ArticlePDF Available

Erica excelsa as a fire-tolerating component of Mt. Kilimanjaro's forests


Abstract and Figures

The altitudinal and ecological distribution of Erica excelsa on the southern slopes of Mt. Kilimanjaro was studied at 407 sampling plots, using the method of Braun-Blanquet (1964). Erica excelsa occurs in the altitudinal range between 1600 and 3500 m where it was found on dry ridges as well as in riverine and even swampy forests. On Mt. Kilimanjaro the upper tree line is represented by pure stands of Erica excelsa, which cover large areas of the subalpine zone. However, this tree is also a component of the montane mixed forests with a preference of the upper, and to a lesser degree of the lower montane zone. However, in the central montane zone Erica excelsa is very rare. The boundary between the Podocarpus Latifolius-dominated montane forests and the Erica excelsa-dominated subalpine forests is very sharp and presumably a result of fire. After a fire, Erica excelsa regenerates regularly by reprouting rather than from seeds. Whether and where the Podocarpus forests of the upper montane zone are thus replaced by an Erica forest is mainly a question of intensity and frequency of fire. The life strategy and competitive strength of the light-demanding Erica excelsa is thus obviously based on its tolerance of fire. The location of the upper forest line on the south slopes of Mt. Kilimanjaro therefore appears as a result of fire rather than of climatic factors. Unusually disastrous fires during the last years pushed the extant upper forests line downhill by approximately 300 m and recovery of the former tree line would require several decades of undisturbed growth.
Content may be subject to copyright.
Berlin-Stuttgart, December 14, 2001
449 -475
31 (4)
Erica excelsa as a fire-tolerating component of
Mt. Kilimanjaro's forests
by Andreas Haup and Erwin Brcr, Bayreuth
with 13 figures and 1 table
Abstract. The altitudinal and ecological distribution ol Erica excelsa ot the southern
slopes of Mt. Kilimanjaro was studied at 407 sampling plots, using the method of Bn¡ur-
Brerqunr (1964). Erica excelsa ocans in the altirudinal range between 1600 and 3500m
where it was found on dry ridges as well as in riverine and even swampy forests. On Mt.
Kilimanjaro the upper tree line is represented by pure stands of Erica exceha, which cover
large areas of the subalpine zone. However, this tree is also a component of the montane
mixed forests with a preference of the upper, and to a lesser degree of the lower montane
zone. Hov¡ever, in the central montane zone Erica excelsa is very rare. The boundary
between the Pod.ocarpus latifulius-dominated montane forests and the Erica excebd-doml-
nated subalpine forests is very sharp and presumably a result of fire. After a Êire, Erica
excelsa regenerates regularly by resprouting rather thân from seeds. Vhether and where
the Pod.ocarpøs forests of the upper montane zone are thus replaced by an Erica forest is
mainly a question of intensity and frequency of fire. The life strategy and competitive
strength of the light-demanding Erica excelsa is thus obviously based on its tolerance of
fire. The location of the upper forest line on the south slopes of Mt. Kilimanjaro therefore
appears as a result of fire rather than of climatic factors. Unusually disastrous fires during
the last years pushed the extant upper forest line downhill by approximately 30C, m and
recovery of the former tree line would require several decades of undisrurbed growth.
Keywords: fire-regeneration, tropical montâne and subalpine forests, forest line, Tanza-
1 Introduction
Erica excelsa (Alm & Fries) Beentje (formerly Pbilþpiø exccelsa (Alm Sc
Fries)) has been considered as an East and Central African endemic, inhabi-
ting montane and alpine regions (Arru & FnlEs 1927). On the southern
slopes of Mt. Kilimanjaro it covers extensive areas of the subalpine zone,
and pure stands of it form the upper tree line. This contrasts with the upper
forest belts on other East African high mountains as e.g. on Mt. Kenya and
the Aberdare Mts., which are dominated by the Rosaceae Høgeniø abyssi-
nicø (BussvrANN 8a Bncx 1995, Scnurrr 1991).In this work we report on
the altitudinal range and on the contribution of the tree-like Ericø excelsa
to different forest communities. Furthermore, we comment on the sharp
interface between the broad-leaf montane and the subalpine Erica-forests,
and on the upper tree line formed by Erica excelsa. A subalpine belt formed
mainly by shrubby Erica species such as E. trirnera (various ssp.), E. arbo-
03 40 -269X/ ol / 003 r -04 49 s 6.7 s
@ 2001 Gebrüder Borntraeger, D-14129 Berlin ' D-70176 Stuttgart
Þ Þ ø)ì À
Þ5oF t¡r
* Q-É è O
h ^ g-s)
:: ÞÞ-
ã l{l
s, o ã'*
d?Þ s
ã :: È s
r-¿ ! l- s:
83 P;'
88.Q s
3* a'il "
^ Y
y 3
= f I 3
E ( O-
l(l r¡¡
3E È +
ro ã ã'Þ
\o.:Ø i.r
grdlõ >
P,^á F
o coô ã
=Y ÈD' Ë
*F2 R
Ñ P'; t"l
qËä *
? ='? o-e-a ø(rz r..) N)
ó: &E'ã E'9 N3à i
3igãgãiin r{ v,
äiË äi råis E i
tr3ocpõoþ-Ë'å' <
ãâ.F FE-a,* <
tàq1äË;çr E
-E.qc Íõ ã I a.dx
ìeqs F E.F"
V . È H êv h Ù ^
8 Fe'ã H-Tq a-B
< ã'i f ,.9 co uro
Ë¡ 3 ã sXq ìr
ûc õoa - * ã õ-i-
Ë 1t: Í q i r,9
ä.þILË È et i
1l ã'[ È'-.p:+
Ë-:-tF qÞE-Ë'3
5 :r'-' - (o - *l ã'P
f i e ¡ätr
¡¡ o s 5 v s^ I
O-;r-P ã ti 5 n =i<
o --^ .0 9 p a r Q
é * 19 ^ H ^ L- re (9
t ¿ 6 gH r! : ^ H
ñli€ s 5ä+\
Ëq"" F'r g û
-:-__J v J J
ts ät< Þ ts Þ a Ð
. E Ø X rU
Þ o i'¡¡ ö i ã o-
.i åili ËE l$Ì: irs il
; äËË3i s 3:n f s il: .
írgl [;:: f å1qi
räBäiË gFii ä ilg
a'x õr-P5'{ ß oGp ã-ñ
EÉËFg*g 5 e $E;Ä
Ê{ t{3æãÊ> !'.l
;¡'Þ-=-:;-Fr ã \o5
g Et il [as 3 à'
' I o -.E 3 *i å
ilã tl Ê [ f q g
<e-ã 3 É + a
\ q'A ã - ã o o 'iì
S F'ä F,Ã *
ñ 3.pä å.'cü
ì-: g€ ã: È
Èjo S - 5 e ï
s-ã * o x gì-N <
ìPUNA (l
\ j ù ñl - W5 Oe
È LS õ Ë î $
l.Y s.ã Þ-P Þ å 9.
R ¡, \ i .'% Þ ? o
è3 * trrlÌ 5
Þ Ê 6' ñ é 71
¡úE"'3 b¡ d ô =
X g É-o id F+,ä
;.¡i ô ô +''<.
P ->Ò *-"o È
lÌ\¿ =-Ø o c)U v
õ .'aE.H o3 ä'
;'ÈË 3lE Ë.qs.
-- L 6 Vu.6 ¿
(D Ñ'É - =9q -d.-
5'e'$ã i -õx
- È_ô (D ö ..ã z
P ^ . U -. V^ :
lÈs trË:{;
È <-cri S.9;-1 grÉ
I r.^ ! L È P H
I \ J" :' ã aD ô Ø
Kibo t-¡:
6000 m
5000 m
@ shira
4000 m
2000 m
A..- €;ên"
1 000 '3
¡Fig. 1. Distribution of Erica exceha over the altitudinal vegetation zones on the southern_ slope of Mt. Kilimanjaro: Areas of
do-itt"rrt E. e. are underlaid dark grey and of scattered E. e. light grey. A: colline zone, B: submontane zone, C: montane zone, D:
subalpine zone, E: alpine zone.
452 A. Hemp & E. Beck
per parts it is associated with Podocørpøs latiþliøs. Above 2800 m asl Erica
excelsa is the dominant rree species. -
In the range berween 3100 and 3500 m the foresrs of Erica excelsø are
gradually replaced by the so-called Erica-bushwhich is dominated by Erica
arboreø and E. trimera-, Protea. caffrø and Eøryops dacryd.ioides. A iypical
moorland vegetation, formed by tussock grasi ãnd intêrspersed wiih the
characteristic giant lobelias, fringes the forest in the Soutlieast. At an alt-
itude of about 3900 mthe Erica bush grades into a Helichrysurn scrub vege-
tation that extends up to 4500 m. Atiigher altitudes a ríountain dereri is
encountered and the top of Kibo is covered with glaciers (for details of
the subalpine and alpine vegetation of Mt. Kilimanjaio see H¡onsn c 1951,,
Krörzrr 1958, Bncx et al. 1lS:).
3 Methods
Vegetation was analysed by relevés which were established at 100 m inter-
vals along 24 transects during the years 1996-2000 (for the locarion of the
transects see Fig. 2). Using the method of BneuN-BLANeuEr (1964) 407
relevés were made in the forests.- Special attention was given to homoge-
neity and representation of typical fõrest stands. The relevé size was choien
with respect to the minimum areas and was 1000 m2 in forests and 2OO m2
in the shrub. To encompass all imporranr foresr types of a given altitudinal
interval, the number of ?elevés p"itOO m altitude'rräried in iccordance with
the diversity of habitats in the ?espective range.
. In additiôn to.the ground layeilwithour l;bels in Table 1) a shrub layer
(S, 1 to 10 m tall) and a tree layer (l > 10 m tall) were differentiated.
Epiphytes were- treated as an additional storey (E). For all plots the typical
parameters such as altitude, exposition and inclination wãre determined.
For selected plots temperature and pH of the soil were measured.
Above about 3200 m Erica excella trees do nor exceed a heieht of 10 m.
Nevertheless, those stands were treated as forests as well (MlrËa & MrpHB
The relevés were clusrered according to floristic similarity and the resul-
ting plant communities are presented ñ Täble 1 as relative frequency of the
species using the following constancy classes: V=81-100 y",iY=6i-BOo/o,
III=41 - 60 o/o, II=21 - 40 y", I=1 | - 20 yo, +=6 - lO o/o, r<6"/". Species with
low constancy .were omitted. The entire species list may be requested from
the authors.
To determine the significance and relative importance of Erica excelsø in
the various communities, the relevés were analyìed and the so-called "ave-
rage percentage cover value" (in German referred to as "mittlerer Gruppen-
mengenanteil", cp. DrenscHrr 1994) was calculated. In this calculatión the
cover value of Erica was expressed as percentage of the cover value of the
whole vegetation in a plot. Applying tlis calculation then to all plots of a
plant communit¡ the average was cãlculated. An average percenåge cover
value of,...g: 5-0 o/o means that hal{ of the vegetarion ðorrìrage oia plant
community is formed by Erica excelsø.
Ericø excelsa as e component of Mt. Kilimanjaro's forests 453
5 Ë ãF,É ÌÀ
e* Ë
5_Ê Ëå å€
þRP- È E P;'
;åËË Ër¡gË
ffiffiffiHffi |
Table 1. Sociological value of Erica excelsa in montane and subalpine forest communities
on the southern slopes of Mt. Kilimanjaro.
1: Lower montans Agauia-Ocotea forcsl 7: Subalp¡ns r¡vsrinê forest
2: M¡ddle montane Ocotbs forest 8: Subalpinê Hagen a forest
3: Middle montane gorge and riverine forest 9: Subalpine Enba axc€lsa forest
4: Upper montane Podoc€rp./s-OcofBa forEst 10: Subalpine Eøba êxcelsa swamp forest
5: Upper montane Podocarpøs forest 1 1: M¡ddle subalpine Enca excelsa forest
6: Uppêr montane Eñca excelsa forest 12: Subalpine Enba ûl¡nela and arbore€ bush
Dotted fields: Enba excelsa dominated communities and fire regeneEtion stages
E=epiphyte, Èshrub, SL=liana rqch¡ng into the shrub layer, TL=liana .eaching into the lÍee layer, T=tree, no label=herb layer
Constancyclassos: V=81¡00%, lV=61-80%, lll=4'1-60%, ll=2140%,1=11-2O%, +=ô10tß, F<6% constancy
community number
Number of re'eves
AveÞge altitude (m asl:10)
AveEge spæi€s number
AveEge number ofùee specis
AveEge number of shrub spæ¡es
AveEge number of lianes
AveEgê number ofEscular epiphyts
AveEge cder velue of the tree layer (%)
AveEge hight ofth6 tre€ layer (m)
AveEge @ver €lu€ of the shrub layer (96)
AveEge height of the shrub layer (%)
AveEge @verElue ot the herb layer (%)
AveEge æver of the moss layer (%)
AveEge covervalue of the epiphyte layer (%)
14 12
51 31
3,4 3,8
8,1 5,3
2,2 1,3
5,6 4,3
33 59
't8 18
5,5 3,4
80 55
77 35 45
49 65 46
7,1 6,3 5,3
13,0'13,0 9,5
4,4 4,9 4,a
15,0 15,0 14,0
74 s6 70
?3 29 29
48 5't 31
4,5 5,2 4,2
44 65 67
34 51 48
Er¡caexceßa T
Ericaexceßa s
Er¡ca èxcê,sd
Montane and subalp¡ne forest speqies
Asplen¡um âelh¡opicum E
Cypes lùus sylv$tÌs
Lepistus excavalus E
Plæpelt¡s madædrya E
Athyrium 8ændic¡nuñ
San¡cula elata
Dryoptqis Íaden¡i
Amaurcpelta bqg¡ana
Rapanæñelanophl@ T
Rapanæmelanophl@ s
Rapanæ ñeldnophl@
Rubus sleudnqi
lmpat¡ens k¡l¡manjari
Nothope ne ñ a squ d m¡seta
Elaphoglæsuñ dæken¡¡ E
Dryoplqis k¡lemensis
Hynenophyllum æpillarc E
Montane forel specles
Ilymenophyllum kuhni¡ E
Eñbel¡d sch¡mpqi
Eñbel¡a sch¡ñpqi TL
Podæaryus lal¡fol¡us T
Pdæaryus lal¡lol¡us S
Podæarpus lalifolius
Maylenus acuminata S
Xymalx monæpffi l
Xyñalæ monNpffi S
Asplen¡um smedsi¡ E
Asplen¡um smedsii
llex ñitìs ï
llex ñ¡ts S
Schemævolkens¡¡ TL
Schenævolkens¡¡ T
Scheñlffivolkensi¡ s
Asplen¡um qælum vr. usambüense
Senæio syring¡fol¡us SL
Asplehiuñ lries¡flm E
Asplen¡um tr¡es¡qum
Psycholria cyalhìælyx S
Psychobia cyêth¡calyx
P¡læ usambænsis wt. englq¡
Elaphoglæsuñ angulaluñ E
Blotella glabre
Hyñenophylluñ lunbñgense E
Pepqom¡a lffiandopoiana E
Pepæñ¡è lffiandopo¡ana
M ¡m u I o ps ¡s k¡ I ¡ ña nd sch âi æ
Asplenium ell¡otli¡
EIê þ s u ñ aßlichoidæ E
ll +
tv lil
ilt lt¡
ilt t
lll +
ll +
ilt tv Iv
ililt il
+ll t
ilt lt
ll lv tv
ilt tv
ll lr rv
tv tv ilt
lv tv ilt
ilr rv ilt
ll + lll
ilt tvv
ilt tv
Vegetation zone
Comnunity number
Number of relæ
Lower and mlddle montane foÞst speci€s
Ocþtæusambarens¡s T
Ocolæusambaensis S
Ocofæ usambáêtsß
Pepqom¡aebyss¡niæ E
Cyathæ ñdnn¡ana S
SchenemYr¡antha TL
Blol¡ella st¡p¡lata
B€{,on¡a meys¡-johann¡s TL
Begonia meyqi-iohann¡s SL
Beg o n ¡ a m eyqi -i o h a n n ¡
Asplen¡um pþtensum
Las¡anlhus kiliñandschü¡cus S
Las ¡a nlhu s k¡ li mand Echüicus
cryplotaàn¡a alricana
PneumèÍopls¡s un¡la
Table 1 (cont¡nued)
TI 5
52 34
35 il5 7
14 I
12 12
Sleplæüpus mûlanus E
Asplen¡um hypoñeles E
Chassêt¡a pay¡fôlia S
Blæhnum allenuatum E
Gal¡n¡ffi sdihaga S
Iñpatiens pseudov¡old
Psycholria lßcl¡n*afa S
Canlh¡uñ ol¡gæâtpuñ ssp. câPfuø S
Dracaena alrcmonlana S
Dmæena âlromontana
Asplen¡um sandæonii E
/sog/6sa /aclæ
Aphtoia lhe¡fm¡s T
Aphlo¡a lhe¡Íqm¡s S
syzyg¡uñ guinænse T
Pepsom¡aletaphylla E
Drynaiêvolkensi¡i E
Pauríd¡natha pauc¡nfl¡s s
Cyphætemma masukuense
Macaßngak¡l¡mêndschaicê T
Routæthoñsoni¡ TL
TabffiêeñØlanapachysiphon T
Asplen¡um l¡nck¡i
lsachne mauit¡ana
Elaphoglæsum lasu¡ E
Elaphc{læsum aubdTi¡ E
Elaphoglæ6um aubql¡¡
Spec¡es of Ágaulaocolæ forests
Villaiavolkens¡¡ E
Agauña sal¡c¡fol¡a T
Agauda sal¡c¡fol¡a S
Ruûdæ luscsæ¡s S
My¡ca sal¡cifol¡a T
Myrica sal¡cifol¡a S
Smilax anæps SL
Pan¡cum monl¡cola
Ps eudæh i n ol aenê polystac hya
Kæt¡a gue¡nz¡¡ s
Kætia gueinz¡i TL
Kæl¡a gue¡nzi¡
Dahlb*g¡a lêctæ
Alb¡z¡a guñm¡|ffi T
spec¡es of communlV 1a
Drymaile cddata
Justicia llava
Anisopappus olivænus
spec¡es of middle montane Ocofæ foresG
Garc¡n¡dvolkens¡¡ T
Garciniavolkensi¡ S
Species of rlver¡ne forests
selag¡nelld kreuss¡ana
Asplenium aby66¡n¡cum
cysloplq¡s lngilis
Pil6 rivuleils
Alrøan¡a volkens¡i T
Elatxtemña monticolum
Depaia bqyana
lil l lll
tv v lll
tv rr lll
ill Il
tv lll ll
ililt l
v lll ll
ilt il+
ilt lll l
lil tv ll
ilt lv rv
v tv lll
il lll ll
tv lv lll
ilr illl
tv tv ll
lt I
I l'
ilt ill
lll ll t
ll l+
lll ll I
ll lt
tf ilt lll
Vegetaüon zonè
lll +
It v
lv ilr
llt llt
llt ilt
lll ilr
tv ilt
ilt tv
Vegetaüon zone
Communltlr number
Numbe¡ of relêves
Upper montane and subalp¡ne forest spec¡es
Asplen¡um volkens¡¡
Xiphoplqis nabeil¡fffils E
Galium apañno¡dæ
H¡sl¡oplqis inc¡sê
Aspleñ¡um pnegßc¡le
As pleni u m I oxæca p ho¡d æ
Polysl¡chum wilson¡i
Prunus aft¡cana T
Ptunus alricdnd S
Subalpine forest specles
Hagen¡a abyssin¡ca T
Hagen¡a abyss¡n¡ca s
Hypqicum rcvolulum ssp. kerlerse
Peucdanum I¡nd*i
Polystichum volkens¡i
serærc suôsessr76
V¡ola em¡n¡¡
senæ¡o cyaneus
Hypolepis gæae¡
Cffist¡um afromonlanum
Conyzavqhonio¡d$ S
Cynog 16 s u m a m p lifol i u m
Hel¡chrysum fm6¡asimum
C¡nffiña delto¡dæ
Cèrduus nyassanus
Kn¡phofia lhomson¡i
Specles of comnun¡ty 6 and 9
Lycopd¡um clavalum
Deschamps¡a llexuæa vat. êfromontana
Myß¡ne alr¡cana S
Mtrs¡ne afriæna
Spec¡es of commun¡ty 6-12
Senæ¡o schwe¡nfuThii
Gffin¡uñ areb¡cum
Luzula abyss¡n¡ca
Species of commun¡ty 7
Dendtæene¡o johnston¡í S
Lobeliadækenii ssp. ¡ncip¡ens S
Lþbel¡a dæken¡¡ ssp. ¡nc¡p¡ens
Stegnqmmña pozo¡
Planlago pdlñdta
Anthriscus sylv$tis
Alchemilla l¡schqi
Pilæ johnston¡i ssp. johnston¡i
Specles of Eniå forests and bush
Penlasch¡sl¡s chrysùrus
Helichryaum nandense
Alchem¡Ia johnston¡¡
Heb ensbetia a ng ol e n s ¡ s
Anthæpqñum usambarense S
Spec¡es of communiV l0
Carex monælâchyê
Hyd@olyl e s i bthqp ¡o ¡d æ
Carcx @nfd7a
Specles of conmunity 1l
Oendr6enæ¡o Øllon¡¡
Hupæ¡a saururus
Fætuca obluúans
Dryoptqis ønlarct¡ca
Hel ic hrys u m ñey*¡ -joh a n n ¡ s
Ba¡1sia dæurya
Prolæ ælÊ ssp. k¡l¡mandscharica S
Spec¡es of community l1a
My6otis abyss¡n¡Ça
Fumü¡a abyss¡n¡cê
Spec¡es ofcommun|ty l2
Eri@ahM S
Hel¡chrysum splend¡duñ
Eri@ triñffi S
Erí@ b¡ñffi
Hel¡chrysum newi¡
Blaqid johnstonii
Teble I (continued)
77 35 45
52 78
14 't2
Ericø excelsa as a component of Mt. Kilimanjaro's forests 457
The occurrence of Erica excelsa was mapped into the UTM grid, with a
resolution of 1 km2. To portray the.structure of the foresrs, stand-profile
diagrams according to FIAMMEN et al. (tlSl) were produced.
The nomenclature of the plants follows Acr.rcV & Acxrw (1994), Bn-
ENrJE (1994), Harxns & Lvr (1983) and Husneno et al. (1.952-).
4 Results
4.1 Distribution of Erica excelsa
Based on 407 relevés Fig. 3 presents the consrancy of Ericø excelsa in the
forests over an altitudinal range of 24OO m, as well as the average percenrage
cover values. Below 1600 m Erica excelsa.was nor encountered in the study
a¡ea. Both parameters show a first small peak at 1600 m asl. The consrancy
then rises rapidly to 80 % at 2600 m whilè the percentage cover value beco-
mes significant (over 30%) only above 3100m. In accordance with the
field observations the analysis of a SPOT satellite image shows rhe bimodal
distribution of Ericø excelsa on the sourhern slope of Mt. Kilimanjaro with
its gap in the middle of the forest belt (Fig. a).
01100 1300 1500 1700 1900 2100 2300 2500 2700 2900 3100 3300 3500
altitude (m asl)
+ Constancy + average peroentage Cover value
Fig. 3. Altitudinal distributio n of Erica excelsa in the forests of the southern slope of Mt.
Kilimanjaro, based on the evaluation of 407 relevés.
4.2 Plant communities with Erica excelsa
As Erica excelsa is absent below 1600 m only forest communities of the
montane and subalpine zones will be presented here. The term "subalpine"
is applied to the transition zone between the broad-leaf montane forest
and the alpine Helicbrysam scrub vegetation. It corresponds closely ro rhe
"ericaceous belt" of Flrnnnnc (1951). For the altitudinal zonation of the
vegetation on Mt. Kilimanjaro see also Hrup (in press a).
õ =.
458 A. Hemp E¿ E. Beck Erica exceba as a component of Mt. Kilimanjaro's forests 459
the successional stage correlates with the degree and time point of distur-
bance by fire. In Table 1 the communities dominated by E. excelsa and the
various regeneration stages from fire are shadowed.
lüíith respect to the occurrence of Erica on Mt. Kilimanjaro three major
vegetation formations were differentiated: the forests of the montane zone,
the forests of the subalpine zone, and the subalpine Ericø bush.
.= d
'¿; e
o -.
!.1 o
l¡ì -t
tr{ O
4.2.1 Lower montane zone
Between 1600 and 2000 m asl Agauria-Ocoteø forests (community 1) repre-
sent the dominant vegetation. Tree species typical of these forests are Agøu-
riø saliciþliø and Myrica salicifoliø, together with Møcørangø kilirnønjaricø
and Syzygiøm guineense. Erica excelsa appears in these forests with remar-
kable constancy in the tree layer. Agauria and Myrica are also components
of the subalpine forests.
4.2.2 l[,|.iddle montane zone
The very dense forests of the middle montane zone between 2100 and
24OOm (communities 2 and 3) lack Erica excelsø more or less completely.
Due to an extraordinarily high humidity the African camphor tree Ocotea
asamba.rensis producing maþstic trees, has im highest abundance in this
altitudinal zone. This holds also for a tree fern (Cyathea rnanniãna) and a
great variety of other pteridophytes, especially filmy ferns (Huur', in press
Table 1 shows a survey of the major forest units of the southern slopes
of Mt. Kilimanjaro. To'provide more information about the ecological
range of Ericø excelsø and-its significance in the forest all {orest types above
160Ó m are represented. For differentiation between the subalpine Ericø
excelsa. communities and the adjacent ericaceous scrub' subalpine Erica tri-
mera and E. arboreø scrubs are shown as well.
In Table 1 the vegetation units are arranged along an aldtudinal gradient'
The floristic compoiition of the plant communities, however' is also deter-
mined by soils, humidity and successional stage of the vegetation. \ùøith
respect to the occurrence of Erica excelsa in the forests of the study area,
4.2.3 Forests of the upper montane zone
Community 4 is a representative of the forests of the upper montane zone.
Prominent species are Ocotea usamba.rensis, Xymalos ntonospora, Podocar-
pus løtifuliui and llex rnitis as well as the shrub Psycbotria qtathicølyx.The
'^urr^gá tree height is significantly lower (29 m) thán in the middle montane
foresti. The light climate of these forests favours the light-demanding Ericø
excelsø, whictrwas recorded with a high constancy of 38 %" tn the ree layer.
Pod.ocørpus latifulius is the dominant tree of another type of the upper
montane forests (community 5), where trees and shrubs typical of lower
altirude-forests (e.g. Ocotea usømbørensis and Lasinnthas þilimøndscbari-
cus) are absent. È.ð"nr. of the high altitude (this forest type extends up to
over 3000 m) the average tree height is only 24 m (Table 1) and Erica excelsa
trees, not reaching that high, are present with a constancy o{ g+y".
Community 5 sharply borders community 6 without any'transition.
Vhile Ericø excelsa is well present but subdominant in communities 1-5,
it is the only dominant tree species of community 6. Vhile Erica excelsa is
still single ,í.--.d in the Podocarpus forests the individuals of community
6 show ramification right from the ground level. The branches are predomi-
460 A. Hemp E¿ E. Beck
nantly vertically orientated and thus render a v-shaped habirus to the trees
(cp. Fig. 5). Montane tree species such as Podocarpus løtifulius, Ilex mitis
and Scbefflera aolþensü are frequent companions but do not produce big
trees. Théy r¡\¡ere mosdy encountered as juveniles.
4.2.4 Subalpine zone
The uppermost forest belt on the southern slopes of Mt. Kilimanjaro, pro--
ducinfi ìhe upper tree line, belongs to the subalpine zone. The pâttern of
soeciei and thê veeetation structuie of the forest communitv 9 with domi-
ì.ant Erica excelsa"trees are similar to community 6. But .tîlik" the latter
the accompanying species from the montane forests are rather rare and
both covef abúndãncì and species numbers of vascular epiphytes are consi-
derably lower (Table 1). This rype of an Erica forest resembles the þcopo-
dium cløztatum variety (Alchernilk fischeri stbcommunity) of the Ericø ex-
celsa-Rubas steadneri var. aberens¿i community reported by ScHrurrrr
(1991) for the Aberdares Mts.
Between 3100 and 3500m asl Ericø excelsa normally does not exceed a
height of 10 m but nevertheless produces groves (community 11) in which
the montane forest species lack completely (Fig. 5).
\ü(/ith increasing altitude these patches of Erica excelsø forests become
gradually replaced by sclerophyllous bush vegetation formed by Ericø ar-
borea and E-. trimera which grow up to an altitude of 3900 m (community
12, Fig. 6). At 3500 m groves of Erica excelsa ffees were seen only in the
shelter of big rock faces.
4.2.5 E. excelsa as a constituent of azonal forest types and of
forests with special edaphic conditions
Together with Hageniø abyssinicø, Rapanea melønophloeos and Prunus
fficana, Erica excelsa f.orms the shrub as well as the tree layers in the
izonal riverine and mostly gorgeJike forests of the subalpine and the upper
montane zone. In this type of vegetation (communiry 7), the characteristic
species are the conspicuènt giant rosette plants Dendrosenecio johnstonü
sip. johnstonii and Lobelia ãecþenü ssp. decipient, representatives of the
typical afroalpine zone.
- -In a wide ïalley at 2870 m altitude, south of the "Shira cathedrale", an
Erica excelsa foreét with trees about 6 m high was encountered in a Carex
rnonostacliyø bog (community I2). There the Carex rufts reach a height of
2 m. Although only represented by one relevé in Table 1 this well-defined
community covered several hectares of the valley bottom.
A broaá-leaf forest type, encountered mainly on basaltic screes and all-
uvial soils at high aldrudes (up to over 3200 m asl), is presented-as commu-
nity 8. The treã canopy repreiented by Erica excelsa, Prunus øfricdna, Rø-
poiro melønophloeoiínd i{agenia abyssinica, has an average height of 18 m
(Table 1).
Erica excelsa as â component of Mt. Kilimanjaro's forests 46t
Fig. 5. Subalpine Erica excelsa forest at 3200 m altirude (community 11) in Southwest
Kilimanjaro, February 1997.
Fig. 6. Ericaceous shrubland in the Shira area in February 1997 before the fire of April
1997. Remnanæ o1 Erira excelsa forest were found up to 3500 m asl.
5 Discussion
5.1 Phytosociology of the forest communities dominated by Erica ex-
Heonrnc (i951) attributed the upper vegetation on Mt. Kilimanjaro into 3
belts, which he considered as "climax formations": The montane forest belt,
the ericaceous belt (consisting of the moorland zone and the ericaceous shrub
zone) and the alpine belt. KNepp (1965,1968) established (without presenting
vegetation tables) the phytosociological class H a g eni e te a abys s i ni c ae
(corresponding to HEonnxc's Hagenia-Hypericum zone) and a second class
Alchemillo-Ericetea termed lateron (1963) by the same author Cliff or-
tio-Ericetea, corresponding with Fleo¡anc's ericaceous shrub zone).
MrnHn E¿ Mrnnr (1994), however, showed for the Bale Mts. that the erica-
ceous belt, as defined by HEonnnc (1951), consists of several groups of com-
munities, which can be associated with different vegetation units of a high
phytosociological rank and that the character species of KNRpp's class Clif -
fo r ti o - E r i cetea are not useful for their study area in Ethiopia.
Based on the altitudinal distribution of pteridophytes (HErurr,, in press a)
and trees (Hrun, unpublished data) the boundary between the upper mon-
tane and the subalpine zone (corresponding in general with the boundary
between the "Oberer Höhenwald" and the "subalpine Region" of ENcrrn
(1,925), and that of the ericaceous belt of Hnosrnc (1951), respectivel¡ is
located at around 2800 m. This boundary is indicated in Table 1 by the
large group of montane forest species. The lowermost forest type which is
dominated by E. excesta and which is represented by community 6, clearly
belongs to the montane forest communities. In the second forest type do-
minated by Erica excelsa, represented by community 9, these montane spe-
cies are completely absent in farow of species typical o{ a Hagenia forest.
Thís Ericø forest type belongs to the subalpine forests. \,Øhether these fo-
rests should be separated as an own class ("Hagenietea", Klqepp 1965, 1968,
BussivreNN & Brcr 1995) or as a sub-type of the forest belt ("Høgenia-
Hypericum zone", HBo¡rnc 1951), is a matter of further investigation.
More difficult is the phytosociological classification of the scattered Erica
excelsa groves (community 10 and 11). The only wide spread montane and
subalpine forest species, occurring in these communities,is Rapønea melø-
nophloeos. Subalpine forest species, such as Xipbopteris flabelliforrnis or
Senecio q)a.neus suggest an affiliation with the subalpine Hagenia forests.
The Erica trimera and E. arborea scrub (community 12) is clearly diffe-
rentiated from all other communities by invaders from the alpine Helicbry-
sam dwarf-shrub lands, such as Helicbrysam newü or Pentøscltistis boras-
sica. This part of the ericaceous belt probably could be given the rank of a
vegetation class, although general character species are difficult to idendf¡
as pointed out by MrrHr E¿ Mrpnr (1,994).
In summar¡ the Ericø excelsa forests represent a sub-unit of the upper
montane and the subalpine forests and are floristically clearly differentiated
from other ericaceous communities.
Erica excelsa as a component of Mt. Kilimanjarot forests 463
5.2 Factors effecting the altitudinal distribution of Erica excelsa
Many environmental factors change with the altitude, in particular tempera-
ture'and precipitation. In the forest plots of the study area the mean soil
temperature (measured at 2Q cm depth where the bulk of the roots are
found) decreased from 16 "C at 1800 m to 9 'C at 3200 m and precipitation
increases up to 2400 m asl and then decreases again with increasing altitude.
As these fãctors change more or less gradually with elevation, a gradual
change in tree species ðomposition could be expected. FIowever, rather ab-
ruDr ¿hanses inihe floristið composition are obvious in the forests berween
ZSbo-¡OOö m asl, which still receive a high amount of rain (about 1500 mm).
The factor causing these discontinuities on Mt. Kilimanjaro is fire' Du¡i1g
the end of the dry season in February and March the whole ericaceous belt,
which covers neárly 300 km2, is usuãlly so dry that it may easily be set on
fire by people (espécially by honey colíectors,'cp' I-{rvrn 1'999) lighte-
ning.'Big iires, devasting iarge areas of the subalpine forests have been
repeatedl-y reported (e.g. Mnvrn 1890, Becr et al. 1986, Bror 1'99-9,Fi1.7).
The fires'of ihe vears |gge and 1,997 d,estroved nearlv 90 km2 of the Ericø
forests and shruÉ land of the southern slopeá of Mt. Kilimanjaro. Therefore
special attention is being paid in the following discussion of the distribution
of Erica excelsa to the influence of fire.
5.3 The distribution of Erica excelsa in the forests of Mt. Kilimaniaro
On Mt. Kilimanjaro Erica excelsa forms monotonous stands in the subal-
pine zone, but also a component of the forests of the lower and uPPer
^-otrt"ne zone. However, it is rare in the central montane zone (Fig. 3 and
a). This distribution results from the pattern of rainfall on the one hand and
the occurrence of fires on the other. Fires, mostly lit by man, are frequent in
the subalpin e zone and also, less frequent in the lower montane zone bet-
ween 1600 and 2OOO m. Here, in the Agauria-Ocorca forests and in the
degraded forest edges contacting the cultivated areas fires are almost exclu-
sivËly lit by man (F"ig. S). Frequðntly burnt areas in this altitude are obvious a puíe Erici büthico*tit"nity ta) whose vegetation structure is simi-
lar to the subalpine Erica excelsa communities.
Some trees such as Agauriø salicifolia (Ericaceae), Myricø salicifolia or
Høseni.a. abyssinicø and some ferns lìke the fire-resistant Pteridium aquili-
nnñ. ^nd Diyopteris pentheri (Hriran, in press a) show the same attituãinal
distribution'as'Ericø excelsa with a gap the central forest belt.
In the central montane zone betweèn 2100 and2400 m asl fire is uncom-
mon due to the extremely high precipitation and Erica excelsa is rare in
these forests, too. There ii onþ one eiception, represented by community
4a (2300 m asl), on the drier south-eastern slope above Marangu. This com-
munity has a completely different stn¡cture. The tree layer is dominated by
Erica'excelsa, Schifflerâ volþensü and Xymalot *oroíporo. This commu-
nity is interpreted-ãs an Ericø excelsa serial stage developing after a local
A. Hemp & E. Beck
464 A. Hemp E¿ E. Beck
Fig. 7. Burning subalpine ericaceous shrubland betv¡een Great and Little Barranco on the
north-eastern slope of Mawenzi (February 2000), lit by poachers.
Erica excelsa âs â conìponent of Mt. Kilimanjaro's forests 465
man-made fire in the Podocatpus-Ocotea forest. Young Podocarpus and Ilex
trees indicate that the original type of forest is on the way to recover.
5.4 Erica as a major component of the upper tree line on Mt. Kiliman-
The explanation of the presence of the upper tree line on tropicai high
mountains is still a matter of controversy. Tnorr (1959) suggested that a
combination of edaphic factors such as soil type, porosity or cornpactness
is crucial for the existence of trees at or above the general tree line. \Ø¡r-
rER E¿ MprrNa (1969) on the other hand, attributed the major effect to the
climate, hypothesizing that a minimum average annual soil ternperature of
7 "C may 6e the limit for tree grov/th. In the South American Andes the
7'C isotherme coincides roughly with the 3300-m-isohypse. However,
with respect to Africa, MIrHE & MIBHe (1994) reported ericaceous forest
in an aria subjected to an average annual soil temperature of 3.3"C at
4000 m altitude in the Bale mountains (Ethiopia), whereas on other East
African mountains, regions whose annual average temperatures are well
above 7 oC, carry only a "moorland vegetation" of grass and sedge tus-
Fig. 8. Burned grass- and shrubland (arrows) at the lower forest boundary (the boundary
of the forest reserve is indicated by a planted Eucalyptus-line) above Nt.waa, South Kili-
manjaro (February 2OOO). Such fires, lit by farmers, often spread into the forests as can
be seen by the burned forest patches in the left background (open arrows).
Fig.9. Erica excelsa iorest islands in the Podocarprs forests. Above 3100 m asl Podocatptts
forests are restrictecl to the lower and wetter parts of the slopes v¡hereas Erica forests
(arrow) cover the ridges. The boundary between these two forest types is very sharp as
a result of fire.
466 A. Hemp 8¿ E. Beck
socksl (FnöHrrcH E¿ \üØrrrnn 1977, see also \{¡rNrcnR 1981, KönNan 1999).
MIrHr E¿ MrnHB (1994, 1996) therefore suggested that the narural upper
tree line in tropical mountains is above 4000 m altitude and explained ihe
existence of an upper tree line at lower elevations and of the isolated groves
above the extant forest line as resulting from the activities of man. As repor-
ted by Krure (1920) the forest extended on the south-v/estern slope of Mt.
Kilimanjaro above Machame at the beginning of the rwentieth century up
to 3600 m while an open Erica lorest was reported at altitudes of over
3900 m.
On Mt. Kilimanjaro stmcture and composition of the subalpine vegeta-
tion is strongly influenced by recurrent fires. As can be concluded from
the stunted form of the Ericø plants, the frequency of fires increase substan-
tially with increasing altitude and exposure to wind. Thus on the Shira
plateau in an altitudinal range of ¡SOO to 3850 m bushes of Erica trirnera
hardly reach 2-3 m height, the branches sprouting from short, crippled
and charred stumps whose form indicates frequent burning and subsequent
resprouting (Brcx et al. 1986). On the rocky outcrops of the Shira ridge
and the Shira cathedral (Fig. 6) at altitudes berween 3600 and 3800 m, indi-
vidual ffees of Erica, approximately 6-7 m high, were seen which due to
the scattered and scarce vegetation on the rocks have never been hit by fire.
These indicate a potential natural tree line, formed by Erica, even at higher
altitudes. At the top region the more less pure Ericø excelsø forest (com-
munity 11) covering the upper areas of the southern slopes of Shira consi-
sted of up to 5 m high trees which branched right from the base. '$7e assume
that the branched growth form resulted from burning as on the Shira pla-
teau. Flowever, a diameter of the branches of up to 5 cm and the lack of
crippled stumps indicate long growth periods without any disturbance.
This part of the Erica forest has recently been burned. New rwigs sprouting
from the stem bases were seen on about 80/" of the individuals. Resprou-
ting of burned Erica excelsa contrasts with reports by ScHrvrrrr (1991) and
ScnNonnrr et al. (cit. in Scr¡r"rrrr 1991). Regeneration of Erica from seeds
is at least on Mt. Kilimanjaro of less importance, due to the high potential
of resprouting, even and especially after repeated burning.
However, extensive. rejuvenation by.seedlings of burned. Erica excelsa
was seen in swampy locations where the fire had presumably promoted
seed formation or ripening and simultaneously had eliminated the dense
sedge and grass cover of the soil, thus providing an ample and nutrient-rich
substrate for germination and establishment of the seedlings.
As shown by ScHr'anr (1991), MrrHB 8¿ MIrr¡n (1994), BussueNN E¿
Bncr (1996) and Lelqcs et al. (1997) fire promotes the natural regeneration
of Ericø trimera, Janiperus procera. and Høgenia øbyssinicø and therefore
appears as a major factor for the persistence of the upper montane and
Erica excelsa as a component of Mt. Kilimanjaro's forests 467
subalpine forests of the East African high mountains. The fires of 1996 and,
1997, which burnt the old Ericø excelsa forests, pushed the upper forest
line on the southern slopes of Mt. Kilimanjaro downwards by about 300 m'
Therefore the intensityãnd frequency of fire is of greater influence on the
upper forest limit on Mt. Kilirnanjaro than the microclimate including the
subsoil temperatures.
The dominance of Ericø excelsø in the apparent climax forest near the
actual upper forest line contrasts with the situation on Mt. Kenya and the
Aberdaie Mts., where the mesophytic Hagenia abyssinicø dominates the
subalpine forests. This difference is presumably due to the high moisture
of the upper forest of these mountains.
Abovä 2800 m Erica becomes dominant, although Podocørpas forests
with broad-leaved rees2 would represent the natural climax vegetation of
elevations up to 3100 m asl. Podoiørpøs forests were often found on the
lower and mbre humid slopes and in the valleys whereas Ericø forests cover
the ridges (Fig. 9). In some places, especially in the drier Southeast, Ericø
excelsalotestJextend down io 2450 m. The majority of these Erica excelsa
forests consist of multi-stemmed trees of apparently similar age (Fig. 10 (3)
and 11), suggesting simultaneous sprouting after a fire, Erica forms clear
growth ringï Stem disk analysis of an Eriia excelsø stand showed at least
ó*o ofste- age. Almost all stems were24-25 years old but the base
of the branched inãividuals and very few big individuals showed an age of
about 150 years. Another indicator of the last fire aPproximately 20-30
years ago *ere l"rg. decomposing logs oI Podocarpuî,-Hagenia and other
iree speiies as welfas young individuãls of Podocarpry a1d Hagenia about
20 yeàrs old. Further indications of periodic fires aré clearly separated char-
coJl horizons in the topsoils of Ericø stands and older tpótt óf charcoal in
the subsoil of Erica and Podocarpøs forests. \Øithout disturbance by fire,
Erica excelsa grows as a single-stemmed tree up to- 28 m high and with a
DBH of 70cm. Such trees have been observedin Podocarpus forests. There-
fore, the existence of pure Erica-forests berween 2600 and 3100 m asl
should depend on fire rãther than on other climatic factors. As mentioned
above, firãs are frequent below 2000 m and above 2800 m. In years with
normal .weather conditions only patches of the Erica forests and heather
shrublands burn. In exceptionaily dry years even adjacent patches of the
Podocarpus-forest (forest community 5, Fig. 10 (1)) may catch fire. Most
of the tiees are killed by such calamities except Agauriø sølicifoliø and Erica
excelsa, which are able to resprout from the superficially charred stumps
with the next rains. As a result of such big fires sharp boundaries between
Erica and old Podocarpøs forests develop. Heurrrou 8¿ Prnnorr (1981)
found a similar sharp b-oundary between Erica excelsa thickets and montane
forests on Mt. Elgón. If the iime-period between successive fires is long
enough to allow monta.te forest treès to recover- or regrov¡ from-s.eeds (like
in for"est community 6, Fig. 1O (2)), Podocørpus forest may reestablish itself.
Scattered charcoal pieces were found in the deep subsoil under old Podocar-
I The moorlands of Mt. Kilimanjaro, especially those of the southern slopes are usually
not swampy, but covered with mssocks of Festaca obtarbans, Roeleria. capensis and
Bulbostylis atrosanguinea. Therefore in this case the misleading term "moorland"
should be replaced by "subalpine grassland". 2Even Podocarpils as a Gymnospermae does not produce needlelike leaves.
lm ,\-,' ;
','A .i
Sv Ee As
ØEe Ee Ee As Ee
t, Rm PI
@Ee Ee r" ?o Ee Ee
The higher the percentage Erica is, the higher is the risqué of another fire and the
more difficult is the re-establishment of a Podocarpus foresr (whire arrows). As: Agøuria
salicifolia,F,e; Erica excelsa,Im llex mitis,Pl: Podocarpus latiþlius,F.:m: Rapanea meh-
nophloeos, Sv: Sch efflera volþensü.
Fig. 10. Regeneration of the canopy in upper montane and subalpine foresrs. 7: podocarpus
forest (community 7),2: Erica excelsaforestwith montane foresispecies (community 8), 3:
Pure Erica excelsa Íorest (community 10). vhether the Podocarpus forest is replaced by an
Erica forestìs a question of intensity and frequency of fire (black arrow).
470 A. Hemp E¿ E. Beck
pus forests. In case of fires recurring with only short intervals, monrane
forest tree species are damaged lethally and Erica becomes the dominant
species. The_occurrence of an Erica forest is obviously depending on rhe
intensity and frequency of fires. During longer periods of dry climate the
boundary between Erica and Pod.ocarpus foresti is consecurively pushed
downwards while it climbs again in a series of wet years. Since even fresh
wood of Erica burns very well the danger of recurring burning is increased
with increasing proportions of Erica in a forest. Once Ericà excelsa has
established, regeneration of a broad-leaved forest becomes more and more
Pollen diagrams of other East African mountains supporr the incidence
of a dry climate from 25000-12500 B. P. The vegetation belts of rhe moun-
tains were depressed by 800- 1000 m at this time and the dry climate favou-
red drought resistant plants, tolerant of dry conditions, such as Artemisiø,
Antbospermum, Stoebe and Myrica (LrNo & MonnrsoN 7974). Erica cer-
tainly played an important role as well. Charcoal horizons found in subsoils
of lower and middle montane forests in several dm depth of the study area
suggest that during the ice age the ericaceous belt exteìded or was depres-
sed about 1000 m lower than today.
Due to the open canopy of the Erica forest the microclimate changes as
is obvious from the composition of the herb layer. Monrane forest sþecies
disappear and light-demanding species like Lycopod,iurn cløvøtam ànd a
variety of- mosses become dominant, which can not successfully compete
in shidy forests (Fig. 1O (3) and L1 demonstrare this successiotrai step).
5.5 The floristic composition of the potential upper forest zone on
Mt. Kilimaniaro
In contrast to Flrosrnc's (1951) definition of the Ericaceous belt, Erica
excelsa stands and moorland vegetation neither are gradually merging into
one another nor does the moorland zone form a complete girdlè aiound
the mountain. Rather it is restricted to the sourh-easre¡n sloþes. There, ar
an àldtude of 2800 m Ericø excelsa stands and the "moorland" iussock vege-
tation producg very abrupt boundaries (Fig. 12). Tree-islands consistingof
a core of Podocarpøs forest are surrounded by a fringe o{ Erica trees and
various shrubs, such as Conyza vernonioides and Hypericurn revolutam.In
this area, Podocarpus forest (community 5), Erica forest (community 6)
and subalpine grassland occur at the same altirude (Fig. i3). Substantial
microclimatic differences e.g. in precipitation, frequency and strength of
frost and soil temperatures can thus ruled out as an explanation óf the
extant course of the upper treeline. Ratheq recurrent fires may be the cru-
cial factor pushing the forest back from the subalpine to lower and moister
regions. The above-mentioned forest islands are therefore to be interpreted
as remnants of the former forest, rather than outposts of the recent ones.
This idea is corroborated by the srunted growth foims of the woody species
of the islands' fringe. Similar observations were made by Dowsrrr-Lr-
MAIRE (1985) in Malawi. On the Nyika plateau sharply bordered forest
Erica excelsa as a component of Mt. Kilimanjarot forests 471.
10 15m
Fig. 11. Profile (16 x 2 m) of a subalpine Erica excelsa forest at 2930 m asl (community 9)'
A: Antbospermum usambarense, As. Agauri.a salicifolia, P: Pittosporatn spec.; trees and
shrubs not labelled are Erica excelsa, crosses mark dead trunks.
472 A. Hemp & E. Beck
Fig. 12. Sharp boundary line between subalpine forest (background) and moorland vege-
tation in Southeast Kilimanjaro at 2700 m asl.
patches of Agaaria and Pbilþpiø (Erica) forests in a surrounding grassland
are considered as palaeoclimatic relicts. Krssrpn (2000) found a similar si-
tuation near the tree line in the Bolivian Andes where groves of Polylepis
grow in an altitudinal zone dominated by fire-maintaineã grasslands. í" Ëtr.
mountain fynbos of South Africa, comparable forest patches are remnanrs,
which have been isolated by the frequent fires (ver.r \ØrrceN et aL.1992).
As far as Erica excelsa is concerned, recurrent burning can decrease its
size and change its habitus from a monocaul tree to a branched bush, but
would not exterminate the individual. Sprouting of charred individuals of
other subalpine trees could not be observed. If an¡ rejuvenation of these
species may take place from enhanced seed producdon and germination
after a fire, as has been described by LeNcr et al. (1997) for the Kosso-tree
(Høgenia abyssinica). Høgeniø, however, is a dioecious tree and male and
female individuals do not always grow in vicinity. Groves of males were
encountered e.g. in the Aberdare Mts. and may also be met on the other
East African Mts. Conceivabl¡ a viable seedbank cannot be expected in
those groves and a strong fire must eradicate Hagenia in these areas.
From such considerations we conclude that the original vegetation of
the present subalpine tussock grassland with its undulating borderline and
the tree islands must have consisted of a mixed type forest in which Ericø
was a component, but not the dominant tree species. 'Síith increasing alt-
itude the share of Erica may also have increased and finally representatives
of this genus may have formed the upper treeline at altirudes close to 4000 m
resembling community 11, as it has been described for the Ethiopian Bale
Mts. (MrrHr 8¿ MrEHr 7994,1996).
Erica excelsa as a component of Mt. Kilimanjaro's forests 473
Acknowledgements. We gratefully acknowledge support by the Deutsche Forschungsge-
meinschaft for grants to A. H., by the VW Foundation for a grant to E. 8., the Tanzanian
Commission for Science and Technology for permitting research (permit no. CST/RCA
96/44/2040/98) and rwo referees as well as Dr. Reiner Zrr¿Ir¡rnIr¡e¡{N, Bayreuth, for valua-
ble comments on the manuscript.
'\ü(ie further thank the keepers of the East African Herbarium, Nairobi, Dr. Beatrice
Knevore and Kew Herbarium, England, Prof. Dr. Ov¡Ns for permission to study their
collections. For support of our field work we owe gratitude to the Chief Park \Øarden
of Kilimanjaro National Park, Mr. Morn¡Nn and Mr. Musur, Moshi.
Agnew, A.D.Q. Ec Agnew, S. (1994): Upland Kenya Vildflowers. - East Africa Natural
History Society. 374 pp.
Alm, C. G. & Fries, T. C. E. (1927): Monographie der Gattungen Philippia Klotzsch,
Mitrastylus nov. gen. und Ericinella Klotzsch. - Kungl. Sv. Vet. Akademiens Handlin-
gar.4 (4)z 1-49.
Beck, E., Scheibe, R. Ec Schulze, E. D. (1986): Recovery from fire: Observations in the
alpine vegetation of western Kilimanjaro (Tanzania). - Phytocoenologia 14: 55-77.
Beck, 8., Scheibe, R. & Senser, M. (1983): The vegetation of the Shira plateau and the
western slopes of Kibo (Mount Kilimanjaro, Tanzania). - Phytocoenologia 11: 1-30.
Fig. 13. Mosaic of Podocarpus forests, Erica excelsa forests and Festøca obtarbans tussock
grassland (recently burned) in the moorland zone in Southeast Kilimanjaro at2T0Qmasl
(Jan'tary 1997).
474 A. Hemp & E. Beck
Beentje, H.J. (199a): Kenya trees, shrubs and lianas. - National Museums of Kenya,
Nairobi. 722 pp.
Blot, J. (1999): The incidence of forest fire in Kilimanjaro. - In: Mount Kilimanjaro:
Land Use and Environmental Management. - French Institute for Research in Africa.
IFRA Les Cahiers 16: 85-86.
Braun-Blanquet, J. (1964): Pflanzensoziologie. - \lien. 865 pp.
Bussmann, R. \ø. s¿ Beck, E. (1995): The forests of Mt. Kenya (Kenya), a phytosociologi-
cal synopsis. - Phytocoenologia 25: 467 -560.
Coutts, H. H. (1969): Rainfall of the KilimanjaÍo ^rea. - 'W'eather 24:66-69.
Dierschke, H. (1994): Pflanzensoziologie: Grundlagen und Methoden. - Ulmer, Stuttgart.
683 pp.
Dowsett-Lemaire, F. (1985): The forest vegetation of the Nyika Plateau (Malawi-Zambia):
ecological and phenological studies. - Bull. Jard. Bot. Nat. Belg. 55: 301-392.
Engler, A. (1925): Die Pflanzenwelt Afrikas Vl. Die Vegetation der Erde IX. - Leipzig.
Fries, R. E. Er Fries, C. E. (1948): Phytogeographical researches on Mt. Kenya and Mt.
Aberdare, British East Africa. - Kungl. Sv. Vet. Akademiens Handlingar. 25 (5):1-
Fröhlich, V. & Viller, M. (1977): Bodentemperatur und obere \ùüaldgrenze. Vorläufiger
Bericht über Untersuchungen an den Hochbergen Ostafrikas. - Die Erde l08- 347-
Flaines, R. \í. S¿ K. A. Lye (1983): The sedges and rushes of East Africa. - East Africa
Narural Societ¡ Nairobi. 404 pp.
Hamilton, A. C. & Perrott, R. A. (1981): A study of altitudinal zonation in the montane
forest belt of Mt. Elgon, Kenya/Uganda. - Vegetatio 45t lQ7-125.
FIammen, T. v. d., Mueller-Dombois, D. & Litle, M. A. (1989): Manual of Methods for
Mountain Transect Studies. - IUBS, Paris.
Hastenrath, S. (1973): Observations on the periglâcial morphology of Mts. Kenya and
Kilimanjaro, East ,A.frica. - Z. Geomorph. N. F. 16t 1,61-179.
Hedberg, O. (1951): Vegetation belts of the East African mountains. - Svensk Bot. Tids-
krift.45: 140-202.
- (1964): Features of afroalpine plant ecology. - Acta Phytogeographica Suecica 49: 1-
Hemp, A. (1999): An ethnobotanical study of Mt. Kilimanjaro. - Ecotropica 5z 147 -165.
- (in press a): Ecology of the pteridophytes of the southern slopes of Mt. Kilimanjaro.
Part I: Altitudinal distribution. - Plant Ecology.
- (in press b): Ecology of the pteridophytes of the southern slopes of Mt, Kilimanjaro.
Part II: Habitat selection. - Plant Biology.
Hemp, 4., Hemp, C. E¿'lØinter, J. C. (1999): Der Kilimanjaro - Lebensräume zv¡ischen
tropischer Hitze und Gletschereis. - Natur und Mensch 1998:5-28.
Hubbard, C. E., Milne-Redhead, E., Polhill, R. M. E V. B. Turrill (eds. 1952-): Flora of
Tropical East Africa. - Crown Agents, London.
Kessler, M. (2000): Observations on a human-induced fire event at a humid timberline in
the Bolivian Andes. - Ecotropica 6:89-93.
Klötzli, F, (1958): Ztr Pilanzensoziologie des Südhanges der alpinen Stufe des Kiliman-
jaro. - Ber. Geobot. Inst. Rübel 1957:33-59.
Klute, F. (1920): Ergebnisse der Forschungen am Kilimandscharo 7912. - Berlin. 136 pp.
Knapp, R. (1965): Pflanzengesellschaften und höhere Vegetations-Einheiten von Ceylon
und Teilen von Ost- und Central-Afrika. - Geobot. Mitt. 33: 1-31.
- (1968): Höhere Vegetations-Einheiten von Äthiopien, Somalia, Natal, Transvaal, Ka-
pland und einigen Nachbargebieten. - Geobot. Mitt. 56: 1-36.
Erica exceha as a component of Mt. Kilimanjaro's forests 475
Körner, C. (1999): Alpine Plant Life. Functional Plant Ecology of High Mountain Ecosy-
stems. - Springer, Berlin-Heidelberg. 338 pp.
Lange, S., Bussmann, R. rüø. Ec Beck, E. (1997): Stand Srructure and Regeneration of the
Subalpine Hagenia abyssinica Forests of Mt. Kenya. - Bot. Acta lTOz 473-480.
Lind, E. M. 8¿ Morrison, M. E. S. (197!: East African Vegetation.- London. 257 pp.
Lösch, R. & Fischer, E. (1994): Vikariierende Heidebuschwälder und ihre Kontaktgesell-
schaften in Makaronesien und Zentralafrika. - Phyrocoenologica 24; 695 -720.
Meyer, H. (1890): Ostafrikanische Gletscherfahrten. Forschungsreisen im Kilimand-
scharo-Gebiet. - Leipzig. 376 pp.
Miehe, G. Ec Miehe, S. (1996): Die obere \Øaldgrenze in tropischen Gebirgen. - Geo-
graph. Rundschau 48: 670-676.
Miehe, S. 8¿ Miehe, G. (1994): Ericaceous Foresrs and Heathlands in the Bale Mountains
of South Ethiopia. - Ecology and Mant Impacr, Hamburg. 206 pp.
Morris, B. (1970): The ZonaI Vegetation of Kilimanjaro. - African \Øild Life 24t 157-
Naveh, Z. (197\: Effects of Fire in the Medirerranean Region. - In: Kozlowski T. T. &
Ahlgren, C.E. (eds.): Fire and Ecosysrems, pp.401-434. - Academic Press, New
Oliver, E. G. H. (19SS): Srudies in the Ericoideae (Ericaceae). VI. The generic relationship
berween Erica and Philippia in southern Africa. - Bothalia 18,1: 1-10.
Pócs, T. (1994): The altitudinal distribudon of Kilimanjaro bryophytes. - In: Seyani,
J.H. & Chikuni, A. C. (eds.): Proc. XIII Plenary Meeting AETFAI, Malawi, pp.797-
Salt, G. (1951): The Shira plateau of Kilimanjaro. - Geogr. l. 177 1,50-164.
Sarmett, j.D. 8¿ Faraji, S. A. (1991): The hydrology of Mount Kilimanjaro: an examina-
tion of dry season runoff and possible factors leading to its decrease. - In: Newmark,
\Ø. D. (ed.): The Conservation of Mount Kilimanjaro, pp. fi-7A. - IUCN.
Scandrett, E., Pearce, T., Gita¡ H. & Collins, H. (unpublished): Plant communities in
the ericaceous belt of the Aberdare Range (Kenya).
Schmitt, K. (1991): The Vegetation of the Aberdare National Park Kenya. - universitäts-
verlag lVagner, Innsbruck. 259 pp.
Troll, C (1959): Die tropischen Gebirge. Ihre dreidimensionale klimatische und pflanzen-
geographische Zonierung. - Bonner Geogr. Abh. 252 l-93,
lValter, H. E¿ Medina, E. (1969): Die Bodentemperarur als ausschlaggebender Faktor für
die Gliederung der subalpinen und alpinen Scufe in den Anden Venezuelas. - Ber.
Dtsch. Bot. Ges. 82, H.3 / 4:275-281.
Viniger, M. (1981): Zur thermisch-hygrischen Gliederung des Mr. Kenya. - Erdkunde
Address of the authors:
Dr. A. HEup and Prof. Dr. E. Bncr, Deparrment of Plant Physiolog¡ University of
Bayreuth, Universitätsstr. 30, 95440 Bayreuth, Germany.
... Standing at close to 5900 m.a.s.l, the Erica rossii (formerly Erica excelsa) treeline sits at around 3500 m.a.s.l. (Hemp and Beck, 2001). Treelines also exist in South Africa on the Drakensberg (29°S) but are absent from their climatic range due to a lack of substrate on steep cliffs and regular burning events (Körner and Paulsen, 2004). ...
... impacting the treeline location, as forests are often replaced by grassland post-fire (Wesche, 2003 (Hemp and Beck, 2001). The severity of the dieback of the upper montane forest will likely result in the forest stands not recovering for several decades, assuming there are no further disturbances. ...
Full-text available
The position of climatically limited treelines is primarily determined by the duration and minimum temperature of the growing season. An increase in temperature brought on by anthropogenic climate change should see treelines migrate to higher elevations and latitudes. This change may have significant impacts the ecology of alpine areas and affect global carbon storage. This thesis has examined the recent history of treeline migration using remote sensing and meta-data approaches to characterise and interpret the causes of treeline migration and to investigate the likely impacts. First, the methods used to monitor treeline change were examined. This work concluded that dendrochronology provides the most accurate estimates of treeline migration. Vegetation transects are also identified as a useful approach to track treeline movements and assess seedling establishment and mortality over time. However, both methods are labour-intensive, time-consuming, and require field access. Hence, remotely sensed data are more suitable for efficiently tracking treeline changes over large areas. These approaches are particularly suitable where site-access is difficult, and where high-resolution imagery is available for several decades. This thesis summarises the current state of the global treeline migration across all continents, where a combination of these methods has been used to track forest expansion over time. Latitudinal and altitudinal treeline migration was assessed at a global scale through metadata analysis. A database was created covering over 477 study sites derived from 142 published articles. Across these study sites, an increase in the uppermost elevational or latitudinal position was detected at 66% of treelines. Not all treelines have responded to climate change by range expansion. This lack of response could be due to either non-climatic barriers or due to the slow rate at which some forests may be able to track changes in their habitat. Importantly, climate change has not caused a uniform increase in temperatures, and some regions have seen more rapid expansion of habitat than others. Latitudinal treelines are already migrating at particularly high rates, with the average horizontal expansion of latitudinal treelines being 47 meters per year. Such a rapid migration can have potentially devastating impacts on the carbon storage potential of the tundra regions, as the establishment of vascular plants can reduce water tables and increase the metabolization of soil carbon. The correlation between climate change and treeline movements was assessed using logistic regression modelling. The model revealed that the rate of temperature change during the autumn, particularly in October, was a significant predictor of treeline movement in the Northern Hemisphere. More specifically, increased minimum temperatures corresponded with treeline migration. However at warmer maximum summer temperatures, treelines were more likely to remain stationary, most likely due to drought effects in shallow alpine soils. While statistically significant correlations were identified at global scales, there were also substantial regional correlations of treeline movement regarding warmer temperatures. For example, Scandinavian treeline migration patterns could not be correlated to climate change, while latitudinal treelines in North America responded strongly to warmer minimum temperatures during spring, summer, and autumn. Hence, while globally significant correlations exist these may not always be applicable at the regional or local scale. While there were insufficient data on treeline migrations from the southern hemisphere, southern hemisphere treelines have shown less vagility than their northern hemisphere counterparts. An overall limitation of treeline migration of temperate treelines in the southern hemisphere may be due to the comparatively poor adaptation of formerly rainforest taxa to cold climates. This could explain why southern hemisphere treelines have responded more slowly to warming climates and why they form at a warmer threshold than northern hemisphere treelines. Remote sensed data, specifically repeat photography was utilised to track treelines in the Arthur’s Pass area, central South Island, New Zealand from 1960 until today. New Zealand was a region where no treeline changes had been detected. Small scale changes were located from sites investigated in this study, indicating a localised spread of trees above the treeline. This investigation showcases the importance of large-scale monitoring to capture changes of treeline elevation, as treelines may show very different responses to warming across short spatial scales. The results presented in this thesis indicate that provided the right climatic circumstances, climate-limited treelines are capable of tracking their suitable habitat across all spatial domains. The understanding of treeline formation and the driving forces behind alpine and arctic forest expansion has increased drastically over the past decades. While the increase in forest cover can lead to reduced alpine habitats, the capacity of vegetation to track their suitable habitat is critical under rapidly increasing temperatures. Alpine and arctic trees, which are some of the most slowly growing forms of vegetation, are so far showing a remarkable resilience and adaptive capacity despite these regions being the most strongly impacted by climate change.
... The long-term fire and disturbance ecologies of the different montane forest types have yet to be fully characterized across the highlands of eastern Africa. Changing fire frequencies on mountains of eastern Africa are key processes that influence vegetation composition and structure (Wesche, 2003) with many effects on patchiness and ecotonal transitions (Hemp and Beck, 2001;Hemp, 2005;Gil-Romera et al., 2019;Courtney Mustaphi et al., 2021a). It is unknown what disturbance ecology processes promote the persistence or hindrance of the spatial and temporal patterns of tree stand compositions over multidecadal, centennial and millennial time scales, and how these processes interact with other ecological processes. ...
... Fire regime changes facilitate ecological changes and could be a consideration for allocating forest fire suppression effort, fire prevention or areas for non-intervention on the mountain. Changing fire regimes at the interfaces of primary forests and agroforestry causes changes at the ecotonal edges (Thijs et al., 2014;Wekesa et al., 2019;Cardoso et al., 2021) as well as at indigenous ecotonal zones within protected areas (Wesche, 2000;Wesche et al., 2003;Hemp and Beck, 2001). Future projects investigating palaeofire in Afromontane forests should be co-produced with land users, and land management and academic stakeholders to align priority research questions (Seddon et al., 2014;Chazdon et al., 2017) and facilitate the longer-term deployment of sediment and charcoal traps in the field to improve the calibration of sediment-charcoal studies to the level of development available in temperate forests (for an example see Adolf et al., 2018). ...
Full-text available
Past forest fire events and fire frequencies are reconstructed with sediment–charcoal records at lake catchment spatial scales. Few quantitative palaeofire analyses exist in tropical montane forests, where fire return intervals are long (decadal and centennial scales) because of the infrequency of fire weather and fuel conditions. Fire return intervals are a key characteristic of fire regimes and changing fire frequencies rapidly alter land cover compositions and vegetation structure. Charcoal records from small lakes with relatively small catchments covered with dense forest provide an opportunity to reconstruct low‐frequency, high‐severity fires through a time series decomposition approach to identify charcoal peaks above a varying background rate as a proxy for palaeofire events. The sediment core from Rumuiku wetland on Mount Kenya, equatorial eastern Africa, accumulated a nearly linear age–depth model and provided a high temporal resolution (10 years cm–1) sieved charcoal count record (>125 µm). Pollen analysis showed a significant change in montane forest assemblage occurred at 21 200 cal a bp from a montane forest with abundant Podocarpus and Juniperus to a forest with more abundant Hagenia. This change in forest altered the vegetation composition and structure with concomitant changes to the fire regime. Forest biomass in the Hagenia forests decreased and it is likely that fire activity qualitatively changed toward lower intensity and lower severity fires. The quantitative fire event reconstruction focuses on the interval from 27 000 to 16 500 cal a bp and the older montane forest that experienced higher severity fires from 27 000 to 21 200 cal a bp, which reconstructed a temporally heterogeneous fire regime with fire return intervals that ranged from 30–430 years and a mean of 120 years (median 160 years) in the catchment. These are the first estimates of fire return intervals of mountain forests in eastern Africa. We then explore the potential for further comparative research and incremental research contributions to improve quantitative and qualitative palaeofire research in tropical forest ecosystems. We discuss the potential to use these types of data for characterizing variables of fire regimes prior to ostensibly significant modification by anthropogenic activity as well as during the recent past as human land use pressures increased within Afromontane forests.
... Variability of fire causes vegetation change and is also itself varying in response to changing vegetation composition, structure and biomass (Archibald et al., 2013). Fire maintains some montane ecosystem types in eastern Africa mountains, including montane forest ecotones; severe fires are known to regenerate monospecific stands of either Erica, Podocarpus, Juniperus, or Hagenia under different conditions (Wood, 1965;Lange et al., 1997;Bussmann, 2001;Hemp and Beck, 2001;Young, 2004). Whether or not fire regime variability maintains a forest type, modifies ecotonal transitions, or facilitates or inhibits forest compositions is spatially complex and only a limited number of studies have focused on the role of fire disturbance ecology in Afromontane forests (Wesche et al., 2000;Wooller et al., 2000;Bussmann, 2002a;Hemp, 2005Hemp, , 2006. ...
... Sediment-based records provide retrospective analyses of multidecadal-to-millennial patterns of fire in tropical montane forests (Sánchez Goñi et al., 2017). A changing fire regime is one of the dominant ecological disturbances controlling forested ecotonal transitions (Supplementary Fig. S1; Hemp and Beck, 2001;Gil-Romera et al., 2019;Courtney Mustaphi et al., 2021) and contributes to the spatially heterogeneous vegetation distributions around the mountains of eastern Africa (Bussmann, 2002a;Hemp, 2005Hemp, , 2006 including compositional and structural patchiness (Wood, 1965;Xu et al., 2016). Understanding how montane forests and fire regime variability have responded to the warming and varying hydroclimatic conditions following the Late Glacial Maximum (LGM; Marine Isotope Stage II), a time interval of rapid global climate change, provides quantitative mechanistic insights useful for comparing proxy data and climate and vegetation models Marlon et al., 2016). ...
Full-text available
Fire is an important ecological disturbance in moist tropical forests influencing vegetation composition and structure. Contemporary and historical records of forest fires in mountain forests of Kenya are limited to the past decades and have a strong anthropogenic influence for ignition patterns and fire suppression activities. Palaeoenvironmental geoarchives provide the temporal depth to investigate long-term (multidecadal-to-millennial) changes in fire activity. Here we use a sediment record from the Rumuiku wetland, located in a volcanic crater on the eastern flank of Mount Kenya that was radiocarbon dated and analysed for diatom, pollen and charcoal subfossils to produce a highly resolved time series of local hydroclimatic change, vegetation, and fire; respectively. This study focuses on the time during and following the global Last Glacial Maximum, a time of rapid warming and changing regional hydroclimate with relatively stable atmospheric CO2 and not yet intensive anthropogenic modification of ecosystems. Charcoal and pollen data support associated changes in vegetation-fire centred around 21,500 cal yr BP when Afromontane forests with predominant abundances of Juniperus, Podocarpus and other montane forest trees changed to Hagenia-dominated forests that are relatively more open and adapted to burn more frequently but with less intense fires. These transitions in ecosystem composition, distribution and structure support the important role of fire in driving and maintaining forest composition in the watershed and contributing to the spatial complexity of forests around the mountain. These changes in composition, structure and biomass occurred during a time of rapid Late Pleistocene climate warming, regional hydroclimatic drying, and slowly rising atmospheric CO2 from 27,000 to 16,500 cal yr BP, during and following the conditions of the global Last Glacial Maximum. Temperature, hydroclimate and atmospheric CO2 are well-known drivers of montane vegetation change in the tropics and the role of fire is shown here to be a contributing driver to the spatial heterogeneity of forest patches at long time scales. Vegetation modelling at spatial scales relevant to land management and conservation should include retrospective evidence of the range of drivers of ecological disturbance regimes.
... In the middle montane zone, natural Ocotea usambarensis forests have been selectively logged until 1984 (Agrawala et al. 2003). In the 1 3 upper montane zone, fires disturb the natural Podocarpus latifolius forest, which is then recolonized by fire-tolerant Erica excelsa trees (Hemp and Beck 2001). Likewise, the natural Erica trimera forest in the sub alpine zone is disturbed by fires. ...
One of the few general patterns in ecology is the increase of species richness with area. However, factors driving species-area relationship (SAR) are under debate, and the role of human-induced changes has been overlooked so far. Furthermore, SAR studies in tropical regions, in particular in multilayered rain forests are scarce. On the other side, studies of global change-induced impacts on biodiversity have become increasingly important, particular in the tropics, where these impacts are especially pronounced. Here, we investigated if area modulates the effect of land use, elevation and canopy on plant species richness. For the first time we studied SAR in multilayered tropical forests considering all functional groups. We selected 13 natural and disturbed habitats on Kilimanjaro in Tanzania, distributed over an elevational range of 3700 m. In each habitat type, we set up three to six modified Whittaker plots. We recorded all plant species in 64 plots and 640 subplots and described SAR using the power function. Area consistently modulated effects of elevation on plant species richness, partly effects of land use but not effects of plant canopy. Thus, area needs to be taken into account when studying elevational plant species richness patterns. In contrast to temperate regions open and forest habitats did not differ in SAR, probably due to a distinct vertical vegetation zonation in tropical forests. Therefore, it is important to consider all vegetation layers including epiphytes when studying SAR in highly structured tropical regions.
... Ta część wulkanicznego masywu Kilimandżaro, szczególnie wulkan Shira, ma najlepszą dokumentację cech jakościowych i ilościowych poszczególnych komponentów środowiska przyrodniczego, włącznie z charakterystyką piętrowości roślinnej ( von Decken 1863;Grant 1872;Gillman 1923;Johnston 1885;Williams 1907;Sjöstedt 1925;Spink 1945;Salt 1951Salt , 1954Hemp 2001Hemp , 2005Schrumpf 2004;Schrumpf i in. 2006ab, 2007Zech 2007;Little, Lee 2006;Lovett, 1993). ...
Full-text available
The aim of this paper is to identify and describe the landscape and geochemical zones on the southern slopes of Kilimanjaro. Characterization of individual zones was carried out based on a literature synthesis and author own study. The article contains description of four main zones and eight subzones. The Kilimanjaro’s altitudinal vegetation zones have proper description, but this overview provides a holistic description of the structure and functioning of the landscape.
... In the middle montane zone, natural Ocotea usambarensis forests have been selectively logged until 1984 (Agrawala et al. 2003). In the upper montane zone, fires disturb the natural Podocarpus latifolius forest, which is then recolonized by fire-tolerant Erica excelsa trees (Hemp and Beck 2001). Likewise, the natural Erica trimera forest in the sub alpine zone is disturbed by fires. ...
Full-text available
One of the few general patterns in ecology is the increase of species richness with area. However, factors driving species-area relationship (SAR) are under debate, and the role of human-induced changes has been overlooked so far. Furthermore, SAR studies in tropical regions, in particular in multilayered rain forests are scarce. On the other side, studies of global change-induced impacts on biodiversity have become increasingly important, particular in the tropics, where these impacts are especially pronounced. Here, we investigated if area modulates the effect of land use, elevation and canopy on plant species richness. For the first time we studied SAR in multilayered tropical forests considering all functional groups. We selected 13 natural and disturbed habitats on Kilimanjaro in Tanzania, distributed over an elevational range of 3700 m. In each habitat type, we set up three to six modified Whittaker plots. We recorded all plant species in 64 plots and 640 subplots and described SAR using the power function. Area consistently modulated effects of elevation on plant species richness, partly effects of land use but not effects of plant canopy. Thus, area needs to be taken into account when studying elevational plant species richness patterns. In contrast to temperate regions open and forest habitats did not differ in SAR, probably due to a distinct vertical vegetation zonation in tropical forests. Therefore, it is important to consider all vegetation layers including epiphytes when studying SAR in highly structured tropical regions.
Full-text available
Fire regimes differ across tropical and subtropical biomes depending on multiple parameters whose interactions and levels of importance are poorly understood, particularly at multidecadal and longer time scales. In the catchment of Lake Victoria, savanna, rainforest, and Afromontane vegetation have interspersed over the last 17,000 years, which may have influenced the fire regime and vice versa. However, climate and humans are most often the primary drivers of fire regime changes, and analysing their respective roles is critical for understanding current and future fire regimes. Besides a handful of radiocarbon dates on grassy charcoal, the timescales of published studies of Lake Victoria sediment chronologies rely mostly on dates of bulk sediment, and chronological disagreements persist, mainly due to variation between estimations of the ¹⁴C reservoir effect. Here, we provide independent ¹⁴C chronologies for three Late Glacial and Holocene lacustrine sediment cores from various water depths and compare them with the biostratigraphy to establish a new chronological framework. We present the first continuous sedimentary charcoal records from Lake Victoria; these suggest that fire activity varied substantially during the past 17,000 years. Our new pollen records reveal the long-term vegetation dynamics. The available evidence suggests that before human impact increased during the Iron Age (ca. 2400 yr BP), biomass burning was linked to climate and vegetation reorganizations, such as warming, drying, and the expansion of rainforests and savannas. Our results imply that climate can trigger substantial fire regime changes and that vegetation responses to climate change can co-determine the fire regime. For instance, biomass burning decreased significantly when the rainforest expanded in response to increasing temperatures and moisture availability. Such insights into the long-term linkages between climate, vegetation, and the fire regime may help to refine ecosystem management and conservation strategies in a changing global climate.
The continuous decline and degradation of tropical rainforests is currently mainly driven by land-use change. Over the long term, climate change will further exacerbate the situation for the remaining forests. To develop sustainable conservation strategies, we need to understand the natural forest-recovery potential after disturbance and the impact of climate on forest recovery. So far, little is known about regeneration of woody species in tropical landscapes, especially in tropical Africa. We investigated the regeneration of woody plant species in six natural and seven anthropogenically disturbed habitat types along a 3.6 km elevational gradient from savanna woodland to afro-alpine vegetation at Mt. Kilimanjaro, Tanzania. We recorded and identified all saplings of woody plants in a total of 65 plots, each measuring 100 m². The total number of recorded sapling individuals amounts to 4,738 saplings from 110 species of 50 families. Along the elevational gradient, sapling numbers in natural plots showed a hump-shaped distribution peaking at mid elevation, where rainfall and overall woody biomass were highest, while species number and diversity of saplings were continuously decreasing with increasing elevation. The sapling layer in the savanna zone had three times as many species as the layer of adult trees and shrubs in the same plots, but the species composition of saplings differed strongly from the one of trees and shrubs, indicating strong spatial or temporal heterogeneity. Active agricultural plots at lower elevations also showed natural regeneration of woody species, even though numbers of sapling individuals and species were much lower than in natural low-elevation plots. Forests formerly disturbed by fire and logging showed either comparable or even increased numbers of sapling individuals and species than natural plots at the same elevation, indicating ongoing succession. Finding regeneration of woody species in all disturbed habitat types, even in agriculturally managed habitats, strongly suggests that natural regeneration is a very powerful tool to recover forest diversity after disturbances. Furthermore, our sapling dataset provides an important baseline for future monitoring of the consequences of climate and land-use change on regeneration.
Full-text available
Deadwood is an important structural and functional component of forest ecosystems and biodiversity. As deadwood can make up large portions of the total aboveground biomass, it plays an important role in the terrestrial carbon (C) cycle. Nevertheless, in tropical ecosystems and especially in Africa, quantitative studies on this topic remain scarce. We conducted an aboveground deadwood inventory along two environmental gradients—elevation and land use— at Mt. Kilimanjaro, Tanzania. We used a huge elevation gradient (3690 m) along the southern slope of the mountain to investigate how deadwood is accumulated across different climate and vegetation zones. We also compared habitats that differed from natural forsts in land-use intensity and disturbance history to assess anthropogenic influence on deadwood accumulation. In our inventory we distinguished coarse woody debris (CWD) from fine woody debris (FWD). Furthermore, we calculated the C and nitrogen (N) content of deadwood and how the C/N ratio varied with decomposition stages and elevation. Total amounts of aboveground deadwood ranged from 0.07 ± 0.04 to 73.78 ± 36.26 Mg ha –1 (Mean ± 1 SE). Across the elevation gradient, total deadwood accumulation was highest at mid-elevations and reached a near-zero minimum at very low and very high altitudes. This unimodal pattern was mainly driven by the corresponding amount of live aboveground biomass and the combined effects of decomposer communities and climate. Land-use conversion from natural forests into traditional homegardens and commercial plantations, in addition to frequent burning, significantly reduced deadwood biomass, but not past selective logging after 30 years of recovery time. Furthermore, we found that deadwood C content increased with altitude. Our study shows that environmental gradients, especially temperature and precipitation, as well as different anthropogenic disturbances can have considerable effects on both the quantity and composition of deadwood in tropical forests.
The climate of a steep equator facing slope? The climate in a poleward exposed avalanche track? The climate of an alpine observatory on a windswept ridge? The climate where alpine life occurs, e.g. in the surface of a cushion plant? In winter or in summer? Under overcast or clear sky conditions? In the tropics or in Alaska?
Full-text available
Problems in placing certain species satisfactorily in Philippia or Erica have led to an investigation of characters delimiting the genera mainly in southern African species. The only character used in placing problem species was partial versus total recaulescence of the bract. This separation breaks down completely in a few species in which ericoid and philippioid flowers occur within the same inflorescence. Not all species of Philippia are closely related, some being more closely related to various sections within Erica. It is evident that Philippia is an unnatural polyphyletic group. It is concluded that Philippia should be placed in synonymy under Erica.
Full-text available
:The uppermost forest belt on Mt. Kenya (Kenya, East Africa), ranges from 2900 to 3400 m a.s.l. and is dominated by the evergreen “Kosso” tree, Hagenia abyssinica (Bruce) J.F. Cmel. (Rosaceae). The ecology of this tree, with emphasis on regeneration, was investigated. Twenty-five phytosociological relevés, representing several types of Hagenia forests were produced, and attributed to various associations of the alliances Hagenio abyssinicae — Hypericion revoluti Bussmann 1994 and Hagenio abyssinicae — Juniperion procerae Bussmann 1994 (Bussmann and Beck 1995a). Biometric data of young and adult Hagenia trees were collected.Kosso trees of the individual relevés were either of almost equal size, and presumably age, or could be grouped into only two size categories. This uniformity of the Hagenia populations suggests simultaneous regeneration after major disturbance. Charcoal horizons were found in soil profiles at various sites in the Hagenia zone of Mt. Kenya, indicating former burning of these forests. Therefore, fire is thought to be the disturbing factor, which triggers the regeneration of the Hagenia forests.Germination of Hagenia seeds was investigated under various ecological conditions at Mt. Kenya and in greenhouses. Crucial factors for successful germination were high temperatures and bare soil. Light intensity or potentially allelopathic decomposition compounds, leaching from the abundantly shed leaves, had no significant effect on the germination rates. Fire apparently promotes germination by clearing and heating the prospective seed bed. In areas suffering from a high activity of herbivores, the regeneration capability of Hagenia is greatly decreased or abolished. A fire-requiring regeneration cycle of the Hagenia forests of Mt. Kenya is concluded from the phytosociological and ecological data.
Soil temperatures (annual means in 50-70cm depth) and mean annual rainfall data are the the elements of the thermal-hygric zoning of Mt. Kenya, which is discussed with reference to its climatic causes as well as its ecological consequences. -from English summary
Current concepts of the upper forest line ecology of tropical mountains were revised in the light of recent research results: Clear cut upper forest lines and isolated tree groves inmidth of tropical alpine grasland or dwarf-shrublands are typical for man-made tropical mountain environments: The upper forest line was depressed by man through repeated fire-clearing down to 3 200 m, whereas the natural upper forest line is above 4 000 m a. s. l. Cold-air-ponding, edaphic factors (seasonally waterlogged soils) and wind are ecological factors at the tropical forest line as well as for extratropical forests. Hypothesis refering to the soil temperature as a limiting factor are still not proofed and are increasingly becoming unlikely. Tropical mountains turn out to be widely a man-made environment. Comparative high mountain research has to refer to potentially natural climatically induced altitudinal belts.