Continuum or zonation? Altitudinal gradients in the forest vegetation of Mt. Kilimanjaro Andreas Hemp Lehrstuhl fu��r Pflanzenphysiologie, Universita��t Bayreuth, Universita��tsstr. 30, 95440, Bayreuth, Germany (e-mail: andreas.hemp@uni-bayreuth.de phone: +49-0-921-552630 fax: +49-0-921-552642) Received 8 May 2005 accepted in revised form 23 August 2005 Key words: Biodiversity, Cloud forest, East Africa, Epiphytes, Tropical montane forest, Vegetation zonation Abstract Based on the analysis of 600 vegetation plots using the method of Braun-Blanquet (1964) the compo- sition of the whole vascular forest plant flora with about 1220 species was studied in the forests of Mt. Kilimanjaro. The altitudinal distribution of all strata (trees, shrubs, epiphytes, lianas and herbs) along a transect of 2400 m is discussed with respect to altitudinal zonation and ecological factors. With uni- dimensionally constraint clustering significant discontinuities were revealed that occurred simultaneously in the different strata. Thus even in structurally highly complex, multilayered tropical montane forests distinct community units exist that can be surveyed and classified by the Braun-Blanquet approach. This observed zonation was significantly correlated with altitude, temperature and soil acidity (pH) rainfall was of importance in particular for the zonation of epiphytes. Other key factors were humidity (influenced by stable cloud condensation belts) and minimum temperature (in particular the occurrence of frost at 2700 m altitude upslope). The contrary results of other transect studies in East Africa in respect to continuity of change in floristic composition appear to be caused by different sampling methods and intensities or mixing of data from areas with different climate conditions, whereas species richness did not influence the clarity of floristic discontinuities on Kilimanjaro and other parts of East Africa. Introduction Tropical forests, because of their species richness, belong to the structurally most complex ecosys- tems (e.g. Vazquez and Givnish 1998 Givnish 1999). This diversity displays striking patterns of changes along gradients of ecological factors (Givnish 1999). Forest species inhabit distinct strata or synusiae sensu Barkman (1973). In con- trast to temperate forests epiphytes and lianas form additional strata of varying importance in tropical forests. Although the indicator value of lianas, herbs, epiphytes and shrubs is at least as high as that of canopy trees (Hall and Swaine 1976 Gentry and Dodson 1987), most studies describing the altitudinal changes in the forest vegetation of East African mountains deal only with trees (Hamilton 1975 Clutton-Brock and Gillett 1979 Friis and Lawesson 1993 Lovett 1996, 1998). In contrast, the present work extends such studies for the first time to all strata of the whole vascular forest flora of an African mountain Plant Ecology (2006) 184:27���42 �� Springer 2005 DOI 10.1007/s11258-005-9049-4

to obtain a comprehensive view of the altitudinal changes of such highly diverse habitats. Changes of the floristic composition on tropical mountains are well documented at least for trees and selected taxocenes. The results show that altitude (or factors relating to altitude) is by far the biggest determinant (e.g. Hamilton et al. 1989 Woldu et al. 1989 Gentry 1995 Vazques and Givnish 1998). However, it is still a matter of de- bate whether altitudinal changes are continuous or discontinuous. This question relates to the con- trasting community unit theory vs (individualistic) continuum concepts (Moravec 1989). Bray and Curtis (1957), Curtis (1959), Whittaker (1962, 1967) and McIntosh (1967) described vegetation on temperate mountains as a continuum with staggered and random modes of Gaussian species distributions. For tropical mountains it was sug- gested that the forest tree flora forms a continuum as well (Hamilton 1975 Hamilton et al. 1989 Lieberman et al. 1996 Lovett 1996, 1998 Vazques and Givnish 1998). On the other hand a number of studies empha- size that there are zones or belts of vegetation delimited by relatively narrow boundaries (���critical altitudes���) in which there is elevation-related dis- continuous variation in floristic composition or structure. Hedberg (1951) identified three main vegetation belts each with several subzones: the montane forest belt, the ericaceous belt and the alpine belt. Van Steenis (1984) found altitudes where the floristic composition shows rather abrupt demarcations in the mountain flora of Malaysia, Boughey (1955a) on Mt. Cameroon, Beals (1969), Woldu et al. (1989), Friis (1992) and Friis and Lawesson (1993) in Ethiopia, Kitayama (1992) on Mt. Kinabalu and Kitayama and Mu- eller-Dombois (1992) on Hawaii. These studies deal with phanerogams but cryptogams were also successfully used as tools for identification of altitudinal zones (Jacobsen and Jacobsen 1989 Kurschner 1990 Frahm 1994 Pocs 1994 Kessler 2000 Hemp 2002a). Mt. Kilimanjaro, one of the highest solitary mountains of the world (700 ���5895 m a.s.l.), and overtopping its surroundings by more than 5000 m, is an outstanding object for the investi- gation of altitudinal effects on species composition and diversity. In this study the following questions were addressed: (1) Does the floristic composition of the forest change distinctly with altitude (2) how are changes in vegetation related to the alti- tudinal gradient of climatic factors along the mountain slope (3) differ the strata with altitude or do they react simultaneously? Study area Mt. Kilimanjaro is located 300 km south of the equator in Tanzania on the border with Kenya be- tween 2��45�� and 3��25�� South and 37��00�� and 37��43�� East. Thirty-four transects were established on the whole mountain (Figure 1). Natural forests cover an area of about 1000 km2 on Mt. Kilimanjaro. A detailed description of the forest types together with synoptic tables has been presented by Hemp (2001a, b, inpress c), and a vegetation map was published by Lambrechts et al. (2002). Topography Mt. Kilimanjaro is a more or less eroded relic of an ancient volcano with three peaks (Shira, Mawenzi and Kibo) rising from the savanna plains at 700 m elevation to a snow- and ice-clad summit of 5895 m altitude. Its diameter from northwest to southeast is about 90 km. Climate Mt. Kilimanjaro is characterized by a typical equatorial day-time climate. The distribution of precipitation over the year follows the intertropical convergence zone and is modified by the respective elevations. Due to its equatorial location, two distinct rainy seasons occur in the study area: the long rains from March to May, and the short, but heavy rains around November. The driest period is from July to October, while April and May are the wettest months. According to the climate classifi- cation system of Koppen and Troll/Pfaffen (in Muller 1983) Mt. Kilimanjaro belongs to the zone of a seasonal dry tropical climate. However, rain- fall and temperature vary with altitude and expo- sure to the dominant wind from the Indian Ocean. The northern slopes, on the lee side of the moun- tain, receive much less annual rainfall than the southern slopes. 28

According to Lauer (1976) precipitation on tropical high mountains exhibits two patterns: on extremely wet mountains such as Mt. Cameroon precipitation decreases from the foothills with altitude. Mountains with dry foothills have their maximum of rainfall somewhere below 2500 m depending on the altitude where the wet monsoon interacts with the dry trade winds at higher ele- vations. This condensation level depends on the dryness of the foothill zone. Because of lack of data, rainfall distribution on Mt. Kilimanjaro was until recent time unclear (Hemp 2001a). Our own measurements, which include the first rainfall data from the high pre- cipitation area inside the southern forest belt, show that annual rainfall on the central southern slope increases to about 1900 mm at 1400 m and to about 2700 mm (partly over 4000 mm on ex- posed ridges) at 2200 m in the lower part of the forest belt, thus markedly exceeding precipitation on other East African high mountains. At higher elevation, precipitation declines, reaching 80% of the maximum at 2400 m, 70% at 2700 m, 50% near the upper forest border at 3000 m and only 20% at 4000 m (Figure 2). These details contrast with more general and earlier reports, e.g. by Hedberg (1951), Hastenrath (1973), Lind and Morrison (1974) and Richter (1996) that Mt. Kilimanjaro is in general drier than other East African high mountains. Mean annual temperature decreases linearly upslope with a lapse rate of 0.56 ��C per 100 m starting with 23.4 ��C at the foothills in Moshi (813 m Walter et al. 1975), and decreasing to )7.1 ��C at the top of Kibo (Thompson et al. 2002) (Figure 2). A temperature lapse rate of 0.55 ��C/100 m is typical of the humid tropics (Lauer 1976). From 2700 m upwards frost occurs during clear nights in July and August. Frost at elevations as low as 1500 m have been recorded in the nearby situated Usambara (Moreau 1935) and the Uluguru Mts. (Pocs 1976), which are relatively close to Mt. Kilimanjaro. Due to the ���Massenerhebungseffekt��� of the much higher Figure 1. Study area with location of transects (dotted lines) and division into northern (N), central southern (S) and south western (SW) forest blocks. 29