The structure and composition of the indigenous forest fragments of the Taita Hills of south-east Kenya were analysed. We collected data in all of the large forest fragments using Point Centred Quarter extensive surveys, supplementing these data with intensive surveys of 20 × 20 m plots. Our structural data on basal area per unit area, stem density, canopy cover, shrub density, stratification and extent of herbaceous ground cover clearly show that the largest fragment, Mbololo, is also the least disturbed. The next largest fragment, Ngangao, has suffered from intermediate levels of disturbance, and all other fragments have been very heavily impacted. Tree size class distributions show that nearly all of the large circumference trees have been lost from the smallest fragments, while the Chawia forest is selectively losing its small, easily-cut trees. Finally, we mapped the Importance Values (combined standard measures of abundance, biomass and dispersion) of each forest's tree species. The map indicates the forest biogeography of the region but also shows the extent to which all of the fragments except Mbololo and Ngangao are dominated by a very few secondary successional sub-canopy species. Only Mbololo and Ngangao may be viable forests in the long term.
VEGETATION STRUCTURE AND COMPOSITION OF THE TAITA HILLS FORESTSC. Wilder⟨sup⟩1⟨/sup⟩University of Virginia School of Medicine, Health Sciences CenterBox 458, Charlottesville, VA 22908, U.S.A.cmw6a@virginia.eduT. BrooksCenter for Applied Biodiversity Science, Conservation International2501 M Street NW, Suite 200, Washington, DC 20037, U.S.A.t.brooks@conservation.orgL. LensLaboratory of Animal Ecology, Department of BiologyUniversity of Antwerp, UIA, B-2610, Wilrijk, Belgiumllens@uia.ua.ac.beABSTRACTThe structure and composition of the indigenous forest fragments of the Taita Hills of south-east Kenya were analysed. We collected data in all of the large forest fragments using Point Centred Quarter extensive surveys, supplementing these data with intensive surveys of 20 × 20 m plots. Our structural data on basal area per unit area, stem density, canopy cover, shrub density, stratification and extent of herbaceous ground cover clearly show that the largest fragment, Mbololo, is also the least disturbed. The next largest fragment, Ngangao, has suffered from intermediate levels of disturbance, and all other fragments have been very heavily impacted. Tree size class distributions show that nearly all of the large circumference trees have been lost from the smallest fragments, while the Chawia forest is selectively losing its small, easily-cut trees. Finally, we mapped the Importance Values (combined standard measures of abundance, biomass and dispersion) of each forest's tree species. The map indicates the forest biogeography of the region but also shows the extent to which all of the fragments except Mbololo and Ngangao are dominated by a very few secondary successional sub-canopy species. Only Mbololo and Ngangao may be viable forests in the long term.INTRODUCTIONThe remaining forests of the Taita Hills, south-east Kenya, are poorly known botanically. A number of historical collections exist from the area, but there have been only three recent species inventories (Beentje, 1987; Mwangangi & Mwaura, 1992-3; Imboma, 1997). These show that many of the forests' tree species are shared with both the Kenyan Highlands and the Eastern Arc Mountains, although they are distinct from either, having no less than 13 endemic plant taxa (Beentje, 1987). However, only one study, limited to the single site of Ngangao, has assessed forest structure (Tetlow, 1987). In this paper, we present data on the vegetation of the eight largest surviving forest fragments of the Taita Hills: Sagalla (03°30'S, 38°35'E), 1,500 m, 4 ha; Ronge (03°21'S, 38°25'E), 1,200 m, 1 ha; Mbololo (03°20'S, 38°26'E), 1,800-2,200 m, 200 ha; Ngangao (03°22'S, 38°20'E), 1,700-2,100 m, 92 ha; Chawia (03°28'S, 38°28'E), 1,500 m, 50 ha; Fururu (03°24'S, 38°20'E), 1,400 m, 5 ha; Vuria (03°24'S, 38°17'E), 2,000 m, 1 ha; and Mwachora (03°25'S, 38°22'E), 1,400 m, 2 ha.⟨sup⟩1⟨/sup⟩ All the work was completed while the authors were at the following address: Ornithology Department, National Museums of Kenya, P.O. Box 40658, Nairobi, Kenya.METHODSIn the five largest forests we established transects with 40 equally spaced points (three forest fragments, Fururu, Vuria and Mwachora, were too small to hold 40 independent points). In most forests these transects had to be oriented to fit the shape and size of the patch; therefore point spacing varied from 20 m between points in the smallest forests to 100 m in the largest. We surveyed each of the 40 points using the Point Centred Quarter method (Cottam & Curtis, 1956). After measuring the distance from the point to the closest tree ⟩45 cm circumference, i.e. ⟩ 15 cm diameter at breast height in each of the four quadrants NE, SE, SW and NW, we identified the trees and measured their circumferences at 120 cm height.Following Kent & Coker (1992), we used these data to calculate the Shannon-Wiener diversity index, H':wheres = the number of speciesp⟨sub⟩i⟨/sub⟩ = the total basal area of the i⟨sup⟩th⟨/sup⟩ species as a proportion of the total basal area of all treesWe also calculated the Shannon-Wiener equitability index, J, using:J = H'/ln sThe Importance Values (IVs) for each species (as percentages) were determined by the following formula (Curtis & McIntosh, 1950, 1951):IV = (Relative density + Relative dominance + Relative frequency)/3where:Relative density = Number of individuals/Number of individuals of all speciesRelative dominance = Total basal area/Total basal area of all speciesRelative frequency = Number of quadrants occurring/Total number of quadrantsThese IVs give a combined standard measure of abundance, biomass and dispersion for each species (Kent & Coker, 1992).At each point in our Point Centred Quarter surveys, we estimated canopy cover and took half-sight measurements of shrub layer density at 1 m and 2 m height (Wight, 1938). For each measurement we counted the number of squares more than half visible on a board that was divided into 25 squares of 8 cm × 8 cm, held at a 5 m distance from the observer. Finally, we counted the number of saw-pits near each point.In addition, we chose three points in a stratified random manner at which to establish large (20 m × 20 m) plots. In these, we identified and counted every shrub (1-5 m tall); and identified every tree (⟩5 m tall), measured its circumference at 120 cm and classified its status (e.g., as dead, artificially cut, etc). We noted the presence of buttresses for trees ⟩ 15 m tall, the heights and dominant species of each vegetation stratum, and the plot's disturbance and slope. Finally, we established five 1 m × 1 m subplots in the centre and in each corner of the plot, within which we estimated the % cover of bare ground, leaf litter and herbaceous growth, and measured the leaf litter depth. For the three smallest forest fragments, we surveyed a single large plot in the centre of the patch and measured half-sights in the five subplots of this plot. All identifications were made with reference to Beentje (1994) and to specimens in the East African Herbarium, Nairobi, Kenya.RESULTSForest structure: Point Centred Quarter surveysTable 1 presents the summary results of our Point Centred Quarter surveys and figure 1 illustrates the frequency distributions of size classes of trees as revealed by these surveys. Mbololo is clearly the least disturbed forest, with the highest mean basal area per m⟨sup⟩2⟨/sup⟩ (an indicator of live standing biomass), stem density and canopy cover, and the most open understorey (as indicated by the high half-sight values). It also retains the largest trees of the Taita Hills, although it holds less intermediate-sized trees than do Ngangao and Chawia, and its mean tree size is lower as a consequence. Ngangao has suffered an intermediate level of disturbance. There are numerous small trees at this site, which have presumably regenerated since the termination of pit-sawing here in the 1970s (Brooks et al., 1998), although Ngangao also holds some very large trees. Likewise, some large trees survive in Chawia, but heavy selective logging of easily-transported small trees at this site has prevented regeneration, as indicated by its low number of small trees, high mean tree size, and low canopy cover. Sagalla and Ronge have both been very heavily disturbed, having the lowest basal area per m⟨sup⟩2⟨/sup⟩, mean tree sizes and stem densities, the fewest large trees, and the densest under-storeys.Forest structure: large plotsTable 2 presents the summary results from our surveys of large (20 m × 20 m) plots. These data are derived from small samples (1,200 m⟨sup⟩2⟨/sup⟩ for large forests and only 400 m⟨sup⟩2⟨/sup⟩ for small forests) and should therefore be interpreted with caution. In particular, the basal area per m⟨sup⟩2⟨/sup⟩ of 86 cm⟨sup⟩2⟨/sup⟩/m⟨sup⟩2⟨/sup⟩ for Ngangao is greatly influenced by a single huge tree, and that for Mwachora is vastly inflated by a large Phoenix reclinata Jacq. clone. However, it can be seen that the two lowest altitude and therefore most level sites (Ronge and Chawia) have the deepest leaf litter, despite the degree to which they have been disturbed. The high stratification and number of buttressed trees attest to the quality of the forest in Mbololo and Ngangao. There are also non-significant tendencies for the more disturbed sites to have a more extensive herb layer and proportionately less leaf litter, higher proportions of dead and cut trees, and more trails.Table 1. Structural data collected from Point Centred Quarter surveys (n = 40 for each forest fragment). Our samples of mean basal area/tree are highly skewed (figure 1) and so standard deviations are uninformative measures of dispersion for these samples. For these, we instead give values of skewness (in parenthesis). As a result, values of biomass (mean basal area/m⟨sup⟩2⟨/sup⟩) also lack standard deviations, but we give standard deviations for all other mean values.ParameterSagallaRongeMbololoNgangaoChawiaMean basal area/m⟨sup⟩2⟨/sup⟩ (cm⟨sup⟩2⟨/sup⟩/m⟨sup⟩2⟨/sup⟩)3325775861Mean basal area/tree (cm⟨sup⟩2⟨/sup⟩)858(2.8)827(4.6)1,358(4.5)1,536(2.9)2,041(1.9)Stem density/ha386±116301±116578±148380±134297±95Mean canopy cover (%)57±3260±3172±3460±3251±34Mean half-sight: 1m13±810±818±616±912±9Mean half-sight: 2m12±911±819±612±911±9Total pit-saws102122Figure 1. Frequency distribution of size classes of trees surveyed by Point Centred Quarter survey.Forest compositionTable 3 gives three measures of forest composition derived from our Point Centred Quarter surveys. Mbololo and Ngangao clearly have higher tree species richness (i.e. variety of species) and diversity indices (i.e. variety of species biomass across total biomass) than do the other forests. They also have higher equitability indices (i.e. evenness of biomass distribution across species) than does Chawia, indicating their healthy distribution of diverse tree species. Sagalla and Ronge have equitability indices as high as Mbololo and Ngangao; however, this is largely a product of their low species richness.Table 2. Structural data collected from 20 m × 20 m plots. Our samples of mean litter depth are highly skewed and so standard deviations are uninformative measures of dispersion for these samples. For these, we instead give values of skewness (in parenthesis). We give standard deviations for all other mean values.ParameterSagallaRongeMbololoNgangaoChawiaFururuVuriaMwachoraRange of slopes (°)0-385-1134-4824-364-14305610Range of strata2-333-42-42-3323Mean litter depth (mm)15(1.7)30(3.7)14(1.3)13(2.7)25(1.8)12(-1.7)10(0.6)6(0.3)Mean litter cover (%)57±3081±3075±1958±2947±2560±2978±317±24Mean herb cover (%)30±2514±2111±1234±2853±2537±3211±981±26Dead trees (%)1412912138220Artificially cut trees (%)201420130Buttressed trees/1,200 m⟨sup⟩2⟨/sup⟩041062033Number of trails/1,200 m⟨sup⟩2⟨/sup⟩01135336Basal area/m⟨sup⟩2⟨/sup⟩ (cm⟨sup⟩2⟨/sup⟩/m⟨sup⟩2⟨/sup⟩)25436386534541281Figure 2 presents IVs of ⟩ 9 % for tree species in the five large forests as determined by our Point Centred Quarter survey. The isolation of Sagalla and the low altitude of Ronge set the composition of these two forests apart from the rest of the Taita Hills (figure 2). Thus important trees in Sagalla (e.g. Newtonia buchananii (Bak.) Gilb. & Bout.) and Ronge (e.g. Anthocleista grandiflora Gilg.) are scarce elsewhere in the Taita Hills, which are instead characterised by species like Albizia gummifera (J.F. Gmel.) C.A. Sm., Tabernaemontana stapfiana Britten and Strombosia scheffleri Engl. This pattern is reflected in the avian biogeography of the forests, with the drier, lower altitude forests of Sagalla and Ronge holding very different bird communities to the moist high forests (Brooks et al., this volume).Table 3. Measures of forest composition data collected from Point Centred Quarter surveys (n = 40 for each forest fragment).ParameterSagallaRongeMbololoNgangaoChawiaNumber of species1522363217Shannon-Wiener diversity, H'2.182.242.752.441.68Shannon-Wiener equitability, J0.810.730.770.700.59Nevertheless, the heavy disturbance the forests have suffered has clearly impacted their composition as well as their structure, overlying the background patterns of biogeography. The degree to which Chawia, Ronge and Sagalla have been disturbed is shown by the importance in these forests of just a few tree species, especially early successional subcanopy species of little economic importance like T. stapfiana and P. reclinata. In contrast, Ngangao and especially Mbololo have a more even distribution of species IVs and much greater species diversity.Figure 2. Importance Values (IVs) ⟩ 9 % for tree species in the five large forests as determined by the Point Centred Quarter survey. Areas are proportional to species numbers, with stripes for subcanopy and shading for canopy trees. The thin lines delimit the area above 1,200 m, the thick lines show main roads, and the solid areas show indigenous forests.DISCUSSIONThe structure and composition of the Taita Hills forests indicate the high levels of disturbance suffered here. In the future, only Mbololo and Ngangao, with their relatively natural forest structure and composition, are likely to persist. The dominance of the other forests by a few species and the former heavy cutting of young (in Chawia) and old (in Sagalla and Ronge) trees may well compromise their viability in the long term.ACKNOWLEDGEMENTSWe thank J. Barnes, R. Barnes, A. Cooke, D. Gitau, J. Kageche and S. Karimi for help with fieldwork, G. Mwachala for help identifying specimens in the East African Herbarium, the staff of the Taita-Taveta District Forest Department and the National Museums of Kenya Ornithology Department for support, and N. Burgess and J. Fjeldså for helpful reviews. 1996 fieldwork was funded by National Geographic Society Research Award No. 5542-95 to S.L. Pimm of the University of Tennessee.REFERENCESBeentje, H.J. (1987). An ecological and floristic study of the forests of the Taita Hills, Kenya. Utaflti 1(2): 23-66.Beentje, H.J. (1994). Kenya Trees, Shrubs and Lianas. National Museums of Kenya, Nairobi.Brooks, T., L. Lens, J. Barnes, R. Barnes, J. Kageche Kihuria & C. Wilder (1998). The conservation status of the forest birds of the Taita Hills, Kenya. Bird Conservation International 8: 119-139.Brooks, T., L. Lens, M. De Meyer, E. Waiyaki & C. Wilder (this volume). Avian biogeography of the Taita Hills, Kenya. Journal of East African Natural History 87: 189-194.Cottam, G. & J.T. Curtis (1956). The use of distance measures in phytosociological sampling. Ecology 37: 451-460.Curtis J.T. & R.P. Mcintosh (1950). The interrelations of certain analytic and synthetic phytosociological characters. Ecology 31: 434-455.Curtis, J.T. & R.P. Mcintosh (1951). An upland forest continuum in the prairie-forest border region of Wisconsin. Ecology 32: 476-496.Imboma, T. (1997). Bird research in the Taita Hills-a view from the ground. Kenya Birds 6: 6-8.Kent, M. & P. Coker (1992). Vegetation Description and Analysis. John Wiley & Sons, Chichester.Mwangangi, M. & P.K. Mwaura (1992-3). Taita Hills Forest Survey Report. Kenya Indigenous Forest Conservation Programme, Nairobi.Tetlow, S.L. (1987). Cambridge Conservation Study 1985 Taita Hills, Kenya. ICBP Study Report No. 18. International Council for Bird Preservation, Cambridge.Wight, H.M. (1938). Field and Laboratory Techniques in Wildlife Management. University of Michigan Press, Ann Arbor.