Effects of Grazing and Cultivation on Forest Plant Communities in Mount Elgon National Park, Uganda
 
Reproduced with permission from Blackwell Science Ltd. African Journal of Ecology 38 (2): 154-162.

Abstract

Plant communities in the montane forest of Mount Elgon National Park were studied in order to assess the impact of grazing and cultivation on species composition. Present and former land uses, tree, shrub and herb species, soil properties and the percentage cover and height of trees, shrubs and herbs were determined in 40 plots. An indirect ordination of these plots showed that species composition was primarily determined by successional stage and agricultural disturbance. In forest plots (ordinated separately) where the widest range of former and current grazing intensities had occurred, evidence of grazing history, soil phosphorus and vegetation height correlated negatively with the strongest ordination axis. Least grazed forest plots had fewer tree seedlings and saplings than more intensively grazed plots. This may be due to the increase in Mimulopsis alpina (Acanthaceae) in less grazed forest where tree regeneration might other-wise be more advanced. Tree seedlings and saplings were uncommon in the forest, rarely exceeding 30 cm in height and there was no tree understorey. Although grazing is important for preserving species diversity in Mount Elgon National Park through the maintenance of species-rich grasslands, long-term effects on montane forest communities must be considered in future park management.

Introduction

Mount Elgon is a solitary extinct volcano straddling the border between Uganda and Kenya, 100 km north-east of Lake Victoria (Fig. 1). The upper reaches of Mount Elgon received National Park status in 1992. Prior to this, the area had been a Forest Reserve (gazetted in 1951) with objectives in forest protection and timber extraction (Synnott, 1968). The Mount Elgon Conservation and Development Project, implemented by the Ministry for Natural Resources, has been assisting the National Park authorities with forest and community issues since 1987. The current aim of the project is to ‘promote community development and conserve Mount Elgon’s ecosystem for present and future use’ using a ‘community-based resource management approach’ involving the participation and empowerment of local communities (MECDP, 1995). Working with MENP, park regulations have been formulated with reference to the needs of local people and their resource use levels, and enforced in conjunction with extension programmes.

Collaborative management has been piloted in two parishes. The climate on Mount Elgon shows a bimodal pattern of rainfall, with the wettest months occurring from April to October (van Heist, 1994). The forest zone receives the maximum rainfall (approximately 1500 mm (Synnott 1968)). Average annual temperatures decrease with altitude, but do not reach freezing in the study area (Synnott, 1968), which was located in Benet Parish near the Benet Grasslands (Fig. 1). Plots were located in a 5-km² area to the south and west of the patrol hut between 1°17'19 and 1°18'45N and between 34°32'31 and 34°34'00E. Altitude ranged from 2800 to 2920 m. Soils in the study area were ferralsols and nitisols and generally sandy (clay) loams. They were neutral to slightly acidic, and were moderately deep to very deep (van Heist, 1994). The vegetation in the study area comprised a mosaic of montane forest, bushland and grassland communities:

1 Grassland: <10 cm in height; dominated by Adropogon amethystinus, Pennisetum clandestinum and Digitaria scalarum (Graminae); grass species accounting for 15–60% species composition; occasional shrub and herb species over 10 cm occurring with increasing frequency towards forest edge; grazed intensively by cattle and occasionally by donkeys.

2 Bushland: dominated by woody herbs and shrub species forming a closed layer between 1 and 2 m above the ground, with occasional tree saplings; tree canopy absent or below 5% cover; Erica trimera ssp. elgonenis (Ericaceae) abundant in shrub layer and canopy layer (where present) with Artemisia afra (Compositae), Dichrocephala integrifolia (Compositae) and Senecio lyratus (Compositae) particularly abundant in shrub layer; sparse understorey of forbs, pteridophytes and grasses beneath shrub layer; sporadic grazing by cattle in grassy patches; previously cleared for grazing and left to regenerate for 6 or 15–20 years.

3 Forest: tallest trees reaching 15–25 m; >50% canopy cover with an overstorey dominated by Afrocrania volkensii (Cornaceae), sometimes forming an association with Podocarpus latifolius (Podocarpaceae), and no tree understorey; Dicliptera laxata (Acanthaceae) and Impatiens meruensis (Balsaminaceae) particularly abundant in the field layer, with Mimulopsis alpina (Acanthaceae) more abundant in less grazed plots (where its growth form became increasingly shrubby forming a closed layer in patches (¡40% cover)), and with occasional shrubs occurring in this layer; grazed intensively until either 1983 or 1990, when the human populations in the area were evicted, and grazed by cattle with varying intensity since that time.

4 Forest edge: within 100 m of closed canopy forest and either grassland or bushland communities; tallest trees reaching 10–15 m; <50% canopy cover; shrubs, herbs and tree saplings forming a patchy layer (¡40% cover) sharing species from each community.

5 Formerly cultivated: abandoned for 3 or 6 years; dominated by Rumex ruwenzoriensis (Polygonaceae), Plectranthus laxiflorus (Labiatae) and Pilea tetraphylla (Urticaceae), with Urtica massaica (Urticaceae) more common in plots where grazing by cattle was most intense; plots near to Piswa Patrol Hut had been protected from grazing by rangers since abandonment.

Plots were replicated to represent each vegetation type and land use at least three times, with more replicates in more species-rich communities (Table 1). A plot size of 20 x 20 m was chosen for the study in preference to the 1km x 10 m plot size used in previous studies (Katende et al., 1989; van Heist, 1994), so plots could be positioned randomly within the smallest cultivated areas (50 x 50 m). Within each 20 x 20 m plot, trees >5cm girth at breast height (GBH) were identified and counted. The height of the three tallest trees was obtained using a clinometer.

Each plot was divided into sixteen 5 x 5 m subplots. Four of these subplots were randomly selected for the study of herbs, shrubs and soil. Within each 5 x 5m subplot all herbaceous and shrub species, and tree seedlings and saplings (<5 cm GBH) were identified (non-vascular plants and fungi were not identified). Vouchers of unknown species were taken for later identification. Vegetation height in each subplot was measured at five random points. A visual estimation of canopy cover was taken at each subplot through a 28-mm lens, 1 m from ground level.

The vegetation data was analysed using DECORANA (Detrended Correspondence Analysis) multivariate analysis (Hill, 1979). In order to analyse the full diversity of species against the land use data and other possible factors affecting plant community composition, it was necessary to combine the herbaceous and tree data. As the percentage cover data for trees and herbaceous plants were measured differently, the data set was converted to presence/absence data. The significance of relationships between the DECORANA ordination axes with the highest eigenvectors and all the factors measured that could affect the vegetation were examined using regression analysis.

The DECORANA ordination of plots shows floristic differences between forest, bushland, grassland, forest edge and formerly cultivated plant communities (Fig. 2). Sub-groups within the formerly cultivated group are differentiated by age and proximity to other vegetation types. Eigenvectors were strong for axis 1 (0.591) and axis 2 (0.427). The third and fourth axes had much lower eigenvectors that did not relate to any of the factors measured, and were consequently omitted from further analysis.

Negative correlations emerged between the first axis and maximum tree heights (P <0.01), basal area per plot (P <0.01), average basal area per stem (P <0.01) and canopy cover (P <0.01). From this it was inferred that the primary gradient was successional. High stem densities in forest edge and bushland plots also support a succession gradient. Species richness correlated positively with axis 1 (P <0.01). Formerly cultivated plots may follow a deflected succession that may be sub-stantially different from the succession of noncultivated plots, but it was not possible to find a large enough chronosequence of formerly cultivated plots to represent their succession.

The second axis was largely determined by floristic differences between formerly cultivated and forest edge plots. There was a positive correlation between the second axis and levels of soil calcium (P <0.01), potassium (P <0.01), magnesium (P <0.01) and pH (P <0.01). Axis-2 variation between formerly cultivated plots correlated negatively with the time since cultivation ceased (P <0.01).

The indirect re-ordination of forest plots produced an ordination with a strong first axis (0.643) and a weaker second axis (0.34) (axes 3 and 4 were very weak) (Fig. 3). While the first axis correlated with a number of variables, none of the variables studied correlated with the second axis. Floristic variation represented by the first axis was determined by grazing history. Plots that had been grazed intensively until 6 years before the study (and had been moderately grazed from that point until the time of the study) and plots that were currently being grazed intensively were clustered together in the ordination. Plots that had been grazed intensively until 13 years before the study, and had been lightly or moderately grazed since then, were also clustered together. When these grazing histories were ranked according to current and previous grazing intensity, a negative correlation emerged between the first axis and grazing intensity (P <0.01). In addition to this, the concentration of phosphorus in the organic layer of the soil correlated positively with the first axis (P =0.02), whilst the height of the herb and shrub layers correlated negatively with the first axis (P <0.01).

Species were indexed in a direct ordination according to their frequency of occurrence in plots that had been assigned grazing intensity scores 1–5 (lowest–highest) according to the intensity of grazing experienced at each plot. Species with high scores were thus associated most strongly with grazing. In the direct ordination, 88% of species with physical defences against herbivory were highly associated with grazing (species score >3.3). The majority of tree seedlings and saplings had no association with grazing (1.7 < species score > 3.3). Species associated with the most intensively grazed plots included Carduus kikuyorum (Compositae), Clematis simensis (Ranunculaceae) and Senecio lyratus (Compositae) (species score =5.0). Species associated with the least grazed plots included Arisaema mildbraedii (Araceae), Lobelia giberroa (Campanulaceae) and Pleopeltis excavata (Pteridophyte) (species score =1.0).

A seedling on the forest floor: tree seedlings and saplings were uncommon in the forest,
but least common where communities were recovering from former grazing

Tree seedlings and saplings (<5 cm GBH) were most common in bushland, forest edge and nongrazed formerly cultivated plots. Rapanea melanophloeos (Myrsinaceae) and Olea africana (Oleaceae) were present in 66% and 41.5% of bushland and forest edge 5 x 5m subplots, respectively. Dombeya goetzenii (Steculiaceae) was present in 8% of grazed and nongrazed formerly cultivated subplots that had been abandoned for 6 years. In addition to this, however, 33% and 25% of nongrazed subplots contained seedlings or saplings of R. melanophloeos and Afrocrania volkensii, respectively. With the exception of grazed and more recently abandoned cultivated plots, tree seedlings and saplings were least abundant in the forest. Saplings over 30 cm in height were rare in the forest and there was no tree understorey. Twenty-eight per cent of the most grazed forest subplots (groups A and B in Table 1) contained A. volkensii seedlings or saplings, and Prunus africana (Rosaceae) and R. melanophloeos were present in 11% of these subplots. Dombeya goetzenii and O. africana were present in 8% and 3% of the most grazed forest subplots, respectively. In forest plots where least grazing had taken place (groups C and D in Table 1), A. volkensii seedlings or saplings were only present in 11% of subplots, and D. goetzenii and O. africana were present in 3% of these subplots. No tree seedlings or saplings were recorded in the grassland.

In the indirect re-ordination of forest plots, grazing history, species richness, the concentration of soil phosphorus and the height of herb and shrub layers correlated positively with the strongest axis (P <0.01). The hypothesis that grazing history is the main factor determining plant community composition in the forest was supported by the species richness, vegetation height and available phosphate data, because these factors are likely consequences of herbivory. Grazing reduces the height of herb and shrub layers, and phosphates are readily available in animal manures (Maraikar & Amarasira, 1989; Myers et al., 1994).

Afrocrania volkensii was the only tree species found in intensively grazed plots. This may be because A. volkensii was the dominant canopy species and probably produced more propagules than other trees. Although montane forest does not usually have an extensive understorey tree flora, the abundance of tree seedlings and saplings in this forest was much lower than has been described in undisturbed upper montane forest in Cameroon and the Congo (A. Katende, pers. comm.; Lebrun 1935; 1960a,b).

Two of the most common species in the ground layer, Dicliptera laxata and Mimulopsis alpina are in the Acanthaceae family. Dense stands of Acanthaceae have been observed on a number of African mountains (Richards, 1963; Hall, 1995) including Mount Elgon (Tweedie, 1965). Information about the distribution of M. alpina and D. laxata and their association with grazing in the direct ordination suggest that these species inhibit grazers; D. laxata was found in high densities reaching up to 50 cm in height in intensively grazed plots and adjacent to frequently used cattle paths, and M. alpina is herbaceous in its early stages, but in areas that have been less intensively grazed, it becomes dense, woody and tall. In areas where M. alpina and Plectranthus species dominate the shrub layer, cattle will graze Plectranthus to ground level and leave stands of M. alpina intact (M. S. Reed, pers. obs.). This may be due to the rapid increment of lignin in the stems of M. alpina as it matures. The species has no other obvious physical defences against herbivory and only forms dominant stands when mature.

Mimulopsis alpina is monocarpic, with mass flowering events occurring every 7–9 years followed by mass mortality (Fey, 1964; Hall, 1990). On Mount Elgon this phenomenon is not simultaneous throughout stands of the species. Instead, it tends to occur in patches (Tweedie, 1965). Hall (1990) observed that their dominance appears to inhibit the regeneration of tree species, and suggested that the flowering events may be crucial to the regeneration of tree species in Afromontane forests, which depend on dead patches for seedlings to establish. If tree species are unable to establish under intense herbivory, they might be expected to establish when this intensity is reduced significantly. However, when this happens, M. alpina becomes dominant and further prevents tree establishment until die-back. The inhibition of tree regeneration by these species may be compounded by the infrequency of mast years in some tree species. The coincidence of monocarpic die-back in M. alpina with the mast years may determine the climax species that reach maturity. Afrocrania volkensii and O. africana flower at relatively regular intervals in montane forests, and P. africana flowers regularly at lower altitudes (Synnott, 1985), but it is not known if other tree species in the area set seed in mast years. Tree seedlings and saplings (<5 cm GBH) were extremely rare in areas dominated by M. alpina. Dense stands of M. alpina are difficult to penetrate and are most abundant in least grazed areas, which suggests that M. Alpina rather than herbivory is suppressing tree regeneration. Although the results of this study appear to support Hall’s (1990) hypothesis that some monocarpic Acanthaceae may impede tree regeneration, a more detailed study would be required to examine the relationship between M. alpina and tree regeneration more closely.

Most dominant tree species in afromontane forests do not accumulate seeds in the soil (Teketay & Granstrom, 1995). This suggests that their regeneration from seed would not be likely if mature individuals disappeared, and that if destroyed, restoration of these forests may be difficult. It is difficult to predict the long-term impact of former cultivation on the plant communities studied, but grazing clearly impedes their development through tree seedling removal and trampling, and the establishment of M. alpina. Although grazing is important for preserving species diversity through the maintenance of grasslands within Mount Elgon National Park, the long-term effects of herbivory on montane forest communities must be carefully considered in the formulation of future management plans.

 
Acknowledgements

This study was conducted under the tenure of the Department of Plant and Soil Science, Aberdeen University, as part of Project Elgon (http://www.abdn.ac.uk/elgon/),and was supported financially by the British Council, the British Ecological Society, the Royal Geographical Society, the Royal Scottish Geographical Society, the Cross Trust, the British Soil Association, the Gilchrist Educational Trust, Edinburgh Trust no. 2, the Adriane Ashby-Smith Memorial Trust, Aberdeen Medico-Chirurgical Society, Aberdeen Journals, Dr G. Argent and Aberdeen University.

ADEDEJI, F.O. (1984) Nutrient cycles and successional changes following shifting cultivation in moist semi-deciduous forests in Nigeria. Forest Ecology & Management 9, 87–89.

ANDRIESSE, J.P. (1987) Monitoring Project of Nutrient Cycling in Soils used for Shifting Cultivation in Various Climatic Conditions in Asia. Final Report. Royal Tropical Institute, Amsterdam.

BULLOCK, A.A. (1932) New species from Mt. Elgon. Royal Gardens, Kew: Bulletin of Miscellaneous Information 10, 487–509.

BRADY, N.C. (1990) The Nature and Properties of Soils, 10th edn, Macmillan Publishing, Basingstoke.

DAVENPORT,T., HOWARD,P.& DICKENSON, C. (1996) Mount Elgon National Park Biodiversity Report. Forest Department, Kampala, Uganda.

FEY, V. (1964) Cloud Over Kenya. Collins, London.

FRIES, T.C.E. (1923) Beitrage zur Kenntnis der Flora des Kenia, Mt. Aberdare und Mt. Elgon. Notizbl. Des Bot. Gart. U. Mus. Berlin 8, 347–353.

HALL, J.B. (1990) Succession in a Natural Forest at Mazumbai. In: Research for Conservation of Tanzanian Catchment Forests (Eds I. HEDBURG and E. PERSSON). Uppsala University, Uppsala.

HALL, J.B. (1995) Botanical Research Needs and Options Relating to the Mount Cameroon Project. Overseas Development Administration, London.

HEDBERG, O. (1964) Features of Afroalpine Plant Ecology. Almpvist & Wiksells Boktryckeri, Uppsala.

VAN HEIST, M. (1994) Land Unit Map of Mount Elgon National Park. IUCN Technical Report.

HILL, M.O. (1979) DECORANA: a FORTRAN Program for Detrended Correspondence Analysis and Reciprocal Averaging. Section of Ecology and Systematics, Cornell University, New York.

HOWARD, P.C. (1991) Nature Conservation in Uganda’s Tropical Reserves. Forest Department/Ministry of Environment Protection, Entebbe.

KATENDE,T.,IPULET,P.,RODRIQUES,R.&DRANZOA, C. (1989) Birds and Woody Perennials Inventory, Mount Elgon Forest Reserve. Sustainable Development and Forest Conservation in Uganda, IUCN Technical Report no. 1.

LANGDALE-BROWN,I.,OSMASTON,H.A.&WILSON,J.G.(1964)The Vegetation of Uganda. Government Printer, Entebbe.

LEBRUN, J. (1935) Les essences forestieres des regions montagneuses du Congo oriental. Publ. Inst. Agron. Congo Belge Ser. Sci. 1, 1–264.

LEBRUN, J. (1960a) Etudes sur la flore et la vegetation des champs de lave au nord du lac Kivu (Congo Belge). Exploration du Parc National Albert, Mission J. Albert (1937–8). Fasc. 2. Institute Parcs National Congo Belge, Brussels.

LEBRUN, J. (1960b) Sur la richness de la flore de divers territoires africains. Bull. Acad. Royal Sci. d’Outre-Mer (Bruxelles) N.S. 6, 669–690.

LUGARD,E.J.&BULLOCK, A.A. (1933) The Flora of Mt. Elgon. Bull. Misc. Info. Roy. Bot. Gdns. Kew 2, 49–106.

MARAIKAR,S.&AMARASIRA,S.L.(1989)Effectofcattle and poultry dung addition on available P and K of a red-yellow podzolic soil. Trop. Agric. 144, 51–59.

MOUNT ELGON CONSERVATION AND DEVELOPMENT PROJECT (MECDP) (1995) Phase III Objectives. Internal Document. IUCN, Mbale.

MOOCO,L.(1986)Analytical Methods of the Laboratory for Soil, Plant and Water Analysis, Part I: Methods for Soil Analysis. Royal Tropical Institute, Amsterdam.

MYERS, R.J.K., PALM, C.A., CUEVAS, E., GUNATILLEKE, I.U.N. & BROSSARD, M. (1994) The synchronisation of nutrient mineralisation and plant nutrient demand. In: Biological Management of Tropical Soil Fertility (Eds P. L. WOOMER and M. J. SMITT). Wiley-Sayce, New York.

RICHARDS, P.W. (1963) Ecological notes on West African vegetation. III. The upland forests of Cameroon Mountain. J. Ecology 51, 529–554.

RICHARDS, P.W. (1996) The Tropical Rain Forest,2^nd edn. Cambridge University Press, Cambridge.

SYNNOTT, T.J. (1968) Working Plan for Mount Elgon Central Forest Reserve. 1st Revision. Period 1968–78. Uganda Forest Department, Entebbe.

SYNNOTT, T.J. (1985) A Checklist of the Flora of Budongo Forest Reserve, Uganda, with Notes on Ecology and Phenology. Occasional Papers. Commonwealth Forestry Institute, Oxford.

TEKETAY,D.&GRANSTROM, A. (1995) Soil seed banks in dry Afromontane forests of Ethiopia. Veg. Sci. 6, 777–786.

TWEEDIE, F.M. (1965) Periodic flowering of some Acanthaceae on Mt. Elgon. E. Afr. Nat. Hist. Soc. 25, 111.

WHITE, F. (1983). The Vegetation of Africa. UNESCO, Paris.

Project co-ordinators: Mark Reed and Martha Clokie
Field work team: Martin Waliwulya, Martha Clokie, Olivia Maganyi and Mark Reed
Field Assistant: Sabila George Paul
Plant identification: Ms. O. Maganyi and Mr. A. Katende, Makerere University Herbarium
Soil Analysis: A. Bukenda, Makerere University Department of Agriculture
Project advisors: Dr. M. D. Swaine, Dr. D. Burslem and Mr. A. Katende
Report Writer: Mark Reed

For photographs of other plants found in this study, click here.