Distinguishing forest tree communities in Kibale National Park, western Uganda using ordination and classification methods Patrick Mucunguzi1,*, John Kasenene1, Jeremy Midgley2, Paul Ssegawa1 and John R. S. Tabuti1 1Department of Botany, Makerere University, PO Box 7062, Kampala, Uganda and 2Department of Botany, University of Cape Town, Private Bag X3, Rondesbosch 7701, South Africa Abstract A study of spatial variation in tree community structure and species composition in the Kibale National Park, western Uganda was conducted. Tree communities were compared at five sites namely K-14, K-15 and K-30 at Kanyawara in the north, Ngogo in the central part of the forest and Mainaro in the southern part. All trees ‡10-cm diameter at breast height were censused along belt transects covering a total of 15 ha in all sites. Cluster analysis and principal component analysis were used to identify forest tree communities and species associations. Using cluster analysis, two species assem- blages emerged: the Mainaro, Ngogo and K-15 cluster and the K-30 and K-14 cluster. Principal component analysis revealed the descriptive species for the northern and southern sites. Key words: ordination, species associations, tree commu- nity Introduction Trees are responsible for much of the biomass in tropical forest ecosystems. Their distribution and abundance strongly influences the abundance of other organisms in these ecosystems (Davies, 1994; Hart, 2001; Brugiere et al., 2002). Many studies of tree community structure and species composition conducted throughout the tropics indicate that tropical forest tree community structure and species composition varies widely between forests in different continents (Phillips et al., 1994), on the same continent (Ter Steege et al., 2000) and different sites within the same forest (Proctor et al., 1983). Most studies of intra forest variation in tree structure and composition have compared sites that differ in forest type (e.g. lowland versus montane, undisturbed versus disturbed). It is therefore not entirely surprising that the identities and densities of tree species generally differ markedly between sites in these studies (Parthasarathy, 1999; Swamy et al., 2000). Comparisons between sites ‡5 km apart of similar forest types within the same forest have been less common (Butynski, 1990; Chapman et al., 1997). Nevertheless, studies conducted in the Kibale forest, Uganda suggest that even within the same forest type, considerable spatial heterogeneity in tree community structure and composition can exist (Butynski, 1990; Chapman et al., 1997). Traditional approaches to investigate plant communities often use ordination and classification methods (Okland, 1990; Kent & Coker, 1992). These include direct and indirect gradient analysis techniques (Jongman, Ter Braak & Van Tongeren, 1995). Indirect gradient techniques of data analysis such as principal Components Analysis (PCA) and cluster analysis (CA) provide means of gener- ating and evaluating a classification of plant communities and enable the detection of patterns (Okland, 1990; Jongman et al., 1995). Such methods work best when environmental data are collected to determine factors affecting species composition. Although there have been studies carried out in Kibale National Park by Butynski (1990) and Chapman et al. (1997) to identify the various vegetation types, the methods and approaches used were different. The aim of *Correspondence: E-mail: pmucunguzi@botany.mak.ac.ug All authors declare no conflicts of interest � 2007 African Journal of Ecology, Afr. J. Ecol., 45, (Suppl. 3) 99–108 99 this study was to determine the tree communities in different predetermined forest types, to establish the degree of similarity between sites, and the different tree species associations. The most widespread and restricted tree species were also identified. We predicted that sites which experienced similar management regimes in the past will tend to have similar species assemblages and would be reflected in the species asso- ciations. Materials and methods Study area Kibale National Park includes a medium altitude (1110– 1590 m above sea level.) transitional, moist forest interposed between dry tropical and wet tropical rain forest in the Albertine region (0o13¢–41¢N and 30o19¢– 30o32¢E) in western Uganda (Howard & Davenport, 1996; Struhsaker, 1997). The protected area, which covers 766 km2, was declared a National Park in 1993. Kibale forest covers numerous hills, valleys, swamps and streams. The rainfall ranges from 1490 mm at Ngogo to 1622 mm at Kanyawara and distributed in two wetter seasons that include March–May and Septem- ber–November (Struhsaker, 1997). Mean annual tem- perature of the reserve is 20.5�C and varies little during the year. This study was carried out in five study sites namely K-15, K-14 and K-30 at Kanyawara, Ngogo and Mainaro. These sites were at different locations of the forest (north, central and south) and had differences in dominant species and levels of anthropogenic disturbances (Table 1; Ghiglieri, 1984; Howard, 1991; Kasenene, 1987; Osmaston, 1959; Skorupa & Kasenene, 1984; Skorupa, 1988; Van Orsdol, 1983). Floristic composition assessment The stratified random sampling technique based on forest types and intensity of logging, with completely random samples in each of the strata was used to locate the sam- pling plots (n = 150). Within each forest type, a sample of 30 of 0.1 ha (10 · 100 m) belt transects were randomly established. The belt transects were oriented to run across similar topographical gradient of valley, lower slope, and upper slope to cover identical vegetation in all study areas. All trees ‡10-cm diameter at breast height (dbh) were enumerated and identified to species level. Species nomenclature is a according to Polhill et al. (1952). Data analysis Data for this study was analysed using CA and principal component analysis. CA is a technique that sorts objects (such as sampling units) into groups or clusters based upon their overall resemblance to one another (Ludwig & Reynolds, 1988). PCA is an ordination technique (Pielou, 1984) which partitions a resemblance matrix (variance– covariance or correlation) into a set of orthogonal (perpendicular) axes or PCA ‘components’ (Ludwig & Reynolds, 1988). The first few PCA components will explain the largest percentage of variation in the data set (Gauch, 1982) and ordinations of sampling units on these axes provide information about the ecological relationship between them. Differences in species richness between forest types were analysed using the chi-squared technique at a level of significance of 95%. Tree community structure analysis The presence of 113 tree species was recorded in binary (presence or absence) format for each of the five sites studied. A matrix of species by study sites was arranged to determine similarity of sites in terms of species assemblage. A Jaccard’s Index (JI) was calculated (Ludwig & Reynolds, 1988). JI is expressed as: JI ¼ a=ðaþ bþ cÞ; where a is the number of species that sites A and B have in common, b as the number of species present in site A but absent from site B and c as the number of species present in site B but absent from site A. The JI ranges from near 0 (for sites highly dissimilar with respect to species) to near 1 (for sites that are very Table 1 Characteristics of the study sites Study sites Location Dominant species Level of disturbance K-15 (Kanyawara) North Parinari excelsa Heavily logged K-14 (Kanyawara) North Parinari excelsa Moderately logged K-30 (Kanyawara) North Parinari excelsa Unlogged Ngogo Central Pterygota mildbraedii Unlogged Mainaro South Cynometra alexandri Heavily logged and encroached 100 Patrick Mucunguzi et al � 2007 African Journal of Ecology, Afr. J. Ecol., 45, (Suppl. 3) 99–108 similar). An agglomerative clustering technique (weighted centroid) provided in the Multivariate Statistical Package (MVSP) of Kovach (1999) was used to produce a dendro- gram containing all five sites. A minimum JI of 0.0 was used for defining clusters. Several algorithms were ex- plored and clusters determined from one of the CA meth- ods based on the underlying ‘ecological knowledge of the data’ (Ludwig & Reynolds, 1988). Other measures of similarity and measures of distance between sites were explored along with several different methods of clustering. All techniques provided similar results, with the JI and weighted centroid clustering being most meaningful ecologically. Species associations The two groups of sites produced by CA were considered unique for the analysis of species associations. From the absence–presence data from sampled sites, two new arrays were created based on site clusters A and B. Each array contained binary information for species present in more than 4% of quadrats sampled in sites of its corresponding cluster. The species which occurred in <4% of the quadrats sampled were considered rare and were removed from the data set to avoid introducing unnecessary noise (Mentis, 1983). This is justified for two reasons. First, the occur- rences of rare species are usually more a matter of chance than an indication of ecological condition. Secondly, most multivariate techniques are affected very little by rare species carrying such a small percentage of the overall information of variance (Gauch, 1982). The two matrices produced were of sizes 90 · 69 (cluster A) and 60 · 103 (cluster B) where the rows represented the sampling quadrats and the columns the species. To determine any degree of similarity between species, the two arrays were used separately to calculate simple matching coefficients (SM) of Kovach (1999), for each species pair (A and B) using the following expression: SM ¼ ðaþ dÞ=ðaþ bþ cþ dÞ; whereby a is the number of quadrats containing species A and B; b the number of quadrats where species A was present but species B was absent; c the number of quadrats where species B was present but species A was absent; d the number of quadrats where neither species A or B were present. The SM ranges from 0 to 1. Species which rarely occur together would have a coefficient of 0, while those which normally occur together would have a coefficient of 1. Two separate weighted centroid cluster analyses were conducted to produce dendrograms using MVSP (Kovach, 1999). By this analysis, tree species were grouped according to their frequency of occurrence in the study sites. A minimum SM of 0.0 was used to define species assemblages. As with the CA of the sites, other measures of similarity among species were attempted along with sev- eral different methods of clustering. All techniques pro- vided similar results, with the SM and weighted centroid clustering making the most sense ecologically. Results Sites similarity The study recorded 67, 66, 59, 61 and 54 tree species for Kanyawara K-30, K-14 and K-15, Ngogo and Mainaro respectively. There was no significant difference in the species richness in the five forest types despite the tendency of disturbed forests to have fewer species (v2, P > 0.05). Cluster analysis produced two distinct groups, clusters A and B. Cluster A included Mainoro, Ngogo and Kanyawara K-15 (Fig. 1). Cluster B included Kanyawara K-14 and K-30. The first two principal components (PC 1 and PC 2) from the PCA accounted for 67% of variance in binary species data (Table 2). The decomposition of the principal components provided insight into which species were responsible for defining the groups (Table 3). Species with relatively high loadings >0.062 on PC 1 are shown in Table 3 in bold. These species were descriptive of sites A MAIN NGO K 15 K 14 K 30 B Jaccard’s index 0.4 0.5 0.6 0.7 0.8 0.9 1 Fig 1 Cluster analysis of five sites in Kibale forest based upon presence or absence of 113 naturally occurring tree species. The letters (A and B) indicate clusters defined by using a minimum Jaccard’s Index of 0.7 (dashed line) Forest tree communities in Kibale forest 101 � 2007 African Journal of Ecology, Afr. J. Ecol., 45, 99–108 Kanyawara K-30 and K-14. There were no sites with negative scores on PC 1 (Table 2). Species with relatively high negative scores on PC 1 are shown in Table 3 but were not descriptive of any site as there were no sites with negative scores on PC 1 (Table 2). Species with loadings >0.062 on PC 2 were descriptive of Mainaro, a member of cluster A (Fig. 1). Kanywara K-14 had a relatively high negative score on PC 2 (Table 2) although there were no species with loadings >0.062 on PC 2 (Table 3). Friedman test revealed a significant dif- ference between clusters A and B on the basis of tree species (v2, P < 0.05). Species associations The two site clusters produced by CA were used to cat- egorize tree species into assemblages which represented common associations of species found in the sites of each cluster. Sixty-nine species occurred at more than 4% of 90 sampling quadrats in sites of cluster A. Species paired by CA which had simple matching coefficient >0.9 are shown in Table 4. Four species assemblages were pro- duced by CA for the sites in cluster A (Fig. 1; Table 5). The first species assemblage A1, had species common in the heavily logged Kanyawara K-15. The second assem- blage A2, had common and some species restricted at Mainaro study site only (Baphiopsis parviflora, Celtis mildbraedii and Phyllanthus discoideus). The third assem- blage, A3, had species common in the unlogged forest types at Ngogo and Kanyawara K-30 and the lightly logged forest Kanyawara K-14. The fourth assemblage, A4, had more species than A1–A3 combined. It was characterized by species common in Ngogo, Mainaro and the heavily logged Kanyawara K-15: Piptadeniastrum africanum, Bequaertiodendron oblanceolata, Warburgia ugandensis and Harrisonia abyssinica were mainly at Ngogo site. Olea welwitschii, Kigelia moosa and Myrianthus arboreus were common in the K-15 site. One hundred and three species occurred at more than 4% of 60 sampling quadrats in cluster B. Species paired by CA which had a SM >0.9 are shown in Table 4. Four species assemblages were produced. Species assemblages B1 and B2 had species common in the unlogged K-30 and the lightly logged K-14 sites at Kanyawara with the exception of Uvariopsis congensis which was more common at Mainaro than Kanyawara. Assemblage B3 had many species common in both unlogged K-30 and Ngogo sites, while assemblage B4 had many species common to both Ngogo and Mainaro sites and rare at K-30 site. Discussion Similarity of sites This study has improved the tree species inventory over previous lists. Struhsaker (1997) reported 51 species for K-30, -23, -18, and 52 for K-14, K-15 and Ngogo respectively. Chapman & Chapman (1997) recorded 88 and 29 species for Ngogo and Mainaro respectively. The lack of significant difference in species richness between logged and unlogged forest types indicates forest recovery after 28 years of postlogging. The findings agree with those of Cannon, Peart & Leighton (1998). This species richness of the logged forest is due to the recruitment of secondary colonizers. It may not indicate recovery to the original forest state. The relatively undisturbed forest at Ngogo and the heavily disturbed forest of K-15 had similar floristic com- position. This is possibly due to both forests being mixed forests characteristic of the central block. Heavy logging in K-15 resulted into growth of the secondary forest charac- terized mainly by Celtis–Diospyros–Markhamia–Strombosia– Newtonia mixed forest (Eilu, Hafashimana & Kasenene, 2004). Sheil, Sayer & Obrien (1999) noted that logging increases heterogeneity of the forest microhabitats and provides considerable space for colonization by good dis- persers and disturbance dependent species. The descriptive species typical of cluster A (Table 3) are more abundant in areas that experienced logging. Cluster B had the lightly logged forest (K-14) with the same number and type of tree species as the unlogged (K-30) forest. This indicates that the impact of logging on floristic composition appears to be dependent on the intensity of logging. These sites experienced lower effects of Table 2 Scores of the five sites on principal components axes 1 and 2 (PC 1 and PC 2) and percent of total variance in species richness explained by each PC axis Sites PC I PC 2 K-30 0.567 )0.073 K-14 0.57 )0.101 K-15 0.371 )0.021 Ngogo 0.45 )0.026 Mainaro 0.12 0.992 % of total variance (cumulative in parentheses) 47.4 (47.4) 19.9 (67.3) 102 Patrick Mucunguzi et al � 2007 African Journal of Ecology, Afr. J. Ecol., 45, (Suppl. 3) 99–108 Table 3 Loadings (correlations) of the original variables (species) on principal component axes 1 (PC I) and 2 (PC 2) for the five sites in Kibale forest Species PC 1 PC 2 Species PC I PC 2 Aningeria altissima 0.096 0.041 Irvingia gabonensis 0.008 )0.048 Aphania senegalensis 0.096 0.041 Newtonia buchananii 0.008 )0.048 Blighia welwitschii 0.096 0.041 Oncoba routledge 0.008 )0.048 Celtis africana 0.096 0.041 Strychnos mitis 0.008 )0.048 Celtis durandii 0.096 0.041 Symphonia globulifera 0.008 )0.048 Chaetachme aristata 0.096 0.041 Albizia glaberrima 0.008 0.05 Diospyros abyssinica 0.096 0.041 Zanthoxylum leprieurii 0 0.05 Dombeya mukole 0.096 0.041 Alangium chinense )0.003 )0.044 Fagaropsis angolensis 0.096 0.041 Schrebera arborea )0.003 )0.044 Funtumia latifolia 0.096 0.041 Ehretia cymosa )0.011 )0.043 Markhamia lutea 0.096 0.041 Bersama abyssinica )0.022 )0.036 Monodora myristica 0.096 0.041 Hallea rubrostipulata )0.022 )0.036 Premna angolensis 0.096 0.041 Tarenna pavetoides )0.046 )0.041 Prunus africana 0.096 0.041 Clausena anisata )0.046 0.059 Trema orientalis 0.096 0.041 Piptadeniastrum africanum )0.046 0.059 Trilepisium madagascariensis 0.096 0.041 Pseudospondias microcarpa )0.046 0.059 Uvariopsis congensis 0.096 0.041 Bridelia micrantha )0.053 0.06 Bombax buonopozense 0.085 )0.053 Warburgia ugandensis )0.053 0.06 Cassipourea ruwensorensis 0.085 )0.053 Erythrina excelsa )0.057 )0.034 Craterispermum laurinum 0.085 )0.053 Pterygota mildbraedii )0.057 )0.034 Ilex mitis 0.085 )0.053 Rauvolfia vomitaria )0.057 )0.034 Neoboutonia melleri 0.085 )0.053 Syzygium guineense )0.057 )0.034 Olea welwitschii 0.085 )0.053 Vangueria apiculata )0.057 )0.034 Parinari excelsa 0.085 )0.053 Chionanthus mildbraedii )0.065 )0.034 Strombosia scheffleri 0.085 )0.053 Majidea fosteri )0.065 )0.034 Tabernaemontana pachysiphon 0.085 )0.053 Sapium ellipticum )0.065 )0.034 Dasylepis eggelingii 0.061 0.043 Strombosiopsis tetrandra )0.065 )0.034 Dictyandra arborescens 0.061 0.043 Baphiopsis parviflora )0.088 0.062 Ficus exasperata 0.061 0.043 Bequaertiodendron oblanceolatum )0.088 0.062 Milletia dura 0.061 0.043 Celtis mildbraedii )0.088 0.062 Polyscias fulva 0.061 0.043 Celtis zenkeri )0.088 0.062 Spathodea campanulata 0.061 0.043 Chrysophyllum albidum )0.088 0.062 Ficus sansibarica 0.054 0.043 Cussonia holsti )0.088 0.062 Balanites wilsoniana 0.05 )0.051 Cynometra alexandri )0.088 0.062 Chrysophyllum gorungosanum 0.05 )0.051 Elaeophorbia sp )0.088 0.062 Coffea canephora 0.05 )0.051 Ficus conraui )0.088 0.062 Cordia millenii 0.05 )0.051 Ficus mucuso )0.088 0.062 Leptonychia mildbraedii 0.05 )0.051 Glyphaea brevis )0.088 0.062 Lovoa swynnertonnii 0.05 )0.051 Harrisonia abyssinica )0.088 0.062 Mimusops bagshawei 0.05 )0.051 Lepidotrichilia volkensii )0.088 0.062 Myrianthus arboreus 0.043 )0.05 Trichilia sp1 )0.088 0.062 Rubiaceae 0.043 )0.05 Albizia gummifera )0.1 )0.032 Teclea nobilis 0.043 )0.05 Antidesma sp )0.1 )0.032 Croton megalocarpus 0.043 0.048 Beilschmiedia ugandensis )0.1 )0.032 Chionanthus sp 0.032 )0.046 Craibia brownii )0.1 )0.032 Kigelia moosa 0.032 )0.046 Ficus natalensis )0.1 )0.032 Randia urcelliformis 0.031 )0.043 Ficus ureolaris )0.1 )0.032 Allophylus abyssinica 0.019 0.045 Harungana madaqascariensis )0.1 )0.032 Erythrina abyssinica sp abyssinica 0.019 0.045 Lannea welwitschii )0.1 )0.032 Forest tree communities in Kibale forest 103 � 2007 African Journal of Ecology, Afr. J. Ecol., 45, 99–108 Table 3 Continued Species PC 1 PC 2 Species PC I PC 2 Pancovia turbinata 0.019 0.045 Macaranga schweinfurthii )0.1 )0.032 Phyllanthus discoideus 0.019 0.045 Maerua duchesnei )0.1 )0.032 Pleiocarpa pycnantha 0.019 0.045 Memecylon sp )0.1 )0.032 Xymalos monospora 0.019 0.045 Morus lactea )0.1 )0.032 Albizia grandibracteata 0.008 )0.048 Trichilia sp2 )0.1 )0.032 Antiaris toxicaria 0.008 )0.048 Trichilia splendida )0.1 )0.032 Casearia battisombei 0.008 )0.048 Voacanga thouarsii )0.1 )0.032 Dovyalis macrocarpa 0.008 )0.048 Table 4 Species that occurred at more than 4% of 90 sampling quadrats in sites of cluster A and cluster B paired by cluster analysis with a simple matching coefficient (SM) >0.9 Cluster A Cluster B Albizia glaberrima and Allophylus abyssinica Phyllanthus discoideus and Pancovia turbinata Harrisonia abyssinica and Warburgia ugandensis Symphonia globulifera and Aningeria altissima Bequaertiodendron oblanceolatum and Cussonia holstii Pleicarpa pycnantha and Newtonia buchananii Pleiocarpa pycnantha and Ficus mucuso Randia urcelliformis and Monodora myristica Polyscias fulva and Milletia dura Polyscias fulva and Ficus sansibarica Schrebera arborea and Ficus exasperata Balanites wilsoniana and Antiaris toxicaria Cordia millenii and Balanites wilsoniana Dictyandra arborescens and Craterispermum laurinum Craterispermum laurinum and Bersama abyssinica Dovyalis macrocarpa and Casearia battisombei Teclea nobilis and Hallea rubrostipulata Spathodea campanulata and Rubiaceae Strombosiopsis tetrandra and Rubiaceae Lovoa swynnertonnii and Albizia grandibracteata Trema orientalis and Bridelia micrantha Morus lactea and Mitragyna rubrostipulata Randia urcelliformis Eggeling and Dombeya mukole Memecylon sp and Majidea fosteri Piptadeniatrum africanum and Dasylepis eggelingii Maerua duchesnei and Macaranga schweinfurthii Ehretia cymosa and Bombax buonopozense Lepidotrichilia volkensii and Lannea welwitschii Myrianthus arboreus and Chionanthus mildbraedii Harungana madaqascariensis and Harrisonia abyssinica Sapium ellipticum and Majidea fosteri Glyphaea brevis and Ficus ureolaris Lovoa sywnnertonii and Leptonychia mildbaedii Ficus natalensis and Ficus mucuso Celtis mildbraedii and Baphiopsis parviflora Ficus conraui and Erythrina abyssinica ssp. abyssinica Elaeophorbia sp. and Cynometra alexandri Cussonia holsti and Craibia brownii Clausena anisata and Chrysophyllum albidum Chionanthus mildbraedii and Celtis zenkeri Celtis mildbraedii and Bridelia micrantha Bequaertiodendron oblanceolatum and Bersama abyssinica Baphiopsis parviflora and Beilschmiedia ugandensis Antidesma sp. and Albizia gummifera Trichilia sp1 and Syzigium guineense Strombosiopsis tetrandra and Sapium ellipticum Rauvolfia vomitaria and Pterygota mildbraedii Pseudospondias microcarpa and Piptadeniastrum africanum Vangueria apiculata and Voacanga thouarsii Trichilia sp2 and Trichilia splendida Zanthoxylum leprieurii and Kigelia moosa Warburgia ugandensis and Croton megalocarpus Ehretia cymosa and Chionanthus sp. Fagaropsis angolensis and Albizia glaberrima. 104 Patrick Mucunguzi et al � 2007 African Journal of Ecology, Afr. J. Ecol., 45, (Suppl. 3) 99–108 Table 5 Species associations’ analysis of the tree species of Kanyawara K-15, Ngogo and Mainaro (cluster A) and of K-14 and K-30 (cluster B) Cluster A1 Celtis durandii Diospyros abyssinica Markhamia lutea Cluster A2 Aphania senegalensis Baphiopsis parviflora Celtis mildbraedii Phyllanthus discoideus Trilepisium madagascariensis Uvariopsis congensis Cluster A3 Aningeria altissima Cassipourea ruwensorensis Funtumia latifolia Ilex mitis Leptonychia mildbraedii Lovoa swynnertonnii Monodora myristica Parinari excelsa Premna angolensis Pterygota mildbraedii Strombosia scheffleri Tabernaemontana pachysiphon Cluster A4 Albizia glaberrima Allophylus abyssinica Balanites wilsoniana Bequaertiodendron oblanceolatum Bersama abyssinica Blighia welwitschii Bombax buonopozense Bridelia micrantha Celtis africana Chaetachme aristata Chionanthus mildbraedii Chionanthus sp Cordia millenii Craterispermum laurinum Croton megalocarpus Cussonia holsti Cynometra alexandri Dasylepis eggelingii Dictyandra arborescens Dombeya mukole Ehretia cymosa Fagaropsis angolensis Ficus exasperata Ficus mukuso Harrisonia abyssinica Table 5 Continued Kigelia moosa Majidea fosteri Milletia dura Mimusops bagshawei Mitragyna rubrostipulata Myrianthus arboreus Neoboutonia melleri Olea welwitschii Piptadeniastrum africanum Pleicarpa pycnantha Polyscias fulva Pseudospondias microcarpa Randia urcelliformis Rubiaceae Sapium ellipticum Schrebera arborea Spathodea campanulata Strombosiopsis tetrandra Teclea nobilis Trema orientalis Warburgia ugandensis Zanthoxylum leprieurii Cluster B1 Trilepisium madagascariensis Uvariopsis congensis Cluster B2 Celtis durandii Chaetachme aristata Diospyros abyssinica Dombeya mukole Dovyalis macrocarpa Funtumia latifolia Markhamia lutea Teclea nobilis Cluster B3 Albizia grandibracteata Aningeria altissima Antiaris toxicaria Aphania senegalensis Balanites wilsoniana Blighia welwitschii Bombax buonopozense Casearia battisombei Cassipourea ruwensorensis Celtis africana Chrysophyllum gorungosanum Cordia millenii Craterispermum laurinum Dictyandra arborescens Erythrina abyssinica sp abyssinica Ficus exasperate Ficus sansibarica Forest tree communities in Kibale forest 105 � 2007 African Journal of Ecology, Afr. J. Ecol., 45, 99–108 logging than the sites in cluster A. The species descriptive of cluster B (Table 3) are characteristic of undisturbed areas in Kibale forest representing climax forest tree com- munities (Kasenene, pers. comm.). However, these results do not preclude the fact that both abiotic and biotic factors are important in characterizing tree species assemblages in sites. Due to a paucity of data on several important factors, we can only speculate as to the other potential sources of these differences in floristic composition between sites other than the differences in logging history. They include small variations in rainfall in different sites, soil composition, elevation, temperature and Table 5 Continued Ilex mitis Leptonychia mildbraedii Lovoa swynnertonnii Milletia dura Mimusops bagshawei Monodora myristica Morus lacteal Myrianthus arboreus Neoboutonia melleri Newtonia buchananii Olea welwitschii Pancovia turbinata Parinari excelsa Phyllanthus discoideus Pleicarpa pycnantha Polyscias fulva Polyscias fulva Premna angolensis Randia urcelliformis Rubiaceae Spathodea campanulata Strombosia scheffleri Strychnos mitis Symphonia globulifera Tabernaemontana pachysiphon Tarenna pavetoides Cluster B4 Alangium chinense Albizia glaberrima Albizia gummifera Allophylus abyssinica Antidesma sp Baphiopsis parviflora Beilschmiedia ugandensis Bequaertiodendron oblanceolatum Bersama abyssinica Bridelia micrantha Celtis mildbraedii Celtis zenkeri Chionanthus mildbraedii Chionanthus sp Chrysophyllum albidum Clausena anisata Coffea canephora Craibia brownie Croton megalocarpus Cussonia holsti Cynometra alexandri Dasylepis eggelingii Ehretia cymosa Elaeophorbia sp Erythrina excelsa Table 5 Continued Fagaropsis angolensis Ficus conraui Ficus mukuso Ficus natalensis Ficus ureolaris Glyphaea brevis Harrisonia abyssinica Harungana madaqascariensis Irvingia gabonensis Kigelia moosa Lannea welwitschii Lepidotrichilia volkensii Macaranga schweinfurthii Majidea fosteri Memecylon sp Mitragyna rubrostipulataB4 Oncoba routledge Piptadeniastrum africanum Prunus africana Pseudospondias microcarpa Pterygota mildbraedii Rauvolfia vomitaria Sapium ellipticum Schrebera arborea Strombosiopsis tetrandra Syzigium guineense Trema orientalis Trichilia sp1 Trichilia sp2 Trichilia splendida Vangueria apiculata Voacanga thouarsii Warburgia ugandensis Xymalos monospora Zanthoxylum leprieurii Species assemblages (A1–A4) and (B1–B4) were defined using a simple matching coefficient of 0.0. 106 Patrick Mucunguzi et al � 2007 African Journal of Ecology, Afr. J. Ecol., 45, (Suppl. 3) 99–108 historical differences in the distribution and abundance of large mammals (Struhsaker, Lwanga & Kasenene, 1996; Chapman et al., 1997). Primates and ungulates are known to influence the floristic composition of tropical forests via their roles as seed dispersers (Nchanji & Plumptre, 2003). Species distribution Many tree species had widespread distributions including Aningeria altissima, Aphania senegalensis, Blighia welwitschii, Celtis africana, Celtis durandii, Chaetachme aristata, Diospyros abyssinica, Dombeya mukole, Fagaropsis angolensis, Funtumia latifolia, Markhamia lutea, Monodora myristica, Premna angolensis, Prunus africana, Trema orientalis, Trilepisium madagascariensis and U. congensis. These were recorded in all the five sampled sites. The most abundant species however, included C. durandii, D. abyssinica, M. lutea, F. latifolia and U. congensis. These were recorded in at least 48% of the plots sampled. Species whose distributions were either mostly or entirely restricted to one compartment or site included C. mildbraedii, Chrysophyllum albidum, Cusso- nia holsti, Cynometra alexandri, Elaeophorbia sp., Erythrina excelsa, Ficus conraui, Ficus mukuso, Glyphaea brevis, H. abyssinica, Lepidotrichilia volkensii, Majidea fosteri, Pterygota mildbraedii, Rauvolfia vomitaria, Sapium ellipticum, Strombosiopsis tetrandra, Syzigium guineense, Tarenna pave- toides, Trichilia sp1 and Vangueria apiculata. The least common species included Coffea canephora, Irvingia gabon- ensis, Oncoba routledge, P. africana, Xymalos monospora and Alangium chinense. These were recorded in less or equal to 4% of the sampled quadrats. In conclusion, it is evident that Kanyawara K-30, K-14, K-15 and Ngogo forests found in the north and central part of Kibale National Park, represent a mixed forest lar- gely influenced by varying logging intensities in the past. The climax forest is in the southern part of the park where Mainaro site is located. Within each forest type, there were different species associations and these appear to be a reflection of the species that have regularly occurred to- gether in these sites. Further investigation is required to include other forest parts not considered in this study. Acknowledgements This study was funded by Makerere University, WWF and Makerere University Biological Field Station. Permission to conduct this research was generously given by the Uganda National Council for Science and Technology. References Brugiere, D., Gautier, J.-P., Moungazi, A. & Gautier-Hion, A. (2002) Primate diet and biomass in relation to vegetation composition and fruiting phenology in a rain forest in Gabon. Int. J. Primatol. 23, 999–1024. Butynski, T.M. (1990) Comparative ecology of blue monkeys (Cercopithecus initis) in high- and low-density subpopulations. Ecol. Monogr. 60, 1–26. Cannon, C.H., Peart, D.R. & Leighton, M. (1998) Tree species diversity in commercially logged Borneon rainforest. Science 28, 1366–1368. Chapman, C.A. & Chapman, L.J. (1997) Forest Regeneration in logged and unlogged Forests of Kibale National Park, Uganda. Biotropica 29, 396–412. Chapman, C.A., Chapman, L.J., Wrangham, R., Isabirye-Basuta, G. & Ben-David, K. (1997) Spatial and temporal variability in the structure of a tropical forest. Afr. J. Ecol. 35, 287–302. Davies, A.G. (1994) Colobine populations. In: Colobine Monkeys: Their Ecology, Behaviour and Evolution (Eds A. C. Davies and J. F. Oates). Cambridge University Press, Cambridge. Eilu, G., Hafashimana, D. & Kasenene, J.M. (2004) Density and speciesdiversity of trees in four tropical forests of the Albertine rift, Western Uganda. Biodivers. Res. 10, 303–312. Gauch, H.G. (1982) Multivariate Analyses in Community Ecology. Cambridge University Press, Cambridge, England. Ghigleiri, M.P. (1984) The Chimpanzees of Kibale Forest: A field study of Ecology and social structure, 1st edn. Columbia University Press, New York. Hart, J.A. (2001) Diversity and abundance in an African forest ungulate community and implications for conservation. In: African Rain Forest Ecology and Conservation (Eds W. Weber, L. J. T. Whim, A. Vedder and L. Naughton-Treves). Yale University Press, New Haven. Howard, P.C. (1991) Nature Conservation in Uganda’s tropical forest reserve. IUCN publication, Switzerland. Howard, P.C. & Davenport, T.R.B. (1996) Forest Biodiversity Reports, Vol. 1–33. Uganda Forest Department, Kampala. Jongman, R.H.G., Ter Braak, C.J.F. & Van Tongeren, O.F.R. (1995) Data Analysis in Community and Landscape Ecology. Cambridge University Press, Cambridge. Kasenene, J.M. (1987) The influence of mechanical selective logging, felling intensity and gap size in the regeneration of a tropical moist forest in the Kibale Forest Reserve. Ph.D Dissertation, Michigan State University, Uganda. Kent, M. & Coker, P. (1992) Vegetation description and analysis: a practical approach. John Wiley & Sons, Chichester, England. Kovach, W.I. (1999) A Multivariate Statistical Package for Win- dows, Version 3.1. Kovach computing services, Pentraeth, Wales. Ludwig, J.A. & Reynolds, J.F. (1988) Statistical Ecology: A Primer on Methods and Computing. John Wiley and Sons, New York. Mentis, M.T. (1983) Towards objective veld condition assessment. Proc. Grassl. Soc. South. Afr. 18, 77–80. Forest tree communities in Kibale forest 107 � 2007 African Journal of Ecology, Afr. J. Ecol., 45, 99–108 Nchanji, A.C. & Plumptre, A.J. (2003) Seed germination and early seedling establishment of some elephant-dispersed species in Banyang-Mbo Wildlife Sanctuary, south-western Cameroon. J. Trop. Ecol. 19, 229–237. Økland, R.H. (1990) Vegetation ecology: theory, methods and applications with reference to Fennoscandia. Sommerfeltia 1, 4–233. Osmaston, H.A. (1959) Working plan for Kibale & Itwara forests. Uganda Forestry department, Entebbe. Parthasarathy, N. (1999) Tree diversity and distribution in undisturbed and human-impacted sites of tropical wet ever- green forest in southern Western Chats, India. Biodivers. Conserv. 8, 1365–1381. Phillips, O.L., Hall, P., Gentry, A.H., Sawyer, S.A. & Vasquez, R. (1994) Dynamics and species richness of tropical rain forests. Proc. Natl Acad. Sci. U.S.A. 91, 2805–2809. Pielou, E.C. (1984) Interpretation of Ecological Data. John Wiley and Sons, New York. Polhill, R.M., Milne-redhead, E., Turrill, W.B. & Hubbard, C.E. (1952) Flora of Tropical East Africa (in many parts). Crown Agents, London and Balkema, Rotterdam. Proctor, J., Anderson, J.M., Chai, P. & Vallack, H.W. (1983) Ecological studies in four contrasting lowland rain forest types in Gunung Mulu National Park, Sarawak. I. Forest environ- ment, structure and floristics. J. Ecol. 71, 237–260. Sheil, D., Sayer, J.A. & Obrien, T. (1999) Tree species diversity in logged tropical rainforests. Sciencemag 284, 5420. Skorupa, J.P. & Kasenene, J.M. (1984) Tropical forest management: can rates of natural tree falls help guide us? Oryx 18, 96–101. Skorupa, J.P. (1988) The effects of selective timber harvesting on rain - forest primates in Kibale Forest. PhD. Thesis, University of Cali- fornia Davis, Uganda. Struhsaker, T.T. (1997) Ecology of an African Rain Forest; Logging in Kibale and the Conflict between Conservation and Exploitation. University of Florida, Gainesville. Struhsaker, T.T., Lwanga, J. S. & Kasenene, J.M. (1996) Elephants, selective logging and forest regeneration in the Kibale Forest, Uganda. J. Trop. Ecol. 12, 45–64. Swamy, P.S., Sundarapandian, S.M., Chandrasekar, P. & Chandr- asekaran, S. (2000) Plant species diversity and tree population structure of a humid tropical forest in Tamil Nadu, India. Bio- divers. Conserv. 9, 1643–1669. Ter Steege, H., Sabatier, D., Castellanos, H., Van Andel, T., Dui- venvoorden, J., DE Oliveira, A.A., EK, R., Lilwah, R., Maas, P. & Mori, S. (2000) An analysis of the floristic composition and diversity of Amazonian forests including those of the Guiana Shield. J. Trop. Ecol. 16, 801–828. Van Orsdol, K.G. (1983) The status of Kibale Forest Reserve of Western Uganda and recommendations for its conservation and management. Report to the New York Zoological Society, Bronx, New York. (Manuscript accepted 20 August 2007) 108 Patrick Mucunguzi et al � 2007 African Journal of Ecology, Afr. J. Ecol., 45, (Suppl. 3) 99–108