIES (1948, p. 47), under the headings "Senecio pureus, and Leptodontium tenerascens, as well as keniodendron-Festuca Association", and "Senecio stray specimens of Cladonia pyxidata (L.) FR. and keniodendron-Alchemilla keniensis Association". Peltigera canina (L.) WILLD. On slightly moister Photographs of Senecio keniodendron woodland ground along a streamlet occurred Bryum afrocalo­ have been published, int. al., by FRIES & FRIES, phyllum and Campylium "hispidulum var. sommer­ 1922a, Figs. 1 and 3; R. E. FRIES, 1925, Taf. 39B feltii" .1 and 40; and JEANNEL, 1950, Plates 22 and 24.

1 According to CRUNDELL & NYHOLM (1962, p. 196) " KILIMANJARO this moss is certainly a Oampylium but it seems nearer to 0. stellatum or 0. chrysophyllum than to the hispidulum Only one species of Dendrosenecio seems to occur group." regularly in the alpine belt on Kilimanjaro, viz.

A.cta Phytogeogr. Suec. 49 Dendrosenecio woodlands 103

Fig. 85. Scattered specimens of Senecio cottonii along the valley of a small stream. Note the compact leaf cylinders. Tan­ ganyika, Kilimanjaro, near Peter's Hut in the ericaceous belt, 3800 m. Photo A. Holm, June 1948.

Senecio cottonii (cf. HEDBERG, 1957, p. 230). Another lOO and 200 specimens of Senecio cottonii was species, S. kilimanjari, is largely confined to the reported by SALT (1954, p. 386) from the base of ericaceous belt, though it may reach the lowermost Platz Peak on the Shira Plateau. Another Senecio part of the alpine belt in favourable localities (op. woodland is said to exist on the SW slope of cit., p. 228). The dryness on the upper parts of Kilimanjaro (op. cit., p. 388). I never came across Kilimanjaro leaves very little space for these any collection of Dendrosenecios worthy of being plants, however. They are entirely confined to called a grove in the alpine belt on Kilimanjaro, and such edaphically favoured places where percolating consequently I have no notes from such a com­ soil water reaches close to the surface, as in stream munity. ravines (Fig. 85), at the base of steep rocks (Fig. 10, background), or in more or less strongly sloping MT MERU Carex bogs (Fig. 86; cf. p. 127 below; and GILLMAN, No species of Dendrosenecio occurs in the alpine 1923, p. 10; HEDBERG, 1951, p. 193). Usually they belt on Mt Meru, which appears to be altogether too occur alone or a few together. A grove of between dry for these plants.

Acta Phytogeogr. S'uec. 49 104 Principal afroalpine plant communities

Fig. 86. Sloping Oarex monostachya bog in the upper part of the ericaceous belt, with scattered specimens of Senecio cot­ tonii. Tanganyika, Kilimanjaro, S slope above Peter's Hut, 3900 m. Photo 0. Hedberg 23.6.1948.

HELICHRYSUM SCRUB

Another very striking community in the afroalpine very porous sandy soil. In the following a brief belt is Helichrysum scrub. This is particularly im­ description is given of the types of Helichrysum pressive on Ruwenzori, but similar communities, scrub encountered on each of the mountains in­ though dominated by other species of Helichrysum, vestigated. do in fact occur in the alpine belt on the other rele­ vant mountains as well. They show a wide variation VIRUNGA VOLCANOES in height and denseness, according to the dominant On comparatively dry, stony ridges in the alpine species as well as to differences in habitat and cli­ belt on Muhavura the vegetation usually consists of mate, from the dense and often 2 m tall H elichry­ Helichrysum newii scrub (Fig. 87; cf. ScA:ETTA, sum stuhlmannii scrub in the lower part of the alpine 1934, p. 301 and Pl. XX, Fig. 35; RoBYNS, 1948 a, belt on Ruwenzori (cf. Fig. 88, and DE GRUNNE, p. 29). I have no detailed notes from this community 1937, PI. 34) to the open c?mmunity of prostrate which appears to be frequent also on the other high H elichrysum newii in the upper part of the alpine Virunga Volcanoes (cf. BURTT, 1934). belt on Kilimanjaro (cf. e.g. Fig. 7, and SALT, 1954, Photo 7). The abundance of Helichrysum scrub also RUWENZORI varies a good deal between different mountains. In the alpine belt on Ruwenzori the Helichrysum Common to all types of it described below seems to stuhlmannii scrub is one of the most important plant be that they occur either on rocky ground or on communities. In the lower part of the belt it forms

Acta Phytogeogr. Suec. 49 Heli chrysum scrub 105

Fig. 87. Uganda, Kigezi Distr. , S slope of Muhavura, 4000 m. In the foregroun,d Helichrysum newii scrub on rocky ground, in the background open Dendrosenecio woodland with much Lobelia wollastonii. Photo 0. Hedberg 8.10.1948. thickets up to 2 m high, which may be rather dif­ ericaceous belt from about 3300 m, being here ficult to penetrate where no path has been cut­ particularly common at the margin of Carex bogs, under normal conditions they seem to be practi­ and is very important throughout the alpine belt cally always dripping wet. This community occurs up to at least 4500 m (E slope of Mt Speke). The mainly on moderately to steeply sloping rocky floristic composition of the H elichrysum stuhl­ ground with thin soil cover, extending on the W mannii scrub on the Congo slope of Ruwenzori has slope between 3900 and 4600 m (HAUMAN, l933a, been decribed by HAUMAN (lac. cit.). A small seg­ p. 612). On the E slope it occurs already in the ment of the same community at 4000 m altitude

Acta Phytogeogr. Suec. 49 106 Principal afroalpine plant communities

Fig. 88. Helichrysum stuhlmannii scrub with scattered specimens of Senec1:o adnivalis and (to the right) Lobelia bequaertii in the uppermost part of the ericaceous belt. Uganda, Ruwenzori, Bukurungu Valley,

3700 m. Photo 0. Hedberg 3.4. 1948. in the Mugusu valley on the Uganda side had the lower part of the alpine belt is exemplified in Figs. following composition. 6, 26, and 88; cf. also DE GRUNNE, 1937, Figs. 33-35; Shrub layer: completely dominated by Helichry­ SYNGE, 1937, Pis. 28, 36; JEANNEL, 1950, Pl. 2. sum stuhlmannii (up to 1.5 m tall), mixed with oc­ ELGON casional specimens of Lobelia wollastonii and On Elgon Helichrysum scrub does not cover at all Senecio adnivalis (coiL). such large areas in the alpine belt as it does on Field layer: rather discontinuous, patchwise Ruwenzori. It occur; in the main on rocky ground dominated by Alchemilla argyrophylla ssp. argy­ with thin soil cover, and is most often dominated rophylloides or Carex runssoroensis. Other vascular by Helichrysum citrispinum (coli.) and H. am­ plants observed in this stratum were Agrostis blyphyllum, sometimes by H. newii. Other common gracilifolia, Deschampsia flexuosa, Galium ru­ species are Alchemilla johnstonii, Festuca pilgeri, wenzoriense, and Polypodium rigescens. Kniphofia snowdenii, and Senecio snowdenii, as Bottom layer: usually dominated by Breutelia well as scattered specimens of Senecio elgonensis, stuhlmannii and B. subgnaphalea. On strongly Lobelia telekii, Carduus keniensis, and Peucedanum sloping rocky ground the dominant species is kerstenii (cf. Fig. 89). No detailed analysis was made more often Campylopus stramineus. Other bryo­ of this community. Photographs illustrating it phytes collected here were Rhacocarpus humboldtii, have been published, int. al., by SYNGE (1937, Pl. Leptodontium sublaevifolium (especially below dead 43), and JEANNEL (1950, Pl. 29). Helichrysum stems), and Sphagnum pappeanum. Epiphytes: on the H elichrysum stems grew a ABERDARE number of lichens (undetermined so far), together The distribution of Helichrysum scrub on the small with Anastrophyllum gambaragarae and Antitrichia afroalpine calottes of Aberdare is very little known. curtipendula. On rocky ground in the alpine belt occurs, int. al., In the upper part of the alpine belt the Heli­ H elichrysum chionoides and H. newii, both of which chrysum stuhlmannii scrub becomes low and very appear to form scrub communities. Another per­ open with much bare ground between the rocks, cf. tinent species is H. gloria-dei. No analysis of such HEDBERG, 1955, Pl. 4A. Its normal aspect in the vegetation is, however, available.

Acta Phytogeog1·. Suec. 49 Heli chrysum scrub 107

Fig. 89. Uganda, Elgon, vegetation in the crater at about 3800 m. In the foreground a small hill covered by Helichrysum scrub, with scattered specimens of Senecio elgonensis. Behind the hill extends a stretch of Alchemilla elgonensis scrub, followed by an area of grassland. Photo 0. Hedberg 19.5.1948.

MT KENYA Shrub layer: dominated by Helichrysum citri­ Also on Mt Kenya the most characteristic plant spinum var. armatum, with occasional specimens of communities of rocky outcrops with thin soil cover Alchemilla argyrophylla ssp. argyrophylla, Philippia in the alpine belt are various types of H elichrysum keniensis ssp. keniensis, Blaeria filago ssp. filago, scrub. Below some 4000 m altitude this is often Helichrysum brownei, and H. cymosum ssp. fruti­ formed by Helichrysum chionoides, whereas at cosum. higher altitudes H. citrispinum var. armatum and Field layer: discontinuous, its most conspicuous H. brownei are the most important dominants. Also elements being scattered tussocks of Festuca pilgeri, on Mt Kenya the areas covered by H elichrysum Agrostis trachyphylla and A. leptophylla, and scrub are relatively small. Pentaschistis minor. Less important constituents From a segment of Helichrysum scrub near Naro were Calamagrostis hedbergii, Senecio keniophytum, Moru Tarn in Teleki Valley, at an altitude of about and Nannoseris schimperi. Much of the ground 4200 m, the following notes were made (cf. Fig. 90). consisted of bare rock.

Acta Phytogeogr. Suec. 49 108 Principal afroalpine plant communities

Fig. 90. Helichrysum scrub on ice­ polished rocky ground. The do­ minant plant was Helichr-ysum citrispinum var. armatum; other important constituents were Al­ chemilla argyrophylla ssp. argy- 1·ophylla and Festuca pilgeri ssp. pilgeri (cf. p. 107). In the back­ ground appear, int. al., Philippia keniensis ssp. keniensis, Senecio keniodendr-on, and Senecio brassica. Kenya, lVIt Kenya, Teleki Valley. at Teleki Tarn, 4250 m. Photo 0. Hedberg 4.8.1948.

KILIMANJARO m, whereas H. newii is the commoner of them above In the alpine belt on Kilimanjaro Helichrysum that level, as well as in the upper part of the scrub is again one of the most important plant ericaceous belt (cf. KLOTZLI, loc. cit.). Above an communities. On the SE slope of the mountain it altitude of some 4300 m the Helichrysum newii appears to be the most widespread of all commu­ shrubs are generally prostrate (forma brevicaulis nities bebveen 4100 and 4350 m altitude, extending ENGL.); cf. Fig. 7; HEDBERG, 1957, Pl. lA; and also down into the upper part of the ericaceous SALT, 1954, Photo 7. No detailed analysis was made belt and upwards into the higher part of the alpine of the Helichrysum scrub on Kilimanjaro, but some belt (Figs. 10 and 60; cf. also SALT, 1954, Photos of its most important subordinate species are 6 and 7; and KLOTZLI, 1958, p. 51). On the Shira evidently Pentaschistis borussica, P. minor, Festuca Plateau to the W the "Helichrysum community" pilgeri ssp. supina, F. abyssinica, F. kilimanjarica, was likewise stated to occupy a greater area than Koeleria gracilis var. convoluta, Helichrysum cy­ any other (SALT, op . cit., p. 384 and Photo 3). The mosum ssp. fruticosum, Senecio meyeri-johannis, dominant H elichrysum species on this mountain are Euryops dacrydioides (especially on rocky ground), H. newii (Figs. 7 and 28) and H. citrispinum var. and Carduus keniensis (cf. also SALT, 1954, p. 384; hoehnelii (Figs. 10 and 60). Unlike the Helichrysum and KLOTZLI, 1958, p. 47-49). communities described above from the other mountains those on Kilimanjaro are not confined to MT MERU rocky ground-on the contrary they seem to occur Because of the edaphic monotony and dryness of equally often on rather deep, porous soil. The most of the alpine belt on Mt Meru Helichrysum substrate of this community is usually very dry, scrub is very scarce. Its main occurrence seems to and the vegetation does not as a rule cover more be just at the upper border of the ericaceous belt than half of the ground, often much less (cf. Figs. (Fig. 91), whereas higher up it was only found in a 10 and 60). Solifluction phenomena may be very few places on rocky outcrops, one of which was just important (cf. p. 64 and Fig. 60). Of the two domi­ below the summit of the mountain. The dominant nants mentioned above Helichrysum citrispinum species is Helichrysum citrispinum var. hoehnelii, var. hoehnelii seems to be the most prominent in often associated with Pentaschistis mannii in the the lower part of the alpine belt-up to about 4350 field layer. No analysis was made of this community.

Acta Phytogeogr. Suec. 49 H elichrysum scrub 109

Fig. 91. Helichrysum citrispinum var. hoehnelii scrub at the lower limit of the alpine belt. In the background to the left Adenocar­ yus mannii. Tanganyika, Mt Meru, S slope, about 3450 m. Photo 0. Hedberg 29.10.1948.

Fig. 92. AlchemillaJohnstonii scrub�

about 0.6-0.7 m high, adjoining to Dendrosenecio woodland. Uganda, Kigezi Distr., Muhavura, W slope

3800 m. Photo 0. Hedberg 6.10. 1948.

Acta Phytogeogr. Suec. 49 llO Principal afroalpine plant communities

Fig. 93. Alchemilla johnstonii scrub , with a few branches of Alchemilla elgonensis in the upper le ft corner. Kenya, Elgon, in the crater at lVIaji ya Moto, 3600 m. Photo 0. Hedberg 18.5.1948.

ALCHEMILLA SCRUB

Less conspicuous than the Dendrosenecio woodland VIRUNGA VOLCANOES and the Helichrysum scrub but no less characteristic On Muhavura Alchemilla johnstonii scrub is one of of the afroalpine vegetation are various types of the most important plant communities on gently Alchemilla scrub (Figs. 92-94, 100, 106). These are sloping ground in the alpine belt (Figs. 92 and 95; cf. usually dominated in moist localities by Alchemilla BuRTT, 1934, p. 151; and HEDBERG, 1951, PL IliA), johnstonii s. lat. (Figs. 92, 93), and on slightly drier as well as in the upper part of the ericaceous belt. ground by species of the section Subcuneatifoliae Very often it forms the undergrowth of Dendra­ (DE WILD .) RoTHM., notably A. argyrophylla, A. senecio woodland (cf. p. 97 above, with species list). subnivalis, and A. elgonensis (Fig. 94). Edaphically, Also on the other high Virunga Volcanoes Alche­ Alchemilla scrub seems to favour gently sloping and milla johnstonii scrub appears to be one of the comparatively well-drained ground, such as morai­ principal plant communities in the alpine belt (cf. nes and not too steep lava ridges. Its importance BuRTT, 1934, p. 153-158). varies on different mountains according to dif­ ferences in topography, soil, climate, and frequency of fires.

Acta Phytogeogt·. Suec. 49 Alchemilla scrub 111

TABLE 7. Analysis of Alchemilla scrub. Ruwenzori� Mugusu VallE,Y, 3900 m altitude, 27.3.1948.

Squares 1-6 were laid on fairly dry ground on a moraine in the central part of the valley, squares 7-10 on slightly moister ground bordering a Carex runssoroensis bog. (For further explanation sea p. 132.)

Square number 2 3 4 5 6 7 8 9 10

Alchemilla argyrophylla ssp . argyrophylloides 5 5 5 5 5 5 4 4 2 3 Alchemilla johnstonii 5 5 5 5 Deschampsia flexuosa Festnca abyssinica .Poa schimperiana

Brachythecium spectabile 2 l 3 3 3 '"' Breutelia subgnaphalea 4 3 5 0 5 3 3 2 3 3 Hylocomium splendens 4 Lophocolea cuspidata Lophocolea mu1·icata Lophozia hedbergii M etzgerict elliotii M etzgeria hamata

Bryophytes and lichens, together 4 3 5 4 5 4 4 3 3 3

RUWENZORI favoured the expansion of grassland at the expense In the alpine belt on Ruwenzori various types of of the Alchemilla scrub (cf. Figs. 18 and 23). An Alchemilla scrub are common, usually dominated analysis of 10 one m2-squares of this community is in the lower part of the belt by A. johnstonii in given in Table 8. An interesting detail is that moist patches, and by A. argyrophylla ssp. argy­ Valeriana kilimandscharica (s.lat.) occurs as a rophylloides on slightly drier ground (cf. Fig. 99). regular constituent of the Alchemilla scrub on Elgon In the upper part of the same belt the dominance is but is absent from the corresponding communities usually held by A. subnivalis, which may form on Ruwenzori (where the species is lacking alto­ pioneer communities on fresh moraines etc. The gether) and Mt Kenya (cf. p. 112 below). Alchemilla scrub on the W side of the mountain was described in some detail by HAUMAN (1933a, p. ABERDARE 609). An analysis of 10 one m2 squares from this community on the E slope is given here in Table 7. On Aberdare very little Alchemilla scrub was noted in the alpine belt; it seems to be more im­ ELGON portant on open places in the ericacous belt and Also on Elgon Alchemilla scrub is a very important the upper part of the montane forest belt (cf. FRIES plant community in the alpine belt, especially on & FRIES, 1948, p. 42). Also on Aberdare Valeriana unburnt areas, reaching to the highest summits, and kilimandscharica (s. lat.) appears to be a regular occurring also on open places in the ericaceous constituent of the Alchemilla scrub. No analysis belt. It is usually dominated on moist ground by A. was made on this mountain, but what appears to johnstonii (Fig. 93), on drier ground by A. elgonensis be the same community was analysed at lower (Fig. 94). Both types are common as undergrowth levels on the mountain by FRIES & FRIES (lac. �it.) of Senecio woodland (cf. p. 99, Figs. 42, 47, 100). under the name "The Alchemilla argyrophylla On this mountain the recurrent fires seem to have Roberti-aberdarensis Association''.

Acta Phytogeogr. Sueo. 49 112 Principal afroalpine plant communities

TABLE 8. Analysis of Alchemilla scrub. Elgon.

Squares 1-5: In the crater near Maji ya Moto, alt. 3600 m, on sloping ground not far above the hot spring. 6: In the -crater between the W pass and Maji ya Moto, alt. 3750 m, on almost flat and slightly moist ground. 7: NW slope of Koitob­ bos, on moist ground bordering a Oarex runssoroensis bog, alt. 3750 m. 8-10: In the crater below Jackson's Peak, alt. 4100 m, in open Senecio barbatipes woodland on gently sloping ground. May 1948. (For further explanation see p. 132.)

Square number 2 3 4 5 6 7 8 9 10

Agrostis sp. (ster.) A lchemilla elgonensis 5 5 4 5 5 4 4 5 5 4 Alchemilla johnstonii 4 4 5 2 5 5 5 Oardamine obliqua Oerastium sp. (ster.) Festuca abyssinica Galium sp. (ster.) Koeleria gracilis var. convoluta Oxalis corniculata Poa leptoclada Poa schimperiana 1 Senecio barbatipes, seedlings 1 V aleriana kilimandscharica ssp. elgonensis 2 3 4 2 Veronica glandulosa 2 l Viola eminii

Brachythecium implicatum Breutelia stricticaulis 4 4 4 2 Bryum homalobolax 2 Otenidium squarrifolium 3 H ylocomium splendens Lophocolea moelleri Sematophyllum elgonensq, Trichostomum cylindricum 2 Oladonia sp. 1 Oollema sp. 2 Nephroma sp. Sticta sp.

Bryophytes and lichens, together 3 2 2 4 2 2 4 4 2

Other species occurring in the same community: Arabis alpina, Swe1·tia crassiusculct, Helichrysum amblyphyllum, Eenecio snowdenii .

.1\'IT KENYA stonii also occurs in a mosaic vegetation together On this mountain Alchemilla scrub is again a very with Festuca pilgeri on sloping solifluction soil important plant community, especially in the (Figs. 69 and 96; cf. also pp. 70 and 116). The upper part of the alpine belt. Alchemilla argyro­ floristic composition of the former two variants of

phylla scrub seems to have about the same eda­ Alchemilla scrub IS analysed m Table 9; a line phic requirements as Senecio keniodendron wood­ transect through a patch of Festuca pilgeri-Al­ land, and often forms the undergrowth of the latter chemilla johnstonii vegetation is given in Fig. 96. -(cf. Figs. 33 and 83, and p. 100 above). Alchemilla johnstonii scrub occurs on slightly moister ground, KILIMANJARO -int. al. in moist patches of Senecio keniodendron On the SE slope of Kilimanjaro Alchemilla scrub woodland, along streamlets, etc.; Alchemilla john- (dominated by Alchemilla johnstonii and A.

Acta Phytogeogr. Suec. 49 Alchemilla scrub 113

TABLE 9. Analysis of Alchemilla scrub. Mt Kenya.

Squares 1-4: Teleki Valley, alt. 4200 m, in open Senecio keniodendron woodland on moderately sloping ground (towards N; square 2 near the transition to the Festuca pilgeri grassland). 5: Same locality, but stronger inclination of the ground (c. 20°-25°); almost half of the square was occupied by bare soil with signs of solifluction. 6-9: Teleki Valley, N slope, alt. 4100 m, in open Senecio keniodendron woodland on moderately sloping ground (15°-20° towards NE); square 6 on slightly flushed ground along a trickle, 7 on somewhat drier ground, and 8 on fairly dry ground with stony surface. July-August 1948. (For further explanation see p. I32).

Square number 2 3 4 5 6 7 8 9

A lchemilla argyrophylla ssp. argyrophylla 5 5 5 5 5 5 2 5 Alchemilla johnstonii 5 I 5 Anthoxanthum nivale Ambis alpina Cf. Deschampsia flexuosa ( ster.) Festuca pilgeri I Galium glaciale I Lobelia telekii, juv. 2 JVI yosotis keniensis Oreophyton falcatum Poa schimperiana Senecio keniodend1·on, seedlings Senecio purtschelleri Small, sterile grass (Agrostis sp. ?)

Antitrichia kilimandscharica Brachythecium atrotheca B1·yum argenteum Bryum ellipsifolium Ceratodon purpureus H ylocomium splendens 2 Hypnum cupressifonne 3 5 Leptodontium johannis-meyeri Ortho trichum 1·upest1·e Tortula cavallii 4 3 3 Small liverwort (indet.) Cladonia sp. Peltigem canina (L.) Willd.

Bryophytes and lichens, together 4 3 3 3 5

argyrophylla ssp. argyrophylla) occurs in the alpine existence. Its floristic composition on the Shira belt mainly in small patches along streams and Plateau has been summarized by SALT (loc. cit.). Carex bogs (cf. Fi g. 106), being virtually absent from all other types of localities. On the Shira MT MERU Plateau to the West, bog communities cover large In the alpine belt on Mt Meru only very small pat­ areas, and here Alchemilla scrub is evidently much ches of Alchemilla (johnstonii) scrub were observed. more important (SALT, 1954, p. 382). On Kili­ Evidently the amounts of soil water available here manjaro this community seems to be entirely are insufficient for this community. No analysis is dependent upon a high soil water level for it s available.

8-641464 Hedberg Acta Phytogeogr. Suec. 49 114 Principal afroalpine plant communities

Fig. 94. Alchemilla elgonensis scrub, with scattered specimens of Valeriana kilimandscharica ssp. elgonensis, Senecio snow­ denii, etc. Kenya, Elgon, in the crater at Maji ya Moto, 3600 m. Photo 0. Hedberg 18.5. 1948.

TUSSOCK GRASSLAND

At least on the drier of the mountains concerned and Ruwenzori this type of vegetation is very one of the most widespread types of plant com­ scarce. It seems to thrive much better on such drier munities is tussock grassland. These grasslands are and less steeply sloping mountains as Elgon, Mt dominated by tall, tussock-forming grasses such as Kenya, and Kilimanjaro, and tends to be parti­ Andropogon amethystinus, Pentaschistis spp., Fes­ cularly luxuriant in areas where occasional fires tuca pilgeri, Festuca abyssinica, Festuca kilima­ occur (cf. p. 36). In such localities it will often en­ njarica, Poa leptoclada, Koeleria gracilis, Agrostis croach upon the Alchemilla scrub, which appears trachyphylla, and A. volkensii. The tussocks are to have largely the same edaphic requirements, but usually large and dense, conforming to the life­ which obviously cannot withstand burning. form description given above on p. 86. The importance of grassl!1nd varies very much VIRUNGA VOLCANOES on different mountains, in connection with dif­ On Muhavura very little grassland was seen in the ferences in topography, soil, and climate, as well alpine belt; presumably the topography of this as in frequency of fires. On the moist and steep steep volcanic cone is inhospitable to such com­ slopes in the alpine belt on the Virunga Volcanoes munities. Only along the rim of the summit crater a

Acta Phytogeogr. Suec. 49 Tussock grassland 115

' .. "";:""';� �-�··.

··..

Fig. 95. Uganda, Kigezi Distr., Muhavura, part of the crater lake at the summit, 4100 m. The inner slope of the crater rim in the background is covered in its lower part by Alchemilla johnstonii scrub, in its upper part by tussock grassland. Photo 0. Hedberg 4.10.1948.

few patches of grassland were seen, dominated by Figs. 18, 19, 23). Also the intense grazing by cattle Festuca abyssinica (Fig. 95). On the other Virunga and buffalo in part of the erater may have similar Volcanoes grassland seems to be equally scarce in effects. The most important dominant in the alpine the alpine belt (cf. BuRTT, 1934; SCAETTA, 1934, grassland on Elgon appears to be Festuca pilgeri p. 301). (ssp. pilgeri), but also Andropogon amethystinus seems to be frequent. Less important tussock RUWENZORI grasses on this mountain are, int. al., Koeleria gra­ Also on Ruwenzori grassland is uncommon (HAu­ cilis var. convoluta and Agrostis gracilifolia. No MAN, l933a; ScAETTA, 1934, p. 304). Such grasses as analysis was made of this community. Deschampsia caespitosa (s.lat.), Festuca abyssinica, Poa ruwenzoriensis, Agrostis spp., and Anthoxan­ ABERDARE thum nivale may become locally dominant in small On the small alpine calottes of Aberdare tussock patches, particularly below Giant Lobelias and grassland appears to be one of the most important along lake shores (cf. Fig. 108), but no real tussock communities. Its main dominants seem to be grassland was observed in the alpine belt. Festuca pilgeri (ssp. pilgeri) and Koeleria gracilis ELGON var. convoluta. At least at lower levels (on open On Elgon tussock grassland is unquestionably one places in the ericaceous belt and the upper part of of the most important plant communities in the the montane forest belt) Andropogon amethystinus alpine belt, both inside the huge crater and on its may also become dominant. I have no detailed outside (Figs. 18, 19, 23, 89). Its large importance notes of this community from alpine level. The on this mountain is certainly largely due to the analysis published by FRIES & FRIES (1948, p. 51: frequent fires, which favour the spread of grassland "The Koeleria convoluta-Alchemilla cyclophylla at the expense of Alchemilla scrub (cf. pp. 36and Ill, Association") was evidently made a good deal below

Acta Phytogeogr. Suec. 49 116 Principal afroalpine plant communities

0.3 m

0

Fig. 96. Kenya, Mt Kenya, Teleki Valley, 4200 m. Line transect through a segment of tussock grassland on sloping ground (slope drawn approximately to scale). The densely hatched bodies above the figure 4 represent stones. The letters &ym­ bolize: (a) Festuca pilget·i (grazed young tussock); (b) Alchemilla johnstonii; (c) Festuca pilgeri (full-grown tussocks); (d) mosses (mainly Webera afrocruda); (e) Lobelia telekii. The figures l-5 represent different micro-habitats. Of the other plants recorded in this association Senecio purtschelleri appears to prefer microhabitats 2 and 3; Garduus platyphyllus 4, seedlings of Senecio keniodendron 4 and 5; seedlings of Lobelia telekii 4; Myosotis keniensis l, 4 and 5; Valeriana kilimandscharica ssp. kilimandscharica 3 and 5.

the lower limit of what is here called the alpine belt former association with a list of species have been (cf. HEDBERG, 1951, p. 183). Also on Aberdare provided by FRIES & FRIES (1948, p. 46: "The recurrent fires may be partly responsible for the Senecio Brassica-Festuca Association"). wide distribution of grassland at high level (cf. In the upper part of the alpine belt-above 3800- FRIES & FRIES, 1948, p. 50; and p. 25 above). 4000 m-much of the ground is covered by a drier type of tussock grassland, dominated also here by MT KENYA large tussocks of Festuca pilgeri, but containing no On the W slope of Mt Kenya tussock grassland such moisture-claiming plants as Senecio brassica seems to cover larger areas than any other plant and Carex monostachya. An analysis of an area of community in much of the alpine belt. In the lower lOO square metres from a representative stand of part of this belt, up to about 3800-4000 m, most of this community is given below, from the S slope of the ground is covered by a moist type of tussock the upper part of Teleki Valley at an altitude of grassland, dominated by Festuca pilgeri, and con­ 4200 m, and sloping about 20° towards N. Cf. Figs. taining large amounts of Senecio brassica, Carex 69 and 96, and HEDBERG, 1955, Pl. 3B. monostachya, Agrostis trachyphylla, etc. (Fig. 22). Dominant species of the field layer were Festuca The same association also occurs locally in moist pilgeri ssp. pilgeri and Alchemilla johnstonii, each valley bottoms etc. up to at least 4200 m. Where covering about one third of the ground. The first the ground becomes still moister this c?mmunity one forms tall and firm tussocks, whereas the latter merges almost imperceptibly into Carex mono­ forms carpets on the small terraces between the tus­ stachya bog (cf. p. 123 below). Some notes on the socks (Fig. 96). The other vascular plants found

Acta Phytogeogr. Suec. 49 Tussoclc grassland 117

Fig. 97. Carex runsso1·oensis bog fringing a stream in a valley bot­ tom. In the background open Senecio adnivctlis (coll.) woodland. Uganda, Ruwenzori, c. 3900 m. Photo A. Holm, March 1948. within the area were: Poa sp. (ster.), Cardamine and 60), whereas in the upper part of the same belt obliqua, Arabidopsis thaliana, Myosotis keniensis, they provide the dominant species (Fig. 61; cf. also Galium glaciale, Valeriana kilimandscharica ssp. HEDBERG, 1951, PI. VI A; and SALT, 1954, Photos 8 kilimandscharica, Lobelia telekii, Senecio kenioden­ and 9). At this altitude the vegetation is almost too dron (seedlings), Senecio purtschelleri, and Oarduus open to be called tussock grassland, however; it may platyphyllus. perhaps more adequately be termed "alpine desert" From the bottom layer were collected Brachythe­ (MEYER, 1900, p. 150 etc.-"alpine Wiiste"; cium ugandae, Bryum argenteum, Didymodon HEDBERG, 1951, p. 193 and Plate VI A; SALT, 1954, papillinervis, Encalypta ciliata, and Peltigera canina p. 402 and Photos 8-9). The most important tus­ (L.) WILLD. sock grasses of this community at about 4300 m The sloping soil was dammed into a series of altitude are Festuca pilgeri ssp. supina and F. small terraces by the big grass tussocks. All the abyssinica, Pentaschistis minor and P. borussica, bryophytes mentioned were recovered round the and Koeleria gracilis var. supina. Other vascular base of the grass tussocks or below the edge of the plants of the same association are Satureja kili­ terraces formed by them. On a small stone in this mandschari, Helichrysum cymosum ssp. fruticosum, community was found Orthotrichum rupestre. H. citrispinum var. hoehnelii, H. newii, Senecio A line transect through another segment of the telekii, and Euryops dacrydioides. On temporarily same community is shown in Fig. 96. moist patches were further collected Arabis alpina, Haplosciadium abyssinicum, Senecio schweinfurthii KILIMANJARO and S. meyeri-johannis, Oarduus keniensis, and In the alpine belt on at least the S slope of Nannoseris schimperi. Towards still higher levels, Kilimanjaro tussock grasses play an important above some 4500 m, the dominance is gradually role in the vegetation, especially on more or less taken over by Festuca kilimanjarica. The areas of deep and porous soils. In the lower part of this belt open soil between the grass tussocks often show (up to about 4300 m) they occur as a rule mainly as distinct signs of solifluction (Figs. 16, 60, 61; cf. subdominants in Helichrysum scrub (cf. Figs. 10 also HEDBERG, 1951, PI. VI A; and SALT, 1954,

Acta Phytogeogr Suec. 49 118 Principal afroalpine plant communities

Photo 9), and most of the small plants occurring in 4000 m). Also on this mountain the relevant vegeta­ this community are concentrated to "solifluction tion seems far too open to justify the denomina­ refuges" in the close proximity of grass tussocks, tion grassland; it is only an open alpine desert, Helichrysum shrubs, and stones (cf. p. 65 above). the vegetation of which is extremely poor in species. Thus in the lower part of the alpine belt vast MT MERU areas of the dry screes are covered by a community In the alpine belt on Mt Meru the sparse vegetation composed of 6 species only, namely Pentaschistis is also largely dominated by tussock grasses, the meruensis (dominant), P. mannii, Silene burchellii most important of which are Pentaschistis meruensis (subdominant), Astragalus venosus var., Senecio at lower levels (up to about 4000 m), and P. telekii, and Orepis sp. near 0. carbonaria (cf. HED­ mannii at higher levels (usually dominant above BERG, 1951, p. 194; 1957, p. 371).

CAREX BOGS AND RELATED COMMUNITIES

In such level or gently to moderately sloping loca­ VIRUNGA VOLCANOES lities in the lower part of the afroalpine belt (and in On Muha vura no Oarex bogs were seen by the the ericaceous belt) where the drainage is impeded author in the alpine and ericaceous belts-the to such an extent that the soil water reaches the regular cone of that mountain leaves little space for surface, the vegetation usually consists of strongly such communities. Such bogs have, however, been peat-forming communities dominated in the field reported from the ericaceous belt on Vissoke and layer by tussock-forming Carex species and in the Mikeno, and from both the ericaceous and alpine bottom layer by Sphagna and/or mosses. The peat belts on Karisimbi (SCAETTA, 1934, PI. XVIII, of these communities appears to be as a rule strongly Fig. 30; BuRTT, 1934, p. 155-157; RoBYNS, 1948, acid (cf. p. 28). In the terminology of TANSLEY p. 29-30). These bogs were said to consist of big (1949, p. 674) "wet acid peat formations" are called Oarex runssoroensis tussocks in a carpet of Sphag­ bogs or mosses, and the relevant afroalpine com­ num, harbouring also Lycopodium saururus, Al­ munities are here called Oarex bogs. Such bogs ap­ chemilla spp., Luzula abyssinica, and Luzula pear to be most important in the upper part of the johnstonii. ericaceous belt, and they do not as a rule occur very high up in the alpine belt. Their two main RUWENZORI fielrl. la.yer dominants are Oarex runssoroensis (on On the W side of Ruwenzori Oarex runssoroensis the Virunga Volcanoes, Ruwenzori, Elgon, Aber­ bogs seem to occur mainly in the alpine belt, be­ dare, and Mt Kenya), and 0. monostachya (on tween 3800 and 4250 m altitude (HAUMAN, 1933a, p. Elgon, Aberdare, Mt Kenya, and Kilimanjaro). 613). On the E slope they are common also in the One of the commonest sub-dominants is Alchemilla ericaceous belt, extending from 3300 m to 4300 m. johnstonii, especially in the marginal parts of the They are best developed on flat valley bottoms, bogs. Other, more conspicuo�s though less regularly sometimes fringing small lakes, but may also occur occurring sub-dominants are some Giant Lobelias on gently sloping ground (Figs. 97, 98, and 108; cf. of the Lobelia deckenii group (L. bequaertii on HAUMAN, 1935, PI. XI A; DE GRUNNE, 1935, Figs. Ruwcnzori, L. elgonensis on Elgon, L. sattimae on 39, 41, 93, 94; SYNGE, 1937, PI. 28 and 30; BERE & Aberdare, L. keniensis on Mt Kenya, and L. deckenii HICKS, 1946, PI. IV; HEDBERG, 1955, PI. 1; HEIN­ on Kilimanjaro). The bottom layer on some of the ZELIN & MoLLARET, 1956, PI. Ib and IVb). mountains is often dominated by a few Sphagnum Some notes on the vegetation of these "Marecages species, otherwise by mosses. a Oarex" on the W slope of Ruwenzori were given

Acta Phytogeog?". Suec. 49 Carex bogs 119

Fig. 98. Scattered specimens of Lobelia bequaertii at the margin of a Carex 1'unsso1·oensis bog in the middle part of the ericaceous belt. Note the cylinder of marcescent dry leaves around the stem of the non-flowering Lobelia specimen to the left. Uganda, Ruwenzori, Bujuku Valley near Bigo, 3400 m. Photo 0. Hedberg 21.3.1948. by HAUMAN (op. cit., p. 613-614) . On the S slope of bottom layer was partly dominated by Sphagnum the mountain I made a few brief notes from a pappeanum, and contained also, int. al., Breutelia Carex bog on the shore of the upper lake at Kitan­ subgnaphalea, Campylopus stramineus, Drepanocla­ dara, at about 4000 m altitude. The vegetation was dus uncinatus, Leptodontium tenerascens, and Philo­ dominated by low tussocks of Carex runssoroensis, notis alpicola. between which occurred carpets of Alchemilla It may also be pertinent to add here some more johnstonii, containing scattered specimens of Poa detailed notes from a Carex runssoroensis bog at schimperiana, Agrostis cf. gracilifolia, and Montia about 3400 m altitude near Bigo, on the E slope of fontana. In the marginal parts of the bog were also the mountain. This, of course, is in the ericaceous found some patches of Alchemilla argyrophylla ssp. belt, but practically all of the plants found there argyrophylloides and H elichrysum stuhlmannii. The have been encountered also in the alpine belt. The

Acta Phytogeogr. S1tec. 49 120 Principal afroalpine plant communities

Fig. 99. Small bog area covered by an almost pure Sphagnum carpet, surrounded by Oarex runssoroensis tussocks. In the background to the left a fairly dense Dendrosenecio forest; in the foreground a zone of Alchemilla scrub. Uganda, Ruwenzori, vV. Bukurungu Valley, 3950 m. Photo 0. Hedberg 4.4.1948.

dominant plant of the field layer was of course tered specimens of a few vascular plants, notably Carex runssoroensis, which f9rmed tussocks up to Poa schimperiana, Agrostis gracilifolia (particularly 1 m high and 0.6 m wide, spaced about 1 m apart, in small hollows), Luzula johnstonii, Oerastium octan­ and crowned by a large number of spreading culms drum, Oardamine obliqua, Alchemilla johnstonii, and up to 0.5 m long (see HEDBERG, 1955, Pl. 2A; cf. Senecio transmarinus var. sycephyllus. In the mar­ also Fig. 97 here). At the top of the tussocks oc­ ginal parts of the bog occurred a few colonies of curred a moss layer formed by Sphagnum davidii, Lobelia bequaertii (Fig. 98; cf. also, e.g., FrsHLOCK & Breutelia stuhlmannii, B. subgnaphalea, and Mylia HANCOCK, 1932, Pl. G; DE GRUNNE, 1937, Figs. 36 hedbergii, and on the middle portions of the tus­ and 41; JEANNEL, 1950, Pl. Ill), as well as stray socks grew a sparse bottom layer containing the specimens of Giant Senecios (Senecio adnivalis same Sphagnum species together with Campylopus s.lat.), Helichrysum stuhlmannii, and Alchemilla stramineus and Cephalozia fissa. argyrophylla ssp. argyrophylloides. The ground between the tussocks was very wet The Carex runssoroensis bogs at higher altitudes and covered by a spongy moss mat dominated by conform to the same pattern as described here, Sphagna (S. davidii, S. pappeanurn, and S. mada­ although the Carex tussocks tend to be lower and gassum) and Breutelia stuhlmannii, and containing the number of vascular plants and mosses tends also, int. al., Brachythecium spectabile, Breutelia to decline towards higher levels. Lobelia bequaertii, subgnaphalea, Campylopus stramineus, and Cyclo­ for instance, was not observed above 3950 m, nor dictyon borbonicum. In this moss layer grew scat- were any Sphagna. A little lower one may oc-

Acta Phytogeog1·. Suec. 49 Carex bogs 121

Fig. lOO. Uganda, Elgon, W slope of Koitobbos, c. 4000 m. In the foreground flowering specimens of Senecio barbatipes growing in a mat of Alchemilla elgonensis; in the background to the left a Carex 1·unssoroensis bog, behind which appears an open Senecio barbatipes woodland. Photo 0. Hedberg 11.5.1948. casionally find small bogs with almost pure Sphag­ ELGON num vegetation (Fig. 99). Also on Elgon Car ex runssoroensis bogs occur, being The peat formed by these Carex runssoroensis mainly confined to flat depressions with impeded bogs often becomes more than one m thick (cf. p. drainage (Fig. I 00). From a distance they look 27 above, and HEDBERG, 1955, p. 139 and Fig. I; rather similar to the bogs on Ruwenzori but differ, HEINZELIN & MoLLARET, 1956, Figs. 3 and 4). int. al., in the complete absence of Sp hagnum in the

As a rule it seems to be distinctly acid, with pH ­ bottom layer-I could not find a single specimen of values between 3.5 and 4.5 (cf. p. 28 above). Sphagnum on Elgon. The following notes were made Nevertheless it contains a rich fauna of earthworms, in a bog at 3750 m altitude on the N slope of Koi­ which eat the Sp hagna and mosses and transform tobbos. them into some muddy and more or less amorphous The dominant plant of the field layer was Carex substance, while the firm Carex peat seems to be runssoroensis, forming dense and hard tussocks left almost intact (HEDBERG, 1955, p. 139). about 3-5 dm high and fairly densely spaced (cf.

Acta Phytogeogr. Suec. 49 122 Principal afroalpine plant communities

Fig. 101. Flowering specimen of Lobelia sattimae in a Carex monostachya bog. Note the claw-marks of sunbirds in its bracts. In the lower left corner Alchemilla johnstonii, in the background mist. Kenya, Aberdare, N slope of Kinangop, 3600 m. Photo 0. Hedberg 17.7.1948.

Fig. 100). On these tussocks grew a number of specimens of Subularia monticola, Orassula gran­ mosses, the most important of which was Oam­ vikii, Limosella africana, Oallitriche stagnalis, and pylopus stramineus, with smaller quantities of Oardamine obliqua. Especially in the marginal parts Antitrichia curtipendula, Hylocomium splendens, there occurred numerous specimens of Lobelia Hypnum cupressiforme, and Sematophyllum el­ elgonensis and some Lycopodium saururus. The gonense. In the drier parts of the bog the vegetation sparse and discontinuous bottom layer between the between the Oarex tussocks was dominated by tussocks contained mainly Breutelia stricticaulis, Alchemilla johnstonii, which in the moister parts was B. subgnaphalea, Drepanocladus uncinatus, and

confined to the basal portions of the tussocks, while Marchantia polymorpha. -Similar bogs were seen the intervening wet ground harboured only stray on Elgon up to at least 4100 m altitude.

Acta Phytogeogr. Suec. 49 Garex bogs 123

20

10

Fig. 102. Schematic profile through a Carex monostachya-Agrostis trachyphylla community on gently sloping flushed ground. The letters stand for: (a) Romulea keniensis; (b) Carex monostachya; (c) mosses; (d) Haplocarpha rueppellii; (e) Festuca pilgeri; (f) Haplosciadium abyssiniwm; (g) Ag1·ostis trachyphylla; (h) Ranunculus oreophytus. Kenya, Mt Kenya,

Teleki Valley, 4200 m.

ABERDARE The commonest dominant of the Garex bogs in the TABLE 10. Carex monostachya - Agrostis alpine belt on Aberdare would seem to be Garex trachyphylla community. monostachya (F . G. runssomensis ig 101), although Kenya, Mt Kenya, Teleki Valley, 4200 m, on gently sloping has also been collected here. Other common plants flushed ground, analyzed 13.8. 1948. Square 2 was obtained in this community are Senecio brassiciformis and on slightly drier ground than the other two; square size Lobelia sattimae (Fig. 101), and two of the most l m.2 (For further explanation see p. 132.) important dominants in the bottom layer are Sphagnum pappeanum and Rhacocarpus humboldtii. Square number 2 3 No detailed analysis was made of this association. A list of species from a variant of the same com­ Agrostis leptophylla 3 Agrostis trachyphylla 2 munity at slightly lower level has been published 4 4 Alchemilla johnstonii l 1 by FRIES & FRIES (1948, p. 52: "The Garex mono­ Carex monostachya 3 2 2 stachya Association''). Crassula granvikii l F�stuca pilgeri ssp. pilgeri 2 MT KENYA H aplocarpha rueppellii 5 On Mt Kenya the most important dominant on Haplosciadium abyssim:cum 1 4 Luzula abyssinica boggy ground at high level is Garex monostachya. Pentaschistis mino1· In the lower part of the alpine belt this species forms Ranunculus oreophytus 4 large tussocks, dominating the vegetation over Romulea keniensis " large areas. Lobelia keniensis and Senecio brassica campylium hispidulum var. are conspicuous subdominants of this community, sommerfeltii' ,a Campylopus stramineus of which no detailed notes are available. Ceratodon purpureus In the upper part of the alpine belt Carex mono­ Leptodontium tenerascens .stachya is not tussock-forming but occurs together Polytrichum piliferum with Agrostis trachyphylla as dominant or sub­ Cladonia cfr. pyxidata dominant on gently sloping flushed ground with high ground water level and more or less distinct a See footnote on p. 102.

Acta Phytogeogr. Suec. 49 124 Principal afroalpine plant communities

Fig. 103. Kenya, Mt Kenya, Head of Teleki Valley, with the twin peaks in the background. The narrow pillars on the valley bottom in the foreground are inflorescences of Lobelia keniensis (cf. p. 60). Photo A. Holm July 1948. The thermo­ grams in Fig. 8 were recorded on the other side of the valley, at the little promontory extending into the valley bottom at the middle of the photo. signs of solifluction. An analysis of three 1-m2 of the alpine belt on Mt Kenya there occurs a squares from such a community is given in Table related community, dominated by Lobelia keniensis 10. In the same community were also seen Lobelia (cf. Fig. l 03). A line transect through a stand of this keniensis, Sagina cf. afroalpina, and Scapania community is shown in Fig. 104:; cf. also Figs. 34 hedbergii. A schematic profile through a segment of (background), 49, 56, 57, and 81. The same stand this association is shown in Fig. 102. It is evidently (at an altitude of 4200 m in the bottom of the upper not peat-forming. part of Teleki Valley) was analyzed within a sample

On flat ground with a high soil water level in the square of 10 x 10 m, where the following notes bottom of glaciated valleys etc. in the upper part were made.

Acta Phytogeog1·. Suec. 49 Carex bog8 125

1 m

· :F'ig. 104. Schematic profile through a segment of Lobelia keniensis community. The letters stand for: (a) Lobelia keniensis; {b) mummified inflorescences of the same, serving as mouse nests; (c) Alchemilla johnstonii; (d) Festuca pilgeri; (e) Agrostis trachyphylla, closely trimmed by mice; (f) Carex monostachya; (g) Agrostis trachyphylla, flowering; (h) Haplosciadium abyssinicum; (i) Carduus platyphyllus. Kenya, Mt Kenya, Teleki Valley, 4200 m.

Lobelia keniensis: dominant, with 5 flowering Oampylopus denticuspes, Didymodon rufus, Lepto­ spikes, 32 spikes that had ceased flowering but dontium johannis-meyeri, Meesea kenyae, Tortula were still erect, about 30 fallen and decaying old cavallii, and Oladonia cf. pyxidata. flower spikes, c. 50 non-flowering rosettes, and c. The ground was rather uneven with tussocks 250 small rosettes. and small hollows, displaying distinct signs of frost Senecio keniodendron: 1 half-grown specimen heaving. This community was populated by con­ which had not yet flowered for the first time, and siderable numbers of small rodents, which grazed c. 50 young specimens with leaf rosettes only be­ most of the grasses very closely, making some of tween 5 and 35 cm wide. them rather difficult to identify. In the same com­ Apart from these the field layer contained the munity there also lived at least two different kinds following species: Festuca pilgeri (scattered tus­ of nectarinides, which seemed to feed largely in the socks), Agrostis sclerophylla (a few cushions), flowers of Lobelia keniensis, causing the bracts of Agrostis trachyphylla (subdominant), Oarex mono­ older flowers to become completely worn out by stachya (scattered specimens, but no tussocks), their claws (see p. 38 above and Figs. 27, 81, and Luzula abyssinica (a few specimens), Oerastium 104; cf. also HEDBERG, 1957, PI. 6B). This com­ octandrum var. adnivale (a few specimens), Sagina munity occurs up to altitudes of at least 4300 m. cf. afroalpina (a few specimens), Ranunculus KILIMANJARO oreophytus (scattered specimens), Oardamine obliqua {scattered specimens), Alchemilla johnstonii (sub­ Because of the dryness of the upper part of Kili­ dominant), Haplosciadium abyssinicum (a few manjaro, bogs and related communities tend to be specimens), and Myosotis keniensis (a few cushions). less prominent here than on most of the other high The bottom layer was rather discontinuous. Its mountains. Carex bogs, dominated by tussock­ most important species in the wetter portion of the forming Oarex monostachya (Fig. 105) are on most square was A ulacomnium turgidum var. papillosum, parts of the mountain largely confined to the upper .and in the slightly drier parts "Oampylium his­ part of the ericaceous belt, barely reaching into pidulum var. sommerfeltii'',l intermingled with, the lower part of the alpine belt. On the SE slope int. al., Bryum argenteum, Bryum ellipsifolium, of the mountain they are confined to narrow sloping strips on flushed areas along small streams (Figs.

1 See footnote on p. l 02. 86 and 106; cf. HEDBERG, 1951, PI. IIA). On the

Acta PhytogeogT. Suec. 49 126 Principal afroalpine plant communities

Fig. 105. Oarex monostnchyn tus­ socks in a Eloping bog. Tanganyika, Kilimanjaro, S slope above Peter's Hut in the upper part of the eri­ caceous belt, 3950 m. Photo 0. Hedberg 27.6.1948.

Fig. 106. Sloping Oarex monostachya bog in the upper part of the ericaceous belt. The slopes along the small stream in th. foreground are covered by Alchemilla }ohnstonii scrub. Tanganyika, Kilimanjaro , S slope above Peter's Hut, 3950 me Photo 0. Hedberg 23.6.1948.

Acta Phytogeogr. Suec. 49 Carex bogs 127

Fig. 107. Tanganyika, Kilimanjaro, S part of the Saddle between Kibo and Mawenzi, 4350 m. Small Carex bog, nourished by subsoil water and appearing as an oasis in the desert­ like vegetation of the Saddle (cf. p. 29). Photo 0. Hedberg 23.6. 1948.

Shira Plateau to the West they seem to be much described above mainly in the presence of De­ more important, however; according to SALT schampsia caespitosa (s.lat.) and the absen�e, int. (1954, p. 382 and Fig. 2) "�edge and grass com­ al., of Giant Senecios and Giant Lobelias. munity", dominated by Carex monostachya and In the dry "alpine desert" of the Saddle between Deschampsia sp . (=D. caespitosa, s.lat.), cover here Kibo and Mawenzi there occur, at altitudes of an area of some 5 sq. km on flat ground. about 4250-4400 m, a few small basins with im­ The following notes were made from a strip of peded drainage, nourished by subsoil water, where distinctly sloping (c. 10°-15°S) Carex bog at an the high water table has caused the appearance of altitude of 3800 m on the SE slope of the mountain a particular type of bogs (Fig. 107). Part of such a (a few hundred metres W of Peter's Hut, in the bog at an altitude of 4350 m in the S part of the ericaceous belt; cf. Fig. 86). The vegetation was saddle was occupied by a fairly high vegetation, entirely dominated by tussocks of Carex mono­ dominated by tussock-forming Carex monostachya. stachya, about 0.4 m tall, growing fairly densely Between the tussocks occurred stray specimens of spaced on a loose and mull-like peat (Fig. 105). The M ontia fontana, Ranunculus volkensii, Crassula discontinuous bottom layer between the tussocks granvikii, and Limosella ajricana. Other parts of the contained no Sphagnum but only a few mosses, same bog had a shorter vegetation dominated by notably Breutelia subgnaphalea, Campylopus alpi­ Agrostis trachyphylla and an undescribed species colus, Ceratodon purpureus, Drepanocladus unci­ of Scirpus (cf. HEDBERG, 1957, p. 262), and con­ natus, and Tortula cavallii. Among these grew some taining also Keniochloa chionogeiton var. oread_es, scattered specimens of Cardamine obliqua, Carda­ M ontia fontana, Ranunculus volkensii, Ranunculus mine hirsuta, and Arabidopsis thaliana. More con­ oreophytus, Ranunculus stagnalis, Alchemilla micro­ spicuous though less frequent vascular plants of the betula, Haplosciadium abyssinicum, and Galium same community were stray specimens of Giant glaciale. In other bogs of the same type were also Senecios and Giant Lobelias (Senecio cottonii and found, int. al., Sagina afroalpina, Subularia monti­ Lobelia deckenii, resp.; cf. Fig. 86). Along the edges cola, and Cardamine obliqua. of the bog occurred a few patches of Alchemilla johnstonii and A. argyrophylla ssp. argyrophylla MT MERU (Fig. 106; cf. p. 113 above). The "Sedge and grass community" described by The upper parts of Mt Meru are very dry, and nei­ SALT (lac. cit.) from level ground on the Shira ther water plants nor bogs are known from the al­ Plateau seems to deviate from the type of bog pine belt.

Acta Phytogeogr. Suec. 49 128 Principal afroalpine plant communities

Fig. 108. Vegetation zonation along a lake shore in the alpine belt, with a first zone of Limosella ajr icana just above the

.3915m. Photo 0. Hedberg 8.4.1948.

OTHER PLANT COMMUNITIES

The five types of plant communities described along streamsides (Fig. 109), in caves, on loose screes, .above are without any doubt the most important etc., but these as a rule cover relatively small areas, ones in the alpine belt of the high East African and it would not be of much use to reproduce here mountains. Several other types do of course occur, my fragmentary notes from them. for instance in small lakes and ponds (Fig. 108),

Acta Phytogeog1·. S?tec. 49 Vicarious plant communities 129

Fig. 109. Tanganyika, Kilima­ njaro, W slope of Mawenzi, 4600 m. Patch of closed vegetation along a small brook, dominated by Kenio­ chloa chionogeiton var. oreades, and containing also, int. al. , Montia jontana, Ranunculus stagnalis, Sa­ gina cfr. ajroalpina, Subularia monticola, and Haplosciadium abys­ sinicum, etc. The light spots on the ground were solid ice (the photo­ graph taken about noon). Photo 0. Hedberg 19.6.1948.

VICARIOUS PLANT COMMUNITIES

The high East African mountains have long been relevant mountains and have aimed at a direct famous for the great number of geographically comparison they may perhaps be suitable to de­ vicarious taxa occurring on their upper parts (cf. monstrate to what extent vicarious communities FRIES & FRIES, 1922a, 1922b, 1948; HAUMAN, do occur in the alpine belt on these mountains. 1933b, 1935; HAUMAN & BALLE, 1934; JEANNEL, The most expedient example is available re­ 1950; GOOD, 1953; HEDBERG, 1957, 1958, 1961). It garding the Alchemilla scrub, where Tables 7-9 was furthermore pointed out long ago by TH. C. E. allow a detailed collation of comparable samples FRIES in (unpublished) lectures that they also from three different mountains, viz. Ruwenzori, provide excellent examples of vicarious plant Elgon, and Mt Kenya. The three associations de­ communities (cf. Du RIETZ, 1924, p. 106). Some of scribed in these tables differ not only in their most the vegetation studies from Aberdare and Mt important dominants (Alchemilla argyrophylla ssp. Kenya on which TH. C. E. FRIES founded this argyrophylloides on Ruwenzori, A. elgonensis on opinion have been published after his death by his Elgon, and A. argyrophylla ssp. argyrophylla on brother and collaborator R. E. FRIES (FRIES & Mt Kenya), but also in several other respects. FRIES, 1948). Because of the short time spent by These other differences cannot be due only to dif­ these authors in the alpine belt (sensu mihi) most ferences in climate or in the floras available on the of their descriptions of high altitude associations relevant mountains but must in part have other are rather laconic, however, and very little is said causes. Thus Veronica glandulosa occurs on all the about which of the associations described should three mountains concerned (although on Ruwen­ be considered mutually vicarious. The classification zori it has so far been found only in the ericaceous of the species forming these associations has also belt), but it occurs as a more or less regular con­ lately been modified to a large extent (HEDBERG, stituent of the Alchemilla scrub only on Elgon. 1957). Similarly Valeriana kilimandscharica coll. occurs The vegetation descriptions provided in the ear­ both on Elgon and Mt Kenya, but it appears as a lier part of this chapter are obviously also rather regular subordinate species of Alchemilla scrub only fragmentary, but since they comprise all of the on Elgon (and Aberdare) but not on Mt Kenya (cf.

9-641464 Hedberg .Acta Phy togeogr. S�tec. 49 130 Principal afroalpine plant communities

Fig. 94). There are also important differences in the splitting of an ancestral population into two or bottom layer: Breutelia subgnaphalea is the most more segments, followed by a parallel differentia­ important moss in the Ruwenzori association but tion of all or most of its species into groups of was not found in the other two; Breutelia stricti­ vicarious taxa. The various dominant Alchemilla caulis plays the same role in the Elgon association (Senecio) species must obviously be expected to have but is absent from the others; and Tortula cavalli, descended from a common ancestor, which pre­ which is the dominant moss of the Mt Kenya as­ sumably formed Alchemilla scrub (Dendrosenecio sociation, appears to be absent from the other two. woodland) much as its descendants do. But there In view of the large climatic (and edaphic) dif­ is no evidence for a more continuous earlier dis­ ferences between the mountains concerned, the tribution of the afroalpine flora-as far as the avail­ three relevant Alchemilla dominants may conceiv­ able evidence goes its species must be expected to ably have slightly divergent ecological preferences. have dispersed one by one in "jumps" from one And, to judge from their "social behaviour", mountain to another, as they became inhabitable different mountain populations of some of the for an afroalpine flora (HEDBERG, 1961). Therefore subordinate species (e.g. Veronica glandulosa and the associations can hardly have spread as such, Valeriana kilimandscharica coli.) must obviously but must have been synthesized anew on each be expected to be ecologically differentiated. mountain from the. particular stock of species Equally good examples of vicarious plant com­ becoming available there. This may partly explain munities are furnished by the afroalpine .Den­ the large floristic differences obtaining between drosenecio woodlands. As demonstrated by the de­ those vicarious associations. scriptions given above on p. 96-103 these associa­ The vicarious communities of the East African 0 tions differ not only in the dominant species of Den­ mountains were considered by Du RIETZ (1924, p. drosenecio but also in the composition of the field 106) to constitute the best examples known of true and bottom layers to such an extent that they form vicarious associations, "deren Arten fast alle echte markedly distinct vicarious associations. Vikariisten sind". The description given above may Both the Alchemilla associations and the Den­ demonstrate that this judgment was slightly drosenecio associations may evidently be regarded exaggerated-the dominants may sometimes be as "true vicarious associations" in the sense of Du almost the only vicarious taxa holding those as­ RIETZ (1924, p. 106) and CAIN (1944, p. 272), since sociations together. In view of the probable mode they (or at least their dominant species) must ob­ of differentiation outlined above it might even be viously have developed from a common stock. questioned vvhether these associations should be This cannot be claimed with the same confidence classified as "true vicariads" at all. In actual fact for the other vegetation types treated in the earlier the distinction between "true" and "false" vi­ part of this chapter (Helichrysum scrub, tussock carious associations appears to be as a rule difficult grassland, and Carex bogs), because the resemblance if not impossible to maintain-in the same way as between most of the respective dominants may be the distinction between "true" and "false" vi­ due more to parallel adaptations to an extreme en­ carious taxa (HEDBERG, 1957, p. 372). The more vironment than to very close relationships (as in the we learn about a group of vicarious plant com­ case of Helichrysum stuhlmannii-H. citrispinum, munities, the more difficult it becomes to refer or Carex runssoroensis-0. monostachya). These them to either of these categories. As a rule we groups of associations may therefore correspond know far too little both about the detailed taxo­ better to "false vicarious associations" according nomy of the plants concerned, and about the his­ to the same terminology. tory of their arsociations, to decide whether they It must not of course be expected that the dif­ consist of numerous (true) vicarious species, and ferentiation of such vicarious communities as the have developed from a common stock. A classifica­ Alchemilla (or Dendrosenecio) associations of the tion based on subjective inferences in these matters East African mountains has involved simply a will be of little use to phytogeography.

Acta PhytogeogL Suec. 49 RESUME

Dans les regions eleves de l'Afrique orienta.le, se d'humus dans la partie superieure de la zone alpine rencontrent c;a et la de hautes montagnes isolees, et le plus souvent a fine texture avec un pour­ dont la vegetation est disposee en zones tres nettes. centage eleve d'humus dans la partie inferieure de La flore des etages les plus eleves, denommee flore la zone alpine. Par suite de la frequence des gelees afro-alpine, a intrigue les botanistes, notamment nocturnes, les sols nus, au moins dans la partie par suite du gigantisme montre par certaines superieure de la zone alpine, sont soumis, sur cer­ especes comme les Senecio et les Lobelia geants. Ce taines montagnes, a une solifluxion prononcee livre traite principalement de l'ecologie de la zone provoquant, sur de vastes etendues, particuliere­ afro-alpine ainsi que des formes biologiques parti­ ment au Kilimandjaro et au Mont Kenya, la forma­ culieres des especes qu'on y rencontre. tion de ce qu'on appelle les mobilideserta. Min de se faire une idee des adaptations des Etudiee plus en detail, la flore afro-alpine montre organismes a leur milieu, il convenait d'etudier tout une serie d'adaptations morphologiques, chacune d'abord ce dernier, notamment le climat et le sol visible sur plusieurs especes appartenant a des (Ch. I). Bien que differant considerablement d'une groupes differents (Ch. II). C'est ainsi que la piu­ montagne a l'autre, le climat afro-alpin se CClorac­ part des graminees des hautes altitudes (et cer­ terise essentiellement par !'absence de variations taines cyperacees) forment de grosses touffes denses saisonnieres de temperature et, dans une certaine constituant un excellent isolement thermique pour mesure, de precipitation, alors que les variations leur partie centrale, que plusieurs especes apparte­ diurnes de temperature sont tres marquees, pas­ nant a des genres differents forment d'enormes ro­ sant du gel au degel et l'inverse de nombreux ou settes foliaires denses constituant egalement un bon meme la plupart des jours durant toute l'annee. La isolement thermique pour leur jeune pousse et que pression atmospherique basse, la rarete des vents de nombreuses plantes font preuve de xeromorphie violents ainsi que !'insolation et le rayonnement in­ prononcee (pubescence dense; feuilles epaisses; pe­ tenses constituent d'autres facteurs caracteristiques. tite surface foliaire; etc.). On observe egalement une Sur chaque montagne, s'observe une serie de micro­ tendance generalisee a la reduction de la longeur climats differents, d'autant plus divers que les des entre-nceuds. montagnes sont seches. Certains des facteurs les Un certain nombre d'especes presentant les plus defavorables de ce milieu doivent etre les gelees adaptations mentionnees ci-dessus appartiennent a nocturnes periodiques ainsi que le rechauffement des formes biologiques bien definies, telles les rapide du sol et de la vegetation le matin sous un plantes a rosette geante, les graminees en touffes soleil tres intense. L'equivalent de ce climat afro­ serrees, les especes acaules en rosette, les plantes en alpin se rencontre sur d'autres montagnes equa­ coussinets et les arbustes sclerophylles, toutes ces toriales, comme les Andes dans le nord de l' Ameri­ formes paraissant bien adaptees au milieu afro­ que du Sud et, dans une certaine mesure, dans les alpin (Ch. Ill). A ces formes biologiques appartien­ lles subantarctiques. nent environ 45 % des especes vasculaires de la ve­ Les sols afro-alpins n'ont ete que tres peu etu­ ritable flore afro-alpine, notamment toutes les dies a ce jour. En general, ils semblent etre nette­ especes dominantes des principaux types de vege­ ment acides, tendant a etre souvent a texture tation. Exactement les memes formes biologiques grossiere et poreux avec un faible pourcentage se rencontrent, mais dans d'autres genres, dans la

Acta Phytogeogr. Suec. 49 132 Resume vegetation des paramo et des puna des Andes sud­ heures de la journee avant que les parties gelees americaines. Les formes biologiques traditionelles des plantes n'aient eu le temps de degeler. Les 5 de Raunkiaer ne peuvent guere etre utilisees dans formes biologiques decrites a-dessus representent, de ce milieu ou la << saison defavorable >> n'est ni l'hi­ toute evidence, des moyens differents de resoudre ver, ni la saison seche, mais les nuits. ce pro bleme. La signification des formes biologiques afro­ La plus grand partie de la vegetation afro-alpine alpines speciales apparait particulierement nette si est constitue par 5 types differents de communau­ on considere les formes de chaque espece du point de tes vegetales : la foret a Dendrosenecio, les fourres a vue de leur economie en eau (Ch. IV). Le maintien H elichryS'IJ,m, les fourres a Alchemilla, les prairies du bilan d'eau va evidemment exiger des moyens a grosses touffes denses et les marais a Carex. extraordinaires dans un milieu ou !'absorption de L'aspect general de ces groupements sur chacune l'eau du sol et le passage de l'eau a travers les ra­ des montagnes est brievement decrit dans le Ch. V, cines sont ralentis par de basses temperatures et ou figurent egalement quelques analyses simples de parfois meme bloques par le gel, ou le passage de la vegetation. L'existence de groupements vege­ l'eau a travers toutes les tiges non protegees est taux vicariants est prouvee par des exemples four­ regulierement bloque par le gel la plupart des nuits nis par les fourres a Alchemilla et les forets a Dendra­ et ou !'insolation devient intense et le degre d'hu­ senecio. midite atmospherique baisse dans les premieres

Explanation to Tables 7-10 The vegetation was analysed in a number . of sample

squares of 1 x 1 m. Within each of these the degree of cover of each species found was estimated according to the HuLT-SERNANDER scale (5 =more than half of the sample area covered; 4 =between 1/4 and 1/2 of the area covered; 3 =between 1/8 and 1/4 of the area covered; 2 =between 1/16 and 1/8 of the area covered; 1 =less than 1/16 of the area covered; cf. Du RIETZ, 1930, p. 396). The vascular plants have been determined by the author; the bryophytes were determined by specialists in samples brought by the author. Each record of a bryophyte in these tables cor­ responds to a collection, quoted in .ARNELL, 1956; PoTIER DE LA V ARDE, 1955; or VAND EN BERGHEN, 1953, resp. Within each layer (field layer, bottom layer) the species are enumerated in alphabetical order, bryophytes and lichens being kept apart. Obviously no claim can be made to completeness as regards the smallest bryophytes­ especially some of the liverworts were probably incom­ pletely sampled.

Acta Phytogeog1-. Suec. 49 REFERENCES

ALLAN, H. H. 1938. A consideration of the "biological spectra" of New Zealand. J. Ecol. 25, ll6-152. ALLUAUD, CH. & JEANNEL, R. 1914. Le Mont Kenya en Afrique orientale anglaise. Rev. gen. Sci. 25 (13), 639-644. Paris. ANDERSSON, J. G. 1906. Solifluction, a component of subaerial denudation. J. Geol. 14. ANDERSSON, G. & BIRGER, S. 1912. Den norrlandska florans geografiska fordelning och invandringshistoria. Norrl. Handbibl. 5. Uppsala & Stockholm. ANONYMUS. 1948 a. A note on the climate of East Africa, with special reference to rainfall and temperature. E. Afr. Meteor. Dept. Pamphlet Ser. (1). Nairobi. ANONYMUS. 1948 b. A note on the climate of East Africa, with special reference to flying conditions. Ibid. (4). ARNELL, S. 1956. Hepaticae collected by 0. Hedberg et al. on the East African Mountains. Ark. Bot. Ser. 2, 3 pp. 517-562. BERE, R. M. & HICKS, P. H. 1946. Ruwenzori. Uganda Journal 10 (2), 84-96. BEws, J. W. 1925. Plant Forms and their Evolution in South Africa. London. BoNNIER, G. 1890. Cultures experimentales dans les Alpes et dan s les Pyrenees. Rev. gen. Bot. 2, 513-546. BRASS, L. J. 1941. The 1938-39 expedition to the Snow Mountains, Netherlands New Guinea. J. Arnold Arbor. 22, 271-342, PI. I-VII. BRAUN-BLANQUET, J. 1928. Pflanzensoziologie. Bioi. Studienbucher VII. Berlin . BRYK, F. 1927. Uber die Curculioniden-Fauna des Mmmt Elgongipfels. Societas entomol. 42 (10), 38-39. Stutt­ gart. BURGEON, L. 1937. La Faune. In DE GRUNNE, X. et al., Le Ruwenzori, Mission scientifique beige 1932, p. 261- 273. BuRTT, B. D. 1934. A botanical reconnaissance in the Virunga Volcanoes of Kigezi Ruanda, Kivu. Kew Bull. 1934, 145-165. CABRERA, A. L. 1958. La Vegetacion de la Republica Argentina. VI. La Vegetacion de la Puna Argentina. Rev. Invest. Agric. XI:4, 317-412, Pls. I-XVI. Buenos Aires. CAIN, S. A. 1944. Foundations of Plant Geography. New York and London. CHIOVENDA, E. & CoRTESI, F. 1909. Angiospermae. In LuiGI AMADEO DI SAVOIA, Il Ruwenzori, Parte scientifica, Vol. I, Zoologia, Botanica, pp . 433-473, Pl. 26-65. Milano. CLAUSEN, J., KEcK, D. D., & HIESEY, W. M. 1940. Experimental studies on the nature of species. I. Effects of varied environments on Western North American plants. Carn. Inst. Wash. Publ. 520. CLEMENTS, F. E. 1920. Plant indicators: the relation of plant communities to process and practise. Ibid. 290. CocKAYNE, L. 1921. The Vegetation of New Zealand (1st ed.). Die Vegetation der Erde, XIV. Leipzig. COGNETTI DE MARTIIS, L. 1909. Lombrichi del Ruwenzori e dell'Uganda. In LuiGI AMADEO DI SAvOIA, Il Ruwenzori, Parte scientifica, Vol. I, Zoologia, Botanica, pp. 359-414. Milano. CosTIN, A. B. 1957. The high mountain vegetation of Australia. Aust. J. Bot. 5, 173-189. CosTIN, A. B., WIMBUSH, D. J., KERR, D., & GAY, L. W. 1959. Studies in catchment hydrology in the Australian Alps. I. Div. Plant Industry Techn. Paper 13. Melbourne. CoTTON, A. D. 1944. The megaphytic habit in the tree Senecios and other genera. Proc. Linn. Soc. Lond. Sess . 156 (2), 158-168. CRUNDWELL, A. C. & NYHOLM, E. 1962. A study of Campylium hispidulum and related species. Trans. Brit. bryol. Soc. 4 (2), 194-200. CuATRECASAS, J. 1934. Observaciones geohotanicas en Colombia. Trab. Mus. Nacional Ciencias Naturales, Ser. Bot., 27. Madrid. -- 1958. Aspectos de la vegetacion natural de Colombia. Rev. Acad. Colo:nbiana Cienc. Exact. Vol. 10 (40), 221-268, Pis. I-XXXV. CUFODONTIS, G. 1955. Kritisches Referat uber die Bedeutung der Termiten fur das Verstandnis der afrikanischen Savannen-Vegetation. Ost. bot. Z. 102, 501-519.

Acta Phytogeog1·. Suec. 49 134 References

DAHL, E. 1956. Rondane. Mountain vegetation in South Norway and its relation to the environment. Skr. Norske Videnskaps-Akad. Oslo I. Mat.--Naturv. Kl. 1956:3. Oslo. DARWIN, CH. 1859. On the Origin of Species by Means of Natural Selection, or the Preservation of Favoured Races in the Struggle for Life. London. DAUBENMIRE, R. F. 1943. Vegetational zonation in the Rocky Mountains. Bot. Rev. 9, 325-393. -- 1947. Plants and Environment. A textbook of plant autecology. New York. DAWE, M. T. 1906. Notes on the Vegetation of Buddu and the Western and Nile Provinces of the Uganda Protectorate. J. Linn. Soc. (Bot.) 37, 533-544. DIELS, L. 1934. Die Paramos der aquatorialen Hoch-Anden. S.B. preuss. Akad. Wiss. 1934, 57-68. -- 1937. Beitrage zur Kenntnis der Vegetation und Flora von Ecuador. Bibl. bot. 29 ( 116), 1-190. Stuttgart. DRUDE, 0. 1913. Die Okologie der Pflanzen. Die Wissenschaft, Bd 50. Braunschweig. -- 1914. Die Stellung der physiognomischen Okologie. Bot. Jb. 52, Beibl. 115, 8-13. Du RIETZ, G. E. 1921. Zur methodologischen Grundlage der modernen Pflanzensoziologie. Uppsala (Diss. ). 1924. Studien iiber die Vegetation der Alpen, mit derjenigen Skandinaviens verglichen. Veroff. geobot. Inst. Rubel 1, 31-138. Zurich. 1930. Vegetationsforschung auf soziationsanalytischer Grundlage. ABDERHALDEN, Handb. biol. Arbeits­ meth. XI : 5. Berlin und Wien. -- 1931. Life-forms of Terrestrial Flowering Plants. I. Acta phytogeog. suec. 3 (1) . -- 1954. Sydvaxtberg. Svensk bot. Tidskr. 48, 174-187. DuTT, B. K. & GuHA-THAKURTA, A. 1938. Effect of reduced atmospheric pressure on the growth of plants. Trans. Bose Res. Inst. Calcutta, 12, 1-9, Pl. 1-3. East African J.Vl eteorological Department, Summary of Rainfall, for the years 1946-1958. Nairobi. EDWARDS, F. W. & TAYLOR, G. 1935. British Museum Expedition to East Africa. I. Nat. Hist. Mag. 5, 165- 186. London. ELLENBERG, H. 1959. Uber den Wasserhaushalt tropischer Nebeloasen in der Kustenwuste Perus. Ber. geobot. Forschungs-Inst. Rubel 1958, 47-74. ENGLER, A. 1925. Die Pflanzenwelt Afrikas V: 1. Die Vegetation der Erde IX. Leipzig. EsPINOSA, R. 1932. Okologische Studien "liber Kordillerenpflanzen. Bot. Jb. 65, 120-211. FISHLOCK, C. W. L. & HANCOCK, G. L. R. 1933. Notes on the flora and fauna of Ruwenzori with special re- ference to the Bujuku Valley. J. E. Afr. Nat. Hist. Soc. 44, 205-229. Nairobi. FLUCKIGER, 0. 1934. Schuttstrukturen am Kilimandscharo . Peterm. Geogr. Mitt. 1934. FosBERG, F. R. 1944. El Paramo de Sumapaz, Colombia. J. N.Y. bot. Gdn 45, 226-234. FRANCHETTI, GrovANNA. 1958. Adumbratio Florae Aethiopicae 7. Cruciferae (Trib. Lepidieae, Euclidieae, Sisymbrieae). Webbia 14, 1 6 1-2 11. FRIES, R. E. 1923. Revision der tropisch-afrikanischen Carduus-Arten. Acta Hort. berg. 8, 11-38. -- 1925. Vegetationsbilder von den Kenia- und Aberdare-Bergen (Ostafrika). G. KARSTEN & H. SCHENCK, Vegetationsbilder, 16 (7 ) . Jena. FRIES, R. E. & FRIES, TH. C. E. 1922a. Uber die Riesen-Senecion9n der afrikanischen Hochgebirge. Svensk bot. Tidskr. 16, 321-340. 1922 b. Die Riesen-Lobelien Afrikas. Ibid. 16, 383-416. 1948. Phytogeographical researches on Mt. Kenya and Mt. Aberdare, British East Africa. K. svenska Vetensk.-Akad. Handl. Ill, 25 (5) . Stockholm. FRIES, TH. c. E. 1913. Botanische Untersuchungen im nordlichsten Schweden. Uppsala (Diss. ). GAMS, H. 1918. Prinzipienfragen der Vegetationsforschung. Vierteljahrsschr. N aturf. Ges. Zurich 63, 293-493. GEIGER, R. 1950. Das Klima der bodennahen Luftschicht. Die Wissenschaft, 78, ed . 3. Braunschweig. GILLETT, J. B. 1952. The genus Trifolium in southern Arabia and in Africa South of the Sahara. Kew Bull. 1952, 367-404. GILLMAN, C. 1923. An ascent of Kilimanjaro. Geogr. J. 61, 1-27. GINZBERGER, A. & STADLMANN, J. 1939. Pflanzengeographisches Hilfsbuch. Wien. GJ1EREVOLL, 0. 1956. The plant communities of the Scandinavian alpine snow-beds. K. norske Vidensk. Selsk. Skr. 1956 ( 1 ) . Trondheim. GOEBEL, K. 1891. Die Vegetation der venezolanischon Paramos. In GOEBEL, K., Pflanzenbiologische Schilder­ tmgen 2 (1), 1-50. Marburg. GooD, R. 1953. The Geography of the Flowering Plants, ed. 2. London, New York and Toronto. GRANVIK, H. 1923. Contributions to the Knowledge of the East African Ornithology. J. Orn. 71, Sonderheft. Berlin.

Acta Phytogeog?". Suec. 49 References 135

GREGORY, J. W. 1921. The Rift Valleys and Geology of East Africa. London. GRIESEBACH, A. 1872. Die Vegetation der Erde nach ihrer klimatischen Anordmmg. Ein Abriss der vergleichende Geographie der Pflanzen. Leipzig. GRUNNE, X DE, HAUMAN, L., BURGEON, L. & MICHOT, P. 1937. Le Ruwenzori. Mission scientifique beige 1932. Bruxelles.

HANCOCK, G. L. R. & SOUNDY, W. W. 1931. Notes on the fauna and flora of Northern Bugishu and Masaba (Mt. Elgon). J. E. Afr. nat. Hist. Soc. 36, 165-183. Nairobi. HANN, J. & SuRING, R. 1939. Lehrbuch der Meteorologie, ed. 5. Leipzig. HARE, C. L. 1941. The arborescent Senecios of Kilimanjaro: A study in ecological anatomy. Trans. roy. Soc. Edinb. 60, 355-371, Pls. I-II. HAUMAN, L. l933a. Esquisse de la vegetation des hautes altitudes sur le Ruwenzori. Bull. Acad. Belg. Cl. Sci. V, 19, 602-616, 702-7 17, 900-917. -- l933b. Les "Lobelia" geants des montagnes du Congo beige. Mem. Inst. col. beige 1933, l-52.

-- 1935. Les "Senecio" arborescents du Congo. Rev. Zool. Bot. afr. 28, l-76. HAUMAN, L. & BALLE, S. 1934. Les "Alchemilla" du Congo Beige. Ibid. 24, 301-368. HAURI, H. & ScHROTER, C. 1914. Versuch einer Ubersicht der siphonogamen Polsterpflanzen. Bot. Jb. 50, 618- 656.

HEDBERG, 0. 1951. Vegetation belts of the East African mountains. Svensk bot. Tidskr. 45, 140-202. l952a. Den afro-alpina florans uppkomst och utveckling. Naturvet. Forskningsr. Arsb. 1950-51, 195-199. Stockholm.

l952 b. Vegetation. In HEDBERG, 0., MA.RTENSSON, 0., & RuDBERG, S., Botanical investigations in the Paltsa region of northernmost Sweden, Bot . Notiser Suppl. 3 (2), 45-73. Lund. 1955. A pollen-analytical reconnaissance in Tropical East Africa. Oikos 5, 137-166. 1957. Afroalpine Vascular Plants. A Taxonomic Revision. Symb. bot. upsal. 15. 1958. The taxonomic treatment of vicarious taxa. In 0. lfEDBERG (ed.), Systematics of to-day, Uppsala Univ. Arsskr. 1958 (6), 186-194. 1959. An "open-air hothouse" on Mt Elgon, Tropical East Africa. Svensk bot. Tidskr. 53, 160- 166, PI. l. 1961. The phytogeographical position of the afroalpine flora. Rec. Adv. Bot. I, 914-9 19. Toronto. 1963. Ekologisk specialisering i den afroalpina £loran. (Ecological specialization in the afroalpine flora.) Svensk Naturvetenskap 16, 158-170. Stockholm. 1964. Etudes ecologiques de la flore afroalpine. Bull. Soc . bot. Belgique 97: 5- 18. HEILBORN, 0. 1925. Contributions to the ecology of the Ecuadorian paramos with special reference to cushion­ plants and osmotic pressure. Svensk bot. Tidskr. 19, 153-170. HEINZELIN DE BRAUCOURT, J. DE. 1953. Les stades de recession du glacier Stanley occidental (Ruwenzori, Congo Beige). Exploration du Pare National Albert, ser. 2, l. Bruxelles. HEINZELIN DE BRAUCOURT, J. DE & MOLLARET, H. 1956. Biotopes de haute altitude de Ruwenzori, I. Ibid. , 3. HOLTTUM, R. E. 1953. Evolutionary trends in an equatorial climate. Symp. Soc. exp. Bioi. 7, 159-173. Cambridge. HULT, R. 1881. Forsok till analytisk behandling af vaxtformationerna. Medd. Soc. Fauna et Flora fenn. 8, l-155. Helsingfors.

HUMBERT, H. 1935. Sur un Senecio arborescent nouveau des hautes montagnes du Congo beige et sur les liens phylogenetiques des especes alliees. Bull. Soc. bot. Fr. 81, 830-848. HuMBOLDT, A. VON. 1806. Ideen zur einer Physiognomik der Gewachse. Tiibingen. -- 1808. Ansichten der Natur. Stuttgart. (3 Aufl. 1849.) HuMPHREYS, W. J. 1940. Physics of the Air, ed. 3. New York and London. IvERSEN, J. 1936. Biologische Pflanzentypen als Hilfsmittel in der Vegetations-Forschung. Mitt. Skallings- laboratorium. Kopenhagen. JEANNEL, R. 1950. Haute3 Montagnes d'Afrique. Publ. Mus. Nat. Hist. Nat. Suppl. l. Paris. JENNY, H. 1948. Great soil groups in the Equatorial regions of Colombia, South America. Soil Sci. 66, 5-28. KEAY, R. W. J. 1959. Explanatory notes to vegetation map of Africa, published on behalf of A.E.T.F.A.T. Oxford (University Press).

KERNER, A. J. 1863. Das Pflanzenleben der Donaulaender. Innsbruck. KLOTZLI, F. 1958. Zur Pflanzensoziologie des Siidhanges der alpinen Stufe des Kilimandscharo. Ber. geobot. Forschungs-Inst. Riibel 1957, 33-59. KLUTE, F. 1920. Ergebnisse der Forschungen am Kilimandscharo 1912. Berlin. KRAMER, P. J. 1949. Plant and Soil Water Relationships. New York.

Acta Phytogeogr. Sueo. 49 136 References

1956. Physical and physiological aspects of water absorption. Handb. Pflanzenphysiol. 3, 124- 159. Berlin­ Gottingen-Heidelberg. KuLLENBERG, B. 1962. Some points of view on Microclimatology as an itegrative part of Terrestrial Ecology. Zool. Bidr. Uppsala 35, 457-479, PI. I. LAUER, W. 1952. Humide "Lmd aride Jahreszeiten in Afrika und Si.idamerika "Lmd ihre Beziehung zu den Vegeta­ tionsgiirteln. Bonner Geogr. Abh. 9, 15-98. LID, J. 1938. Mosbollar. Nytt Mag. Naturvid. 78, 101-104. LINDMAN, C. A. M. 1914. Nagra bidrag till fragan: buske eller trad? K. Vetensk.-Akad. Arsb. 12, 231-287. Uppsala.

LINKOLA, K. 1922. Zur Kenntnis der Uberwinterung der Unkrauter und Ruderalpflanzen in der Gegend von

Helsingfors. Ann . Soc. zool. -bot. fenn. Vanamo I ( 7 ) . LYON, T. L. & BucKMAN, H. 0. 1947. The Nature and Properties of Soils. Ed. 4. New York. MACKINDER, H. J. 1900. A journey to the summit of Mount Kenya, British East Africa. Geogr. J. 15, 453-486. MARTIN, W. 1952. Unattached mosses in the New Zealand flora. The Bryologist 55, 65-70. MATTICK, F. 1951. Wuchs- und Lebensformen, Bestand- und Gesellschaftsbildung der Flechten. Bot. Jb. 75, 378-424.

MEYER, H. 1890. Ostafrikanische Gletscherfahrten. Leipzig.

-- 1900. Der Kilimandscharo, Reisen und Studien. Berlin. MICHOT, P. 1937. Geologie et Geographie physique. In GRUNNE, X. DE, et al., Le Ruwenzori, Mission scientifique beige 1932, p. 207-250. MILDBRAED, J. 1922a. Ueber die Paramos-Vegetation der tropisch-afrikanischen Hochgebirge. Verh. bot. Ver. Brandenburg 64, 199-201. -- 1922b. Zur Kenntnis der Senecio-Baume der afrikanischen Hochgebirge. Fedde's Repertorium 18, 227-232. MILDBRAED, J. et al. 1910-1914. Wissenschaftliche Ergebnisse der deutschen Zentral-Afrika-Expedition 1907- 1908, 2. Botanik. Leipzig. MILNE, G. 1936. A provisional soil map of East Africa. Amani Memoirs. M0LHOLM HANSEN, H. 1956. Life forms as age indicators. Copenhagen. MOREAU, R. E. 1934. A contribution to Tropical African bird-ecology. J. Anim. Ecol. 3, 41-69. 1936. A contribution to the ornithology of Kilimanjaro and Mount Meru. Proc. zool. Soc. Lond. 1935, 843-891.

1938. Climatic classification from the standpoint of East African biology. J. Ecol. 26, 467-496. 1944. Kilimanjaro and Mount Kenya: some comparisons, with special reference to the mammals and birds; with a note on Mount Meru. Tanganyika Notes and Records 18, 1-32. 1945. Mount Kenya: a contribution to the biology and bibliography. J. E. Afr. nat. Hist. Soc. 81-82, 61-92. Nairobi. 1957. Variation in the Western Zosteropidae (Aves). Bull. Brit. Mus. (Nat. Hist.) Zool. 4 ( 7 ), 3ll-433. NEGER, F. W. 1913. Biologie der Pflanzen. Stuttgart.

NILSSON, E. 1932. Quaternary glaciations and pluvial lakes in British East Africa. Geogr. Ann. 13, 241-348. Stockholm.

-- 1940. Ancient changes of climate in British East Africa and Abyssinia. Ibid. 22, 1-79. NORDHAGEN, R. 1928. Die Vegetation und Flora des Sylenegebietes. Skr. Norske Vid.-Akad. Oslo, I. Mat.­ Naturv. Kl. 1927 : I. 0DMAN, 0. 1930. Volcanic rocks of Mt Elgon in British East Africa. Geol. Foren. Forh. 52, 455-536. Stockholm. OsMASTON, H. A. 1958. Pollen analysis in the study of the past vegetation & climate of Ruwenzori and its neighbourhood. B. Se. thesis, Oxford (unpublished). PETTERSSON, B. 1958. Dynamik och konstans i Gotlands flora och vegetation. Acta phytogeogr. suec. 40. PHILLIPS, J. F. V. 1930. Fire: its influence on biotic communities and physical factors in South and East Africa. S. Afr. J. Sci. 27, 352-367. PICHI-SERMOLLI, R. E. G. 1939. Aspetti del paesaggio vegetale nell'alto Semit'm (Africa Orientale Italiana). Nuovo G. bot. ital. II, 45 pp. CXV-CXXIV, Tav. 1-4.

-- 1957. Aspetti della vegetazione del Kilimanjaro. Ibid. 64, 760-763. PoLUNIN, N. 1948. Botany of the Canadian Eastern Arctic. Ill. Vegetation and Ecology. Nat. Mus. Canada Bull. 104.

PoTIER DE LA VAR DE, R. 1955. Mousses recoltees par M. le Dr. Olov Hedberg, en Afrique orientale, au cours de la mission suedoise de 1948. Ark. Bot. Ser. 2, 3 pp. 125-204.

Acta Phytogeogr. Suec. 49 References 137

RAUH, W. 1939. Uber polsterformigen Wuchs. N. Acta Leopoldina, N.F. 7 (49), 267-508, Taf. 23-44. Halle (Saale).

RAUNKIJER, C. 1907. Planterigets livsformer og deres betydning for geografien. Kj 0benhavn &Kristiania.

-- 1908. Livsformernes statistik som grundlag for biologisk plantegeografi. Bot. Tidsskr. 29, 42-83.

-- 1934. The Life Forms of Plants and Statistical Plant Geography. Oxford. RICHARD , J. J. 1945. Volcanological observations in East Africa. J. E. Afr. nat. Hist. Soc. 18, 1-12. ROBERT, M. 1946. Le Congo Physique. Ed. 3. Liege. ROBYNS, W. 1947. Flore des Spermatophytes du Pare National Albert. 2. Sympetales. Bruxelles.

-- 1948a. Les territoires biogeographiques du Pare National Albert. Inst. Pares Nat. Congo Beige. Bruxelles.

-- 1948 b. Flore des Spermatophytes du Pare National Albert. 1. Gymnospermes et Choripetales. Bruxelles. RoBYNS, W. & BouTIQUE, R. 1951. Cruciferae. Flore du Congo Beige et du Ruanda-Urundi. Spermatophytes , Vol. 2, 522-543.

RosvALL, S. & PETTERSSON, B. 1951. Gotlands Orkideer. Stockholm. RuNE, 0. 1953. Plant Life on Serpentines and Related Rocks in the North of Sweden. Acta phytogeog. suec. 31. SALT, G. 1954. A contribution to the ecology of upper Kilimanjaro. J. Ecol. 42, 375-423. SAUBERER, F. & HA.RTEL, 0. 1959. Pflanze und Strahlung. Leipzig. SAVILE, D. B. 0. 1960. Limitations of the competitive exclusion principle. Science 132 (3441), 1761. ScAETTA, H. 1933. Les Precipitations dans le bassin du Kivu et dans les zones limitrophes du fosse tectonique. Inst. roy. col. Beige, Mem., 2. Bruxelles.

-- 1934. Le Climat ecologique de la Dorsale Congo-Nil (Afrique centrale equatoriale). Inst. roy. col. Beige, Sect. Sci. Nat. Med., Mem. 4°, 3. Bruxelles. ScHIMPER, A. F. W. 1898. Pflanzengeographie auf physiologischer Grundlage. Jena. ScHIMPER, A. F. W.-FABER, F. C. VON. 1935. Pflanzengeographie auf physiologischer Grtm.dlage. 3 Aufl. Jena. ScHROETER, C. 1908. Das Pflanzenleben der Alpen. Zurich. SELANDER, S. 1950. Floristic Phytogeography of South-Western Lule Lappmark (Swedish Lapland), I. Acta phytogeog. suec. 27.

SERNANDER, R. 1927. Zur Morphologie und Biologie der Diasporen. Nova Acta reg. Soc. Scient. upsal., vol. extra ordinem. U ppsala.

SEWELL, T. G. et al. (The Tussock Grassland Research Committee) 1954. The high-altitude snow-tussock grass­ land in South Island, New Zealand. N.Z. J. Sci. Tech. Sect. A, 36, 335-364. SHANTZ, H. L. & TURNER, B. L. 1958. Photographic Documentation of vegetational Changes in Africa over a third of a Century. Univ. Arizona Coil . Agric. Rep. 169. SKOTTSBERG, C. 1912. The Vegetation in South Georgia. Wiss. Ergebn. Schwed. Siidpolar-Exped. 1901-1903, 6 (12). Stockholm.

1913. A Botanical Survey of the Falkland Islands. Bot. Ergebn. Schwed. Exped. Patagonien Feuerlande 1907- 1 909, 3. K. svenska Vet.-Akad. Handl. 50 (3). Uppsala & Stockholm. 1914. Studien iiber die Vegetation der Juan Fernandez-Inseln. Ibid., 51 (9). 1916. Die Vegetationsverhaltnisse langs der Cordillera de Ios Andes S. von 41 °S.Br. Ibid. 56 (5). 1929. Plant communities of the Juan Fernandez Islands. Proc. Intern. Congr. Plant Sci. Ithaca, New York (1926) l, 565-574. Menascha , Wisconsin.

SMITH, A. C. & KocH, M. F. 1935. The genus Espeletia: a study in phylogenetic taxonomy. Brittonia l, 479-530. S0RENSEN, TH. 1941. Temperature Relations and Phenology of the North-east Greenland Flowering Plants. Medd. Gmnl. 125 (9), 1-305, Pl. 1-15. K0benhavn. SPINK, P. C. 1945. Further notes on the Kibo inner crater and glaciers of Kilimanjaro and Mt. Kenya. Geog. J. 106, 210-2 16.

STEENIS, C. G. G. J. VAN. 1935. On the Origin of the Malaysian Mountain Flora, 2. Altitudinal Zones, General Considerations and Renewed Statement of the Problem. Bull. Jard. bot. Buitenzorg ser. Ill, 13 (3) , 129-257.

1939. Ecological observations on the genus Pleiocraterium in Gajoland, Sumatra. Rec. Trav. bot. Neerl. 36, 446-448, Tab. VI.

1948. General Considerations, in Flora Malesiana ser. I vol. 41 • Djakarta. 1962. The mountain flora of the Malaysian tropics. Endeavour 21, 183-193. STOCKER, 0. 1956. Die Diirreresistenz. Handb. Pflanzenphysiologie 3, 696-741. Berlin-Gottingen-Heidelberg. SYNGE, P. M. 1937. Mountains of the Moon. An Expedition to the Equatorial Mountains of Africa. London.

Acta Phytogeogr. Sttec. 49 138 References

TANSLEY, A. G. 1949. The British Islands and Their Vegetation. Ed. 2. Cambridge. TANSLEY, A. G. & CHIPP, T. F. 1926. Aims and Methods in the Study of Vegetation. London. TOBLER-WOLFF, G. & TOBLER, FR. 1914. Vegetationsbilder vom Kilimandscharo. G. KARSTEN & H. ScHENCK: Vegetationsbilder 12 (2, 3). Jena.

TROLL, C. 1943 a. Thermische Klimatypen der Erde. Peterm. geogr. Mitt. 1943, 81-89. 1943b. Die Frostwechselhaufigkeit in den Luft- und Bodenklimaten der Erde. Meteorol. Z. 60, 161-171. Braunschweig.

1944. Strukturboden, Solifluktion und Frostklimate der Erde. Geol. Rtmdschau 34, 545-694. 194 7. Die Formen der Solifluktion und die periglaziale Bodenabtragung. Erdkunde 1, 162-175. 1948. Der asymmetrische Aufbau der Vegetationszonen und Vegetationsstufen auf der Nord- und Siid­ halbkugel. Ber. geobot. Forschungsinst. Rubel 1947, 46-83. Ziirich. 1952a. Das Pflanzenkleid der Tropen in seiner Abhangigkeit von Klima, Boden und Mensch. Deutscher Geographentag Frankfurt 1951, Tagungsber. und wiss. Abh., 28, 35-67. Remagen/Rh. 1952 b. Die Lokalwinde der Tropengebirge und ihr Einfluss auf Niederschlag und Vegetation. Bonner geogr. Abhandl. 9 (1952), 124-182.

1955. Der jahreszeitliche Ablauf des Naturgeschehens in den verschiedenen Klimagurteln der Erde. Stu­ dium Generale 8, 713-733.

1956. Das Wasser als pflanzengeographischer Faktor. Handb. Pflanzenphysiologie 3, 750-7 86. Berlin­ Gottingen-Heidelberg .

1958. Zur Physiognomik der Tropengewachse. Jber. Ges. Fr. Ford. rhein. Friedrich-Wilhelms-Univ. Bonn E.V., 1958, 1-75.

-- 1959. Die tropischen Gebirge, ihre dreidimensionale klimatische und pflanzengeographische Zonierung. Bonner geogr. Abh. 25.

TROLL, C. & WIEN, K. (t) 1950. Der Lewisgletscher am Mount Kenya. Geogr. Ann. 31 (1949), 257-274. Stock­ holm. TuRRILL, \V. B. 1940. Experimental and synthetic plant taxonomy. In J. HuXLEY (ed.), The New Systematics, 47-7 1. Oxford. VANDEN BERGHEN, C. 1953. Quelques Hepatiques recoltees par 0. Hedberg sur les montagnes de l'Afrique Orien­ tale. Svensk bot. Tidskr. 4 7, 263-283. VOLKENS, G. 1887. Die Flora der Aegyptisch-Arabischen Wiiste auf Gnmdlage antomisch-physiolog.ischer Forschtmgen. Berlin.

-- 1897. Der Kilimandscharo. Berlin. et al. 1915. Berichtigung zu den von R. Muschler in Engl. Bot. Jahrb. XLIII. (1909), XLVI. (1911), XLIX. (1913) und L. Suppl. (1914) veroffentlichten Diagnosen afrikanischer Pflanzen. Bot. ,Jb. 53, 366-375.

WALTER, A. 1948. Observations of atmospheric pressure in East Africa, 1. Results from First Order Stations. E. Afr. meteor. Dept. Mem. II (1). WALTER, H. 1931. Die Hydratur der Pflanze und ihre physiologisch,-okologische Bedeutung. Jena. -- 1950. EinfUhrung in die Phytologie. IlL Gnmdlagen der Pflanzenverbreitung. 1. Standortslehre, 2. Stuttgart.

WARMING, E. 1884. Om Skudbygning, Overvintring og Foryngelse. Naturhistorisk Forenings Festskrift. K0ben­ havn.

1895. Plantesamfund. Grundtraek af den okologiske Plantegeografi. Kj 0benhavn. 1908. Om Planterigets Livsformer. K0benhavn. 1909. Oecology of Plants. An Introduction to the Study of Plant-Communities. Oxford. 1923. 0kologiens Grundformer. Udkast till en systematisk Ordning. K. danske Vid. Selsk. Skr. naturv.­ mat. Afd., 8. Raekke, 4 (2). K0benhavn. WARMING, E. & GRAEBNER, P. 1918. Eug. Warming's Lehrbuch der okologischen Pflanzengeographie. 3 Aufl. Berlin.

WEBER, H. 1956. Histogenetische Untersuchungen am Spross-Scheitel von Espeletia. Abh. Akad. Wiss. Lit. Mainz, math.-naturw. Kl. 1956 (9).

1958. Die Pa.ramos von Costa Rica und ihre pflanzengeographische Verkettung mit den Hochanden Sud­ amerikas. Ibid. 1958 (3).

Acta Phytogeog1·. Suec. 49 References 139

WEBERBAUER, A. 1905. Anatomische und biologische Studien iiber die Vegetation der Hochanden Perus. Bot. Jb. 37' 60-94. 1930. Untersuchungen ii ber die Temperaturverhaltnisse des Bodens im hochandinen Gebiet Perus u;nd ihre Bedeutung fur das Pflanzenleben. Ibid. 63, 330-349. 1931. Uber die Polsterpflanze Pycnophyllum aristatum tmd die Polsterpflanzen im allgemeinen . Ber. dtsch. bot. Ges. 49, 29-34. WHITE, F. 1962. Forest Flora of Northern Rhodesia. Oxford. ZEUNER , F. E. 1949. Frost Soils on Mount Kenya, and the Relation of Frost Soils to Aeolian Deposits. J. Soil Sci. 1, 20-30. Oxford.

Acta Phytogeogr. Suec. 49 Index to Latin names

(Page numbers in italics refer to pages with illustrations showing the relevant taxon.)

Adenocarpus Aralia 85 mannii 89, 109 Argyroxiphium Agave 85 sandwicense 85 Agrostis 115 Asplenium gracilifolia 72, 86, 99, 106, 115, 119, 120 kassneri 97 leptophylla 72, 107, 123 Astragalus sclerophylla 73, 78, 81, 88, 94, 125 venosus 118 trachyphylla 47, 72, 81, 107, 114, 116, 123, 125, 127 Aulacomnium volkensii 86, 114 turgidum var. pap illosum 125 Alchemilla 36, 43, 89, 90, 110, Ill, 114, 118, 120, 129 aberdarensis Ill Bartramia argyrophylla 43, 72, 110, 112 afro-ityphylla 99 ssp. argyrophylla 47, 100, 101, 102, 107, 108, 113, Bartsia 83, 89 127, 129 kilimandscharica 7 2 ssp . argyrophylloides 99, 106, 111, 119, 120, 129 longiflora 89 cyclophylla 115 macro calyx 7 2 elgonensis 38, 43, 54, 90, 99, 107, 110, Ill, 112, 114, macrophylla 89 121, 129 Blaeria johnstonii 70, 79, 90, 97, 99, 100, 101, 106, 109, 110, filago 89 Ill, 112, 113, 115, 116, 118, 119, 120, 122, 123, ssp . filago 107 125, 126, 127 Brachythecium keniensis 102 atrotheca 101, 113 microbetula 99, 127 implicatum 97, 112 stuhlmannii 43 ramicola 97, 99 subnivalis 88, 99, 110, Ill spectabile 97, 99, Ill, 120 Aloe 85 ugandae 117 Anastrophyllum vellereum 97 gambaragarae 99, 106 Breidleria Andreaea 68 africana 97, 99 Andropogon Breutelia amethystinus 47, 114, 115 madagassa var. kivuensis I 02 Anthospermum stricticaulis 112, 122, 130 usambarense 89 stuhlmannii 97, 99, 106, 120 Anthoxanthum subgnaphalea 97, 99, 106, Ill, 119, 120, 122, 127, nivale 38, 100, 113, 115 130 Anthriscus Bryum sylvestris 100 afrocalophyllum 102 Antitrichia argenteum 102, 113, 117, 125 curtipendula 97, 99, 100, 106, 122 capillare 99 kilimandscharica 113 ellipsifolium 101, 113, 125 Arabidopsis homalobolax 112 thaliana 42, 73, 77, 117, 127 Arab is Calamagrostis alpina 38, 42, 80, 91, 97, 99, 100, 112, 113, 117 hedbergii 107

Acta Phytogeogr. Suec. 49 Index 141

Callistemon 83 Dendrosenecio: see Senecio, subgenus Dendrosenecio Callitriche Deschampsia stagnalis 25, 66, I22 caespitosa 42, 47, ll5, I27, 128 Call una flexuosa 42, 9I, 97, IOO, 106, Ill, ll3 vulgaris 83 Dicranoweisia 68 Campylium Didymodon chrysophyllum I 02 papillinervis II7 "hispidulum var. sommerfeltii" I02, I23, I25 rufus I25 ste11atum I02 Drepanocladus Campylopus uncinatus 97, 99, ll9, I22, I27 alpicolus I27 denticuspes I25 Encalypta stramineus 97, 99, I02, I06, 119, I20, I22, I23 ciliata II7 Cardamine Erica 25, 37 hirsuta I27 Ericaceae 72, 89 obliqua 9I, 97, 99, IOO, IOI, ll2, ll7, I20, I22, I25, Espeletia 52, 55, 62, 85 I27 Euryops 35, 42, 89 Carduus dacrydioides 24, 72, 90, I08, ll7 keniensis 35, 37, 45, 6I, ·63, 65, 7I, 84, 85, I06, I08, II7 elgonensis I 00 nanus 73, 87 ruwenzoriensis var. lanuriensis 25 Festuca 65 Carex 48, I03, I05, ll3, ll8, I22, 127, I30 abyssinica 72, 86, 97, 99, 108, Ill, ll2, ll4, ll5, ll7 monostachya 64, 86, 90, 104, 116, ll8, 122, 123, 125, kilimanjarica 47, 72, 86, I08, ll4, ll7 126, I27' I30 pilgeri 47, 70, 72, 81, IOO, 106, I07, ll2, ll3, ll4, runssoroensis 27, 28, 38, 47, 59, 86, I06, ll2, 117, 116, 123, 125 ll8, 119, 120, 121, I23, 128, I30. ssp. pilgeri 36, 39, 44, 46, 47, 108, ll5, ll6, I23 Carica 85 ssp . supina 24, 30, 47, I08, ll7 Carpha eminii 25 Galium 1I2 Cephalozia glaciale IOO, 101, ll3, ll7, I27 fissa I20 ruwenzoriense 97, 99, I06 Cerastium Grimmia 68 octandrum I20 campestris 68, 92 var. adnivale 97, 99, IOO, I25 ovata 68, 92 Ceratodon perichaetialis 99 purpureus I02, ll3, I23, 127 Gyrophora 39 Cladonia ll2, ll3 pyxidata I02, I23, I25 Haplocarpha Collema II2 rueppellii 45, 64, 7I, 73, 74, 77, 87, 88, 94, 123 Conyza Haplosciadium subscaposa 73, 74, 75, 87, 94 abyssinicum 64, 66, 69, 73, 75, 76, 87, ll7, 123, 125, Cotula I27, 129 cryptocephala 73 Hebenstretia Crassocephalum dentata 89 ducis-aprutii 25 Helichrysum 36, 43, 45, 64, 65, 72, 89, I04, 107, ll7, Crassula I30 granvikii 42, 7I, 72, I22, I23, I27 amblyphyllum 43, 72, 89, I06, ll2 Crepis argyranthum 89 sp. near C. carbonaria ll8 brownei 43, 72, 89, 107 Ctenidium chionoides 43, 72, I06, I07 squarrifolium II2 citrispinum 43, 7I, 72, I06, I30 Culcitium 62 var. armatum I07, 108 Curculionidae 50, 6I var. hoehnelii 9, 24, 65, 108, 109, ll7 Cyclodictyon cymosum ssp. fruticosum 72, I07, I08, 117 borbonicum I20 gloria-dei 43, 106 Cyperaceae 90 nandense 89

Acta Phytogeogr. s��ec. 49 142 Index

newii 43, 71, 72, 88, 89, 90, 104, 105, 106, 108, 117 muricata 99, Ill f. brevicaulis 16, 88, 108 newtonii 97 splendidum 43 Lophozia stuhlmannii 15, 21, 40, 42, 59, 71, 72, 99, 104, 105, hedbergii Ill 106, 119, 120, 130 ruwenzoriensis 99 Heracleum Lupinus 61, 85 elgonense 100 alopecuroides 61 Hy1ocomium Luzu1a splendens 99, 100, 102, 111, 112, 113, 122 abyssinica 118, 123, 125 Hypericum 89 johnstonii 99, 100, 118, 120 keniense 89 Lycaenidae 39 Hypnum Lycopodium cupressiforme 97, 99, 100, 102, 113, 122 saururus 89, 118, 122 Hyrax 38 Malva

Iridaceae 90 verticillata 3 7 Marchantia

J uncaceae 90 polymorpha 100, 122 Meesea

Keniochloa kenyae 125 chionogeiton var. oreades 127, 129 Metzgeria Kiaeria 68 elliotii Ill Kniphofia hamata 97, 99, Ill snowdenii l 06 hedbergii 97 Koeleria Montia 129 gracilis 86, 114, 115 fontana 119, 127, var. convoluta 108, 112, 115 Mylia var. supina 117 hedbergii 97, 120 Myosotis 39, 79, 81, 102, Leptodontiopsis keniensis 73, 88, 94, 100, 113, 116, fragi1ifolia 97, 99 117, 125 00 Leptodontium vestergrenii l johannis-meyeri 113, 125 Nannoseris squarrosum 97 schimperi 64, 73, 74, 75, 87, 91, 94, 107, 117 sub1aevifolium 99, 106 Neckera tenerascens l 02, 119, 123 hedbergii lOO Liliaceae 90 Nectarinia Limosella dartmouthii 40 africana 66, 127, 128 johnstonii 38, 40, 47 Lobelia Nephroma 112 aberdarica 83 Nostoc bambuseti 83 commune 68, 69, 92 bequaertii 59, 60, 106, 118, 119, 120 deckenii 59, 60, 62, 83, 84, 94, 118, 127 Oreophyton e1gonensis 60, 118, 122 falcatum 73, 75, 87, 94, 97, 113 giberroa 83 Orthotri chum keniensis 38, 40, 45, 48, 57, 58, 59, 60, 61, 62, 71, 75, rupestre 100, 113, 117 81, 118, 123, 124, 125 Oxalis 1anuriensis 83, 97 corniculata 37, 112 sattimae 118, 122, 123 telekii 55, 56, 57, 58, 60, 83, 100, 101, 102, 106, 113, Parietaria 116, 117 debilis 25, 100 wollastonii 40, 57, 58, 60, 83, 97, 98, 99, 105, 106 Parmelia Lophocolea sp . near P. vagans 68, 69, 92 cuspidata 97, Ill Peltigera moelleri 99, 112 canina 97, 99, 102, 113, 117

Acta PhytogeogT. Suec. 49 Index 143

Pentaschistis 47, 114 Satureja 89 borussica l 08, ll7 biflora 89 rnannii 108, 118 kilimandschari 91, ll7 rneruensis 118 Scapania minor 31, 46, 1 07, 108, 117, 123 hedbergii 124 Peucedanurn Scirpus sp. 127 kerstenii 35, 37, 97 , 99, lOO, 106 Sedurn Philippia 25, 37 ruwenzoriense 38 keniensis ssp. keniensis 107, 108 Sematophy llum Philonotis elgonense 112, 122 alpicola 119 Senecio 43, 89, 97 Pirnpinella Subgen. Dendrosenecio 24, 26, 29, 3.J, 52, 56, 64, 71, kilirnandscharica 91 84, 91, 94, 96, 97, 98, 99, 103, 105, 109, 110, 120, Plagiochila 130 colorans 99 aberdaricus 100 hai1rnanii 97 adnivalis 21, 40, 45, 71, 97, 98, 106 , 117, 120 Pleuropus var. adnivalis 44, 99 sericeus 99, lOO var. alticola 42, 43, 96 Poa var. petiolatus 99 annua 37 amblyphyllus 54 leptoclada 47, 86, 112, 114 barbatipes 43, 54, 84, 99, lOO, 112, 121 ru wenz oriensis 21, 115 battiscombei 54, lOO schirnperiana 91, 97, 99, lOO, 101, 102, Ill, 112, 113, bra3sica 43, 52, 53, 71, 81, 82, 84, 85, lOO, 108, 116, 119, 120 123 Polypodiurn brassiciformis 84, 85, lOO, 123 rigescens 91, 99, 106 cottonii 24, 43, 62, 103, 104, 1 27 Polytrichnrn elgonensis 38, 99, 1 06, 107 keniae 99 erici-rosenii 42, 43, 54, 55, 96 piliferurn 123 friesiorum 43, 52, 84 Pringlea johnstonii 55 antiscorbutica 85 keniodendron 36, 37, 45, 47, 48, 49, 50, 51, 52, 53, Protea 54, 60, 71, 82, 100, 101, 102, 108, 112, 113, 116, kilirnandscharica 89 117, 125 Puya 61, 85 keniophyturn 43, lOO, 107 rairnondii 61 kilimanjari 103 meyeri-joham1.is 65, 108, 117 Radula purtschelleri 38, 113, 116, 117 boryana 99 schweinfurthii 117 rneyeri 99, l 00 snowdenii 72, 106, 112, 114 Rarnalina 3 9 sotikensis lOO Ranunculus telekii 38, 43, 72, 117, 118 cryptanthus 37, 73, 87 transmarinus var. sycephyllus 25, 99, 120 oreophytus 64, 73, 74, 87, 99, 100, 123, 125, 127 Silene stagnalis 127, 129 burchellii 91, 1 1 8 volkensii 127 Sphagnum 38, 118, 120, 121, 127 Rhacocarpus davidii 120 hurnboldtii 106, 123 madagassum 120 Rhacornitriurn 68 pappeanum 106, 119, 120, 123 Rhytidiurn Sticta 112 rugosurn l 00 Stoebe Romulea 42, 72 kilimandscharica 89 keniensis 72, 90, 123 Subularia monticola 42, 65, 66, 80, 122, 127, 128, 129 Sagina Swertia abyssinica l 00 crassiuscula 37, 112 afroalpina 73, 75, 79, 81, 88, 94, 124, 125, 127, 129 subnivalis 73, 88, 94

Acta. Phytogeog1-. Suec. 49 144 Index

Teophrasta 85 Veronica Thesium glandulosa 83, 91, 112, 129, 130 kilimandscharicum 8 9 Viola Tortula eminii 112 cavallii 47, 97, 99, 100, 101, 102, 113, 125, 127, 130 Wahlenbergia Trichostomum pusilla 73 cylindricum 112 Webera Trifolium afro -cruda 116 acaule 73 olgonense 73 Xestorhinus 61 Valeriana kilimandscharica 111, 129, 130 ssp . elgonensis 112, 114 Zygodon ssp . kilimandschttrica 100, 116, 117 intermedius var. strictifolius 99

.Acta Phytogeogr. S�tec. 49 ACTA PHYTOGEOGRAPHICA SUECICA

8. R. SERNANDER, Granskar och Fiby urskog. En studie Zeugnisse der spiitquartaren Vegetationsentwickl�g, over stormluckornas och marbuskarnas betydelse i den hauptsachlich der Wa.ldgeschichte, im nordwestlichen svenska granskogens regeneration. (Summary: The Gotaland, Stidschweden.) 1951. 21: - (14: -). primitive forests of Granskar and Fiby. A study of the 30. M. WERN, Rocky-shore algae in the Oregrund Archi­ part played by storm-gaps and dwarf trees in the re­ pelago. 1952. 33: - (22: -). generation of the Swedish spruce forest.) 1936. 21: ­ 31. 0. RUNE, Plant ,life on serpentines and related rocks {14: -). _ in the North of Sweden. 1953. 18:- (12: -). 9. R. STERNER, Flora der Insel Oland. Die Areale der 32. P. KAARET, Wasservegetation der Seen Orlangen und Gefasspflanzen Olands nebst Bemerkungen zu ihrer Trehorningen. 1953. 12: - (8: -). Oekologie und Soziologie. 1938. 21: - (14: -). 33. T. E. HAssELROT, Nordliga lavar i Syd- och Mellan­ 10. B. LINDQUIST, Da.lby Soderskog. En skansk lovskog i sverige. (Nordliche Flechten in Slid- und Mittelschwe­ forntid och nutid. {Zusa.mmenf.: Ein Laubwald in den.) 1953. 22: - (14: -). Schonen in der Vergangenheit und Gegenwart.) 1938. 34. H. SJoRs, Slatterangar i Grangiirde finnmark. (Sum­ 27: - (18: -). mary: Meadows in Grangarde Finnmark, SW. Dalarna, 11. N. STALBERG, Lake Vattern. Outlines of its natural Sweden.) 1954. 18: - (12: -). history, especially its vegetation. 1939. 12: - (8: -). 35. S. KlLANDER, Karlvaxternas ovre griinser pa fjall i 12. G. EA Du RIETZ, A. G. liANNERz, G. LoHAMMAR, R. sydvastra Jiimtland samt angransande delar av SANTESSON & M. WERN, Zur Kenntnis der Vegetation Hiirjedalen och Norge. [Summary: Upper limits of des Sees Takern. 1939. 12: - (8: -) . vascular plants on mountains in Southwestern Jaint­ 13. Vaxtgeografiska. studier tilHi.gnade Carl Skottsberg pa land and adjacent parts of Harjedalen (Sweden) and sextioarsdagen 1/12 1940. (Geobotanical studies de­ Norway.] 1955. 22: - (14: -). dicated to C. Skottsberg.) 1940. 36: - (24: -). 36. N. QUENNERSTEDT, Diatomeerna i Langans sjpvegeta­ 14. N. HYLANDER, De svenska formerna av Mentha gen­ tion. (Summary: Diatoms in the lake vegetation of the tilis L. coli. (Zusammenf.: Die schwedischen Formen der Langan drainage area, Jamtland, Sweden.) 1955. 1941. 12: - (8: Mentha gentilis L. sensu coli.) -). 22: - (14: -) · 15. T. E. HAsSELROT, Till kannedomen om nagra nordiska 37. M.-B. FLORIN, Plankton of fresh and brackish waters umbilicariaceers utbredning. (Zusammenf.: Zur Kennt­ in the Sodertiilje area. 1957. 20: - (14: -). nis der Verbreitung einiger Umbilicariaceen in Fenno­ 38. M.-B. FLoRIN, lnsjostudier i Mella.nsverige. Mikro­ skandia.) 1941. 12: - (8: -). vegetation och pollenregn i vikar av Ostersjobiickenet 16. G. SAMUELSSON, Die Verbreitung der Alchemilla­ och insjoar fran preboreal tid till nutid. (Summary: 1943. Arten aus der Vulgaris-Gruppe in Nordeuropa. . Lake studies in Central Sweden. Microvegetation and 18: - (12: -). pollen rain in inlets of the Baltic basin and in lakes 17. TH. ARWIDSSON, Studien tiber die Gefasspflanzen in den from Pre-boreal time to the present day.) 1957. 10:­ Hochgebirgen der Pite Lappmark. 1943. 27: - (18: -). (6: -). 18. N. DAHLBEOK, Strandwiesen am stidostlichen Oresund. 39. M. FRIES, Vegetationsutveckling och odlingshistoria i (Summary: Salt marshes on the S.E. coast of Oresund.) Varnhemstrakten. En pollenanalytisk undersokning i 1945. 18:- (12: -). Vastergotland. [ Zusa.mmenf.: Vegetationsentwicklung 19. E. VON KRUSENSTJERNA, Bladmossvegetation och und Siedlungsgeschichte im Gebiet von Varnhem. Eine bladmossflora i Uppsalatrakten. (Summary: Moss flora pollenanalytische Untersuchung aus Vastergotland and moss vegetation in the ·neighbourhood of Uppsala.) (Stidschweden).] 1958. 16: - (10: -). 1945. 27: - (18: -). 40. BENGT PETTERSSON, Dynamik och konstans i Gotlands 20. N . .ALBERTSON, Osterplana hed. Ett alvaromrade pa flora och vegetation. (Resume: Dynamik und Kol.\Stanz Kinnekulle. (Zusa.mmenf.: Osterplana hed. Ein Alvar­ in der Flora und Vegetation von Gotland, Schweden.) gebiet auf dem Kinnekulle.) 1946. 27: - (18: -). 1958. 40: - (27: -). 21. H. SJo:Rs, Myrvegetation i Bergslagen. (Summary: 41. E. UGGLA, Skogsbrandfiilt i Muddus nationalpark. Mire vegetation in Bergslagen, Sweden.) 1948. 36: ­ (Summary: Forest fire area.s in Muddus National Park, (24: -). Northern Sweden.) 1958. 21: - (14: -). 22. S. AHLNER, Utbredningstyper bland nordiska barr­ 42. K. THoMASSON, Nahuel Huapi. Plankton of some lakes tradslavar. (Zusa.mmenf.: Verbreitungstypen unter in an Argentine National Park, with notes on terrestrial fennoskandischen Nadelbaumflechten.) 1948. 30: ­ vegetation. 1959. 21: - (14: -) . (20: -). 43. V. GILLNER, Vegetations- und Standortsuntersu­ 23. E. JULIN, Vessers udde. Mark och vegetation i en igen­ chungen . in den Strandwiesen der schwedischen West­ vaxande lovang vid Bjarka-Saby. (Zusammenf.: Vessers ktiste. 1960. 33: - (22: -). udde. Boden und Vegetation in einer verwachsenden 44. E. SJoGREN, Epiphytische Moosvegetation in Laub­ Laubwiese bei Bjarka-Saby in Ostergotland, Stid­ wiildern der Insel Oland, Schweden. (Summary: schweden.) 1948. 21: - (14: -}. Epiphytic moss communities in deciduous woods on the 24. M. FRms, Den nordiska utbredningen av Lactuca al­ island of Oland, Sweden.) 1961. 24: - (16: -). pina, Aconitum septentrionale, Ranunculus platani­ 45. G. WISTRAND, Studier i Pite Lappmarks kiirlvaxtflora, folius och Polygonatum verticillatum. (Zusammenf.: med sarskild hiinsyn till skogslandet och de isolerade Die nordische Verbreitung von Lactuca alpina ...) fjiillen. (Zusammenf.: Studien tiber die Gefasspflanzen­ 1949. 15: - (10: -). flora der Pite Lappmark mit besonderer Beriicksichti­ 25. 0. GJ..EREVOLL, Sn0leievegetasjonen i Oviksfjellehe. gung des Waldlandes und der isolierten niederen Fjelde.) (Summary: The snow-bed vegetation of Mts Oviks­ 1962. 36: - (24: -). fjallen, Jiimtland, Sweden.) 1949. 18: - (12: -). 46. R. IvARssoN, Lovvegetationen i Mollosunds sooken. 26. H. OsVALD, Notes on the vegetation of British and (Zusammenf. : Die Laubvegetation im Kirchspiel Irish mosses. 1949. 12: - (8: -). Mollosund, Bohuslan, Schweden.) 1962. 30: - (20: -). 27. S. SELANDER, Floristic phytogeography of South­ 47. K. THOMASSON, Araucanian Lakes. PlanktQn studies in Western Lule Lappmark (Swedish Lapland) I. 1950. North Patagonia, with notes on terrestrial vegeta" tion; 21: - (14: -). 1963. 36: - (24: -). 28. S. SELANDER, Floristic phytogeography of South­ 48. E. SJoGREN, Epilithische und epigaische Moosvegeta­ Western Lule Lappmark (Swedish Lapland) II. Karl­ tion in Laubwaldern der Insel Oland, Schweden. vaxtfloran i sydviistra Lule Lappmark. (Summary: (Summary: Epilithic and epigeic moss vegetation in Va.scular flora.) 1950. 18:- (12: -). deciduous woods on the island of Oland, Sweden.) 29. M. FRms, Pollenanalytiska vittnesbord om senkvartiir 1964. 40: - (27: -). vegetationsutveckling, sarskilt skogshistoria, i nord­ 49. 0LOV HEDBERG, Features of afroalpine plant ecology. vastra Gotaland. (Zusammenf.: Pollenanalytische Avec un resume en fran�ais.- 1964. 36: -, Pris 36 kronor Printed in Sweden 1964