Forest Ecology and Management 261 (2011) 949–957 Contents lists available at ScienceDirect Forest Ecology and Management journal homepage: www.elsevier.com/locate/foreco Plant communities, species diversity, richness, and regeneration of a traditionally managed coastal forest, Kenya Staline Kibet ∗ Coastal Forest Conservation Unit, National Museums of Kenya, P.O. Box 596, Kilifi, Kenya a r t i c l e i n f o Article history: Received 12 July 2010 Received in revised form 22 November 2010 Accepted 26 November 2010 Available online 17 January 2011 Keywords: Kaya forests Regeneration Biodiversity Species association Cultural sites Conservation a b s t r a c t The Kenyan coastal forests make up one of the World 25 Biodiversity Hotspots. They consist of over 140 fragments (the majority with areas less than 0.5 km2 ) of the once extensive Zanzibar–Inhambane lowland moist forest. The over 60 known Mijikenda sacred Kaya forests and groves scattered along the coastal hinterland form the greater part of this ecosystem. The forests are of biological and cultural significance, and this has been recognized nationally and internationally, with some now listed as World Heritage Sites. The forests are protected by councils of Kaya elders who regulate use of their resources. Increasing human population and subsequent rise in demand for forest products and land for settlement has put a strain on these relic forests. Farm encroachment and extraction of forest products in different Kaya forests have affected the vegetation ecology at varying levels. This study investigated the spatial species distribution, association and regeneration potential of commonly utilized plants in one of these traditionally managed ecosystems. A modified nested plot method was used to collect data in the field. Using TWINSPAN multivariate, and indicator species analysis, two plant communities (Asteranthe and Bridelia) and an undifferentiated vegetation type were identified. Species association in Asteranthe consisted largely of forest dependant species, with a significant presence of woody climbers. It was comprised of two sub-communities namely Manilkara and Scorodophloeos. In contrast the second plant community, Bridelia, was dominated by light demanding species. It comprised one sub-community (Catunaregam) and a seral stage (Keetia). The species diversity and richness was higher in the Asteranthe community compared to Bridelia. Some of the forest species commonly utilized by the local people were observed to regenerate both in open and closed forest habitats while others had seedling recruitment confined to closed forest. Despite some coastal forests showing physiognomic similarity, detailed study shows intra-variation linked to topography, exposition, type and intensity of human perturbation both currently and in the distant past. Clearly, vegetation patterns of coastal forests of eastern Africa change at fairly short intervals. Recruitment of forest specialists is likely to decline if closed forests are opened up by farm encroachment, however their less specialized counterparts can pioneer in re-colonization of disturbed sites if conservation is strengthened. There is need to invigorate traditional management systems of forests with cultural significance by recognizing and giving increased legal mandates to the local custodians. © 2010 Elsevier B.V. All rights reserved. 1. Introduction The Kenya coastal forest, which constitutes part of the once extensive Zanzibar–Inhambane lowland moist forests, is one example of an ecosystem under threat in eastern Africa (Janzen, 1988; White, 1983). It is estimated at 660 km2 (Burgess et al., 2000; Waiyaki, 1995) with a significant number of the fragments having an area less than 0.5 km2 . Many of these fragments are the Mijik- ∗ Current address: East Africa Herbarium/Botany Department, National Museums of Kenya, P.O. Box 40658-00100, Nairobi, Kenya. Tel.: +254 20 37412131/2/3/4x2205; fax: +254 20 3741424. E-mail addresses: skibet@museums.or.ke, skibet1@yahoo.com 0378-1127/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.foreco.2010.11.027 enda sacred Kaya forests: there are over 60 Kaya forest patches. They range in area from as little as 2 ha to over 500 ha and are scattered over a distance of 200 km in the administrative Counties of Kwale in the south to Kilifi in the North. They owe their existence to the culture and beliefs of nine closely related ethnic groups collectively called the Mijikenda. Historically, the Kaya forests were used by these communities as a refuge against aggression from unfriendly neighbors, so cutting of trees and other form of vegetation destruction inside and around the sites was strictly prohibited (Spear, 1978). The Kaya forests have gain recognition since the early 1990s after the first comprehensive inventory of coastal vegetation was carried out (Robertson, 1987; Robertson and Luke, 1993). They have been recognized nationally and internationally as important natu- 950 S. Kibet / Forest Ecology and Management 261 (2011) 949–957 ral and cultural landscapes as exemplified by their registration as National Monuments under Antiquities and Monuments Act (Cap 215) laws of Kenya and most recently when several were listed as World Heritage Sites (UNESCO, 2008). Many vegetation types along the Kenyan Coast have been described in varying degrees of details; however few studies have targeted the sacred Kaya forests (Clarke and Robertson, 2000). Studies done in the Kaya forests and surrounding areas (Kibet and Nyamweru, 2008; Pakia, 2000; Nyamweru et al., 2008; Tengeza, 1999; Waiyaki, 1995) indicate that these forest patches are biologically and culturally diverse. Unfortunately, several of these sites despite their high conservation value and cultural significance have continued to suffer destruction and degradation. Pressure from increased demand for firewood, timber, mining, and more land for farming and settlement linked to increasing human population, and the development of the tourism industry are threatening these forest patches (Kibet and Nyamweru, 2008; Nyamweru et al., 2008). Fear of divine retribution played a significant role in the enforcement of traditional rules. Unfortunately, the decline in the traditional values and the authority of elders who are the custodians of the forests have made the system less effective. It is of utmost important that these forest patches are effectively protected. To design appropriate and focused conservation measures, there is a need for quantitative data that can be used to empirically define the vegetation communities and identify their uniqueness. Currently these data are inadequate in some areas and completely lacking in other coastal forests (Clarke and Robertson, 2000). In view of increasing threats faced by Kaya forests, this study was carried out to understand spatial species distribution, plant association, and regeneration potential of commonly utilized plant in traditionally managed ecosystem. The specific objectives were to (1) identify plant communities, (2) identify species diversity, richness and associations, (3) assess recruitment and regeneration of commonly utilized species and conservation options. 2. Materials and methods 2.1. Study area The study was conducted in a Rabai sacred forest named Kaya Mudzimuvya, located between 39◦ 34′ E to 39◦ 36′ E and 3◦ 56′ S to 3◦ 57′ S in Rabai location of Kilifi County of coastal Kenya (see Fig. 1).Topographically, the area is gently undulating to undulating, with altitudes ranging between 20 and 200 m above sea level. The area experiences bimodal rainfall patterns with long rains (Mwaka) between March and July and short rains (Vuri) October–December with a dry period between January and March. Average annual rainfall ranges between 1100 and 2000 mm (Boxem et al., 1986). The minimum and maximum temperatures range 22.5–24.5◦ C and 26.5–34◦ C respectively. Kaya Mudzimuvya forest is one of the five remaining sacred forests of the Rabai community; the other four are Kaya Mudzimwiru, Kaya Bomu, Kaya Fimboni, and Kaya Mzizima. These forests are under the protection of Kaya elders who are the custodians of the sacred forests in the area. Regrettably, the authority of the elders has been declining over the years coupled with social, economic, and political dynamics that has allowed some of the sites (e.g. Kaya Bendeje) to be encroached upon. The population density of Rabai area based on 1999 census was 493 persons per square kilometer (Republic of Kenya, 2000). Subsistence agriculture and animal husbandry are the main land activities in areas neighboring the Kaya forests. Despite the enormous influence of new religion and western civilization, a significant portion of the population remains committed to their culture and traditions which are hallmark on the Kaya-based rituals and ceremonies. In 1998, 171.3 ha of Kaya Mudzimuvya forest was registered as a National Monument (NM) under the Antiquities and Monument Act (Cap 215), Laws of Kenya as a measure to protect it. 2.2. Sampling design A random nested plot method modified from Hall and Bawa (1993) was used for sampling. The modifications made included the use of randomised plots instead of plots placed along predetermined transects, reducing sampling area for trees from 1.0 ha to 0.4 ha and using four 2 m × 2 m plots for seedlings instead of one. The plots measuring 20 m × 20 m were used to enumerate trees (individuals with diameter at breast height – DBH > 10 cm), 10 m × 10 m for saplings/shrubs (with DBH < 10 cm and >1 cm) and 2 m × 2 m for herbs and seedlings. One 10 × 10 m plot was placed within one quarter of the 20 m × 20 m plot while four 2 m × 2 m plots were placed at each of the corners of the 20 m × 20 m plots. Random plots were generated using a random number generator and a global positioning system (GPS) receiver was used to locate them on the ground. All individual species were identified following Beentje (1994) procedure. Heights of all individuals from 1 m and above were estimated, basal diameter (BD), DBH, and their densities per plot recorded. Cover for three strata, trees, shrubs, and herbs, was estimated. In each sampling plot, indicators of disturbances (e.g. grazing), species harvested, and presence of bundles of firewood/poles of harvested species were recorded. 2.3. Analysis of data The raw data initially recorded in Excel sheet were subjected to Systat program version 8.0 (SPSS Software Inc., 1998), whereby volumes, basal area and density were calculated per hectare. The figures generated were keyed into statistical software; PC-ORD version 4 program whereby TWINSPAN multivariate analysis (McCune and Mefford, 1999) was used to analyze phytosociological association. Basal area figures were used as cover-abundance values in the analysis. Shannon diversity index (H′ ) was calculated using the  ′ formula H = − {(ni/N) log (ni/N)} (Misra, 1989). 3. Results Two hundred and three plots, covering 8.12 ha and representing 4.7% of the study area, were sampled. A total of 355 woody species were identified and 320 of them run in TWINSPAN multivariate analysis as described by Hill (1979) to determine species association. Some species were, however, expunged to reduce ‘noise’. A two-way ordered matrix table was used to identify plant clusters that were then named based on dominant species. 3.1. Plant communities Two plant communities (Asteranthe and Bridelia) and an undifferentiated vegetation type were identified. The Asteranthe community consisted of two sub-communities namely Manilkara and Scorodophloeos, while the Bridelia community had one subcommunity and one seral stage namely Catunaregam and Keetia respectively (see Fig. 2). 3.1.1. Asteranthe community The diagnostic species for the Asteranthe community were Asteranthe asterias and Combretum illairii. Tree layer (canopy) preferential species included Scorodophloeos fischeri, Manilkara sansibarensis and Haplocoelum inoploeum. The other layers were S. Kibet / Forest Ecology and Management 261 (2011) 949–957 951 Fig. 1. The Rabai Kaya Forests including the study site. covered by shade tolerant species shown in Table 1. Woody climbers such as Ancylobotrys petersiana, C. illairii, and Uvaria acuminata enjoined the lower stratum to the canopy. The Asteranthe community largely occupied the central part of the closed forest along the ridge top, as well as the slopes along River Kombeni to the East. 3.1.1.1. Manilkara sub-community. In the Manilkara subcommunity the indicator species was M. sansibarensis, an evergreen tree that can reach 25 m tall at maturity. The species did not however form the canopy layer as expected, because most individuals were resprouting stumps. In its place, Spiciformis (Brachystegia spiciformis), Gum Copal Tree (Hymenaea verrucosa) and Afzelia (Afzelia 952 S. Kibet / Forest Ecology and Management 261 (2011) 949–957 Kaya Mudzimuvya Forest vegetation Asteranthe community Closed forest Bridelia community Open Wooded Bushland/Shrubland/Grassland N= 65 N= 138 Scorodophloeos sub-community N=37 B Manilkara subcommunity Transitional vegetation (ecotone) N=19 N=9 Catunaregam sub-community Undifferentiated vegetation type N= 52 N= 63 Keetia seral stage N=23 Fig. 2. The plants community clusters identified in study site. Table 1 Species that defined plant communities’ strata. Communities Asteranthe Bridelia Preferential/indicator species Tree layer Shrub layer Herb layer Haplocoelum inoploeum, Manilkara sansibarensis, Scorodophloeos fischeri Bridelia carthatica, Cocos nucifera, Anacardium occidentale, Mangifera indica Asteranthe asterias, Combretum illairii, Vepris trichocarpa, Acalypha fruticosa Allophyllus rubifolius, Aspilia mossambicensis, Premna chrysoclada Psilotrichum sclerethum quanzensis) constituted the canopy layer. Preferential species that formed the understory layers are listed in Table 2. Common woody climbers of the sub-community included Synaptolepis kirkii, Hugonia castaneifolia, Apodostigma pallens, and Artabotrys modestus among others. 3.1.1.2. Scorodophloeos sub-community. The Scorodophloeos subcommunity was characterized by dominant presence of dry evergreen forest species namely; Scorodophloeos fischeri, Mildbraedia carprinifolia and Commiphora eminii. Preferential tree and shrub layer species that dominated this sub-community are listed in Table 2. Unlike the Manilkara sub-community where climbers were significant, only a few species such as Grewia forbesii, G. holstii and Hibiscus faulkenarae were observed in the sub-community. Vernonia hildebrandtii, Agathisanthemum bojeri 3.1.2. Bridelia community The Bridelia community was dominated by Bridelia cathartica, a forest margin or bush land thicket species. Preferential species that defined the tree, shrub and herb layers are indicated in Table 1. Bridelia community occurred in open woody shrub land and thickets, largely in sites previously disturbed by grazing and cultivation that have been undergoing regeneration in the last 40 years. 3.1.2.1. Catunaregam sub-community. The species Catunaregam nilotica characterized the Catunaregam sub-community. Despite the species being common in several habitats such as wooded grassland, several coastal bush lands, and palm woodland, it was not represented in the Asteranthe community at all and Table 2 Indicator species in the vegetation sub-communities. Sub-communities Manilkara Scorodophloeos Catunaregam Preferential/indicator species Tree layer Shrub layer Herb layer Manilkara sansibarensis, Brachystegia spiciformis, Hymenaea verrucosa, Afzelia quanzensis Scorodophloeos fischeri, Commiphora eminii, Gyrocarpus americanus, Manilkara sulcata, Combretum schumannii, Dobera loranthifolia, Lecaniodiscus fraxinifolius Vitex payos, Acacia nilotica, A. etbaica and A. mellifera, Strychnos madagascariensis, S. spinosa Phyllanthus welwitschianus, Suregada zanzibarensis, Heinsia crinita, Ritchiea capparoides Mildbraedia carprinifolia, Tricalysia ovalifolia, Memecylon fragrans, Croton pseudopulchellus, Uvariodendron kirkii Dichrostachys cinerea, Ormocarpum kirkii, Harrisona abyssinica Chazaliella abrupta var. abrupta, paveta stenosepala Vitellariopsis kirkii, Scorodophloeos fischeri Aganthisanthemum bojeri, Vernonia hildebrandtii S. Kibet / Forest Ecology and Management 261 (2011) 949–957 Table 3 Some recorded species that were confined to specific identified sub-communities. Manilkara Scorodophloeos Catunaregam Drypetes natalensis Pleiocarpa pycnantha Chazaliella abrupta Allophylus pervillei Artabotrys modestus Ludia mauritiana Diospyros consolatae Xylopia parviflora Pseudobersama mossambicensis Clerodendrum incisum Strophanthus kombe Erythrina sacleuxii Ophrypetalum odoratum Memecylon sansibaricum Memecylon fragrans Craibia brevicaudata Pycnocoma littoralis Uvaria faulknerae Cola minor Croton pseudopulchellus Ricinodendron heudelotii Grewia stulhmannii Synadenium pereskiifolium Vitex payos Catunaregam nilotica Acacia etbaica Acacia nilotica Acacia mellifera Dichrostachys cinerea Omorcarpum kirkii Hibiscus altissima Sterculia schliebenii Combretum tenuipetiolatum Scorodophloeos fischeri Cynometra suaheliensis only once within the Keetia seral stage. Other species unique to this sub-community included wooded grassland species such as Dichrostachys cinerea and Vitex payos. The vegetation type was characterized by several thorny species, especially in the shrub layer. 3.1.2.2. Keetia seral stage. Unlike the earlier mentioned plant communities, no diagnostic species was identified for Keetia seral stage however; forest margins species such as Keetia zanzibarica and Rytigynia celastroides as well as weedy Stachytarpheta jamaicensis were common. Moreover, short-lived perennials such as Waltheria indica, Agathesanthemum bojeri, Tinnea aethiopica and some agricultural tree crops were well represented. Some species occurred in more than one sub-community while some were exclusive to one sub-community or community. Table 3 indicates some of the species that occurred exclusively in some of the identified sub-communities. 3.2. Species diversity and richness Statistically, the Asteranthe community recorded higher species diversity and richness than the Bridelia community with values of 1.83 and 38 compared to 1.77 and 28 respectively. There was a decline of woody species from 260 in the Asteranthe to 233 in the Bridelia community. On average the Asteranthe community recorded 6 trees and 21 understory species per plot while the Bridelia community recorded an average of 3 trees and 18 understory species. Trees species general diversity declined from 1.65 to 1.61 while understory species declined from 2.01 to 1.93 in Asteranthe to Bridelia respectively. The most species rich plot was found in the Bridelia community with 57 species. There were 7 other plots with more than 50 species each. Of all the 8 plots with over 50 species each, 6 of them were found in the Asteranthe community majority within ecotone zone. The 3 plots with the least species richness (less than 10 species each) occurred in the Bridelia community. In general, closed forest sub-communities (Manilkara and Scorodophloeos) had on average the highest species richness. The Manilkara sub-community with 28 plots was the richest, with species ranging from 21 to 56 (mean = 43, S.D. = 8.1) while the Scorodophloeos sub-community, represented by 37 plots, was second with species ranging from 19 to 49 (mean = 34, S.D. = 7.1). The Catunaregam sub-community represented by 63 plots was species poor with a range of 11–57 (mean = 27, S.D. = 9.5). 953 seedling stage in more than 10 plots sampled; other species occurred in lower frequencies. Some of the taxa had their seedlings exclusively confined to the closed forest and others in the open woody bush land/shrub land. Cynometra suaheliensis, C. webberi, M. sansibarensis and J. magnistipulata seedlings occurred solely within the closed forest under parent plants. In contrast, the seedlings of forest margin species such as Strychnos madagascariensis, Premna chrysoclada, and Uvaria lucida occurred exclusively in the open woody bush land. A. quanzensis, Combretum schumannii, H. verrucosa and V. kirkii seedlings occurred in both the closed forest and open woody shrub land. Over 40% of H. verrucosa and all Parkia filicoidea seedlings recorded in sampling plots occurred in the open woody shrub land under cashew (Anacardium occidentale) trees. Few H. verrucosa seedlings grew under their parent plants, irrespective of whether such mother plant occurred in the closed forest or open shrub land. The majority of P. filicoidea seedlings were found in closed forest under the mother plant outside sample plots. In comparison, understory and climber species had more seedlings than canopy species both in the Asteranthe and Bridelia communities. A majority of commonly utilized species found in the Bridelia and Asteranthe communities regenerated more as resprouters than as reseeders. 4. Discussion Although the lowland coastal dry forest of Kaya Mudzimuvya is currently protected from indiscriminate human destruction, vegetation continues to be impacted by the extraction of forest products as well as by livestock grazing. The past and present extraction of firewood, poles, withers, fibers, and weaving materials has shaped the phytosociological associations, structure and composition of species and this has a bearing on the choice of possible conservation options. Previous studies in the region revealed that variation in vegetation is attributed to ecological factors such as soil characteristics, topography, wind direction, and level of forest disturbance (Clarke and Robertson, 2000; Schmidt, 1991). In this study major vegetation types seemed shaped largely by anthropogenic factors, whereas intra community variation point to other ecological dynamics. Fig. 3 shows a hypothetical diagrammatic illustration of vegetation dynamics as driven by both biotic and abiotic factors. As envisaged in the above diagram, any extensive clearing of forests, coupled with consistent grazing alters Asteranthe closed forest into Bridelia open woodland vegetation type and eventually into grassland. The reverse option through natural succession is possible; however its success rate depends on the frequency, duration and intensity of perturbation instigated. The dynamics are less systemic if human perturbations vary in space and time and this is demonstrated by observed plant communities’ variability. The forest clearing and farm encroachment in the past caused degeneration of Manilkara forest type to a Keetia-Rytigynia formation/vegetation type. Continued grazing and extraction of firewood, poles, withers and other non timber forest products has not only facilitated the establishment of short-lived perennials and other weedy species but also made the formation unstable. Further evolution of this vegetation type is described under Sections 4.1 and 4.3. 3.3. Recruitment and regeneration of species 4.1. Plants communities Commonly harvested species for building houses such as Vitellariopsis kirkii, Scorodophloeos fischeri, H. verrucosa, Julbernardia magnistipulata and Millettia usaramensis were represented at The Bridelia community largely covered sites used in the past as cultivation and grazing fields by the Rabai people. These sites stretched from down slope at the riverine habitat 954 S. Kibet / Forest Ecology and Management 261 (2011) 949–957 Manilkara Sub-community Keetia seral stage Cultivation, NTFPs extraction Low past perturbations & abiotic factors High present perturbations & abiotic factors Asteranthe community Bridelia community Slash and burn cultivation, livestock grazing High past perturbations (logging) & abiotic factors Scorodophloeos sub-community Open grassland Vegetation Low present perturbations Catunaregam subcommunity Human induce vegetation changes Natural vegetation dynamics (succession and natural disturbances). NTFPs – Non Timber Forest Products Fig. 3. Vegetation dynamics due to biotic and abiotic variability. to as high as the ridge-top adjacent to Asteranthe closed forest. It is interpreted therefore that persistent anthropogenic pressures over many years could have created Bridelia community from Asteranthe. The presence of isolated huge (>50 cm DBH) forest canopy species as well as shade tolerant species in dominantly open shrub land vegetation supports this assertion. The diagnostic as well as preferential species found in the Bridelia community were either generalists such as Allophylus rubifolius, Premna chrysoclada and Tinnea aethiopica, and therefore found in several habitats, and/or bushland–grassland specialists (Beentje, 1994). The lack of well defined species association within Keetia seral stage could be ascribed to its unstable state. The Keetia seral stage occurs within close proximity to human settlement and thus suffers from frequent disturbance from grazing and wood harvesting. It is dominated by short-lived perennials and exotic species creating a human-induce vegetation formation whose ecological evolution is less predictable. Dominance of short-lived perennials with few seedlings of climax forest species corroborates this position. Although the exotic species were numerically few, some were significantly huge and dominant so that they greatly influenced their immediate ecological unit. Occasionally, cashew (A. occidentale) trees “hosted” Rytigynia celastroides, Trichilia emetica, Keetia zanzibarica, P. filicoidea and H. verrucosa underneath them. The mutual relationship between the species is not clear. However there are two likely scenarios; first the A. occidentale may have been acting as a ‘nurse plant’ for the associating species, thus promoting their development. Secondly, it may be strategically providing good resting points for dispersers of the seeds of the associating species, thus promoting their propagation in the process. Notably all associating species were juvenile (either saplings or seedlings) as compared to their ‘host plant’ thus indicating later arrival. Whitmore (1991) reported a similar observation in Mexico whereby tree relics on the farms left after cultivation provided perches for birds, some of them seed dispersers, predicted to help the reinvasion of climax species. It is predicted therefore that with enhanced conservation, Keetia seral stage will evolve into a KeetiaRytigynia “sub-community”. Over the years H. verrucosa and P. filicoidea will take over, creating a Manilkara-like type of vegetation. Unlike Keetia seral stage, species association in Catunaregam sub-community is better defined, with assemblages of deciduous thorny species. The dominant presence of Vitex payos and Catunaregam nilotica, both wooded grassland tree species (Beentje, 1994), indicates some level of plant community stability. It might take many years for forest species to take over if natural succession progresses unperturbed. The sub-community can be compared with the Kaya Mtswakara thickets and wooded vegetation (Pakia, 2000) and Acacia thorn-bushland (Moomaw, 1960). The subcommunity has significant population of Acacia spp and other thorny species such as Strychnos spp., Harrisonia abyssinica and Dichrostachys cinerea and could best be described as the “thorn in the flesh – functional group” or association. They are largely deciduous species and sometimes form impenetrable thickets that are highly susceptible to fire during the dry season. The herbaceous layer commonly consists of annuals. Unlike the Keetia seral stage, where anthropogenic influence is conspicuously displayed by the presence of huge exotic fruit trees, less than 10 such individuals occurred in the Catunaregam sub-community. The Rabai people used tree crops to mark and lay claim of ownership to a piece of land, which were planted soon after one has cleared the forests and cultivated it for a couple of seasons, to thwart potential ‘squatters’ given that the land was communally owned (Kibet, 2002). The presence of fewer exotic plants in this secondary vegetation could mean that the Catunaregam sub-community is a creation of a relatively “more recent” human disturbance later than 1970, when most of the western part of forest had already been cleared for cultivation. Despite earlier perturbations the Catunaregam sub-community seemed to have steadily regenerated from the early 1980s, after individu- S. Kibet / Forest Ecology and Management 261 (2011) 949–957 als who had encroached on the forest were forcefully evicted by the elders with the help of the government (Nyamweru et al., 2008). The close proximity of Catunaregam and Scorodophloeos subcommunities and thus the possibility of shared edaphic factors points to a possible succession toward later sub-community type of vegetation if future perturbations are curtailed. The presence of secondary forest species such as Commiphora eminii and Gyrocarpus americanus in the two sub-communities supports this argument. It also presupposes that any form of cultivation, grazing or extraction of non-timber forest products from Scorodophloeos will cause the sub-community to degenerate into a Catunaregam type of vegetation dominated by thorny generalists species but devoid of forest dependant species, and ultimately into a grassland if the frequency and intensity of perturbation persist. In contrast, the Asteranthe community occupied areas with cultural sites where plant extraction, grazing, and cultivation are prohibited, meaning that the vegetation has remained ‘relatively’ undisturbed for more than 100 years compared to other sites. Taxa associations showed concentrations of forest, dense bush land, or thicket specialists. This observation agrees with Pakia (2000) description of dense canopy forest of Asteranthe consisting of Scorodophloeos and Hugonia communities in Kaya Mtswakara forest, 25 km away from the study site. Given that most of the taxa are trees, any extraction could greatly alter the microclimate and subsequently floristic composition and structure. The association of species in the Scorodophloeos sub-community closely relates to legume-dominated dry forest (Clarke and Robertson, 2000) due to fair representation of Caesalpiniaceae (S. fischeri, Cynometra webberi, C. suaheliensis and Julbernardia magnistipulata), Manilkara sulcata and M. sansibarensis. The subcommunity is stable, as indicated by dominance of S. fischeri at the tree as well as the shrub layer (Clarke and Robertson, 2000). Other members of the shrub layer included M. carprinifolia, Memecylon fragrans and Grandidiera boivinii, which form a thick understory. The sub-community occurred on the eastern side of the forest stretching from the mid slope downwards and forming the riverine ecosystem along Kombeni River. Contrary to expectation this sub-community had a low population of woody climber species, unlike the situation in the nearby Kaya Kambe forest, where selective logging of huge trees had triggered massive invasion by lianas (Hawthorne, 1984). The Manilkara sub-community was largely confined within the ridge top at the middle of the Kaya but with a small patch surrounded by the Scorodophloeos sub-community at the lower slopes. In terms of anthropogenic disturbance, the sub-community has not experienced significant extraction of large trees due to the presence of prayer shrines and burial grounds – cultural sites highly revered by the local community (Nyamweru et al., 2008). However, occasional natural tree fall would occur especially among the old Brachystegia trees, creating tree gaps. Unlike in the Scorodophloeos, where lianas were scarce, Manilkara sub-community had strong presence of S. kirkii, H. castaneifolia, A. pallens and A. modestus forest woody climbers. As earlier conceptualized in Fig. 3 and expounded in Section 4.1, Manilkara sub-community is by no means static, natural tree fall and selective pole harvesting instigates ecological cascade within the ‘equilibrium’, unless greater perturbations where huge trees are removed and seedlings destroyed to cause significant change in species composition and structure. Physiognomically, Manilkara and Scorodophloeos subcommunities appeared similar and thus without detailed sampling it is easy to miss out species unique to each vegetation type. Despite sharing several species, each support species that are strictly confined to either of the sub-communities (see Table 3). The presence of species exclusive to one sub-community though adjacent to each other is subject to debate given that the altitudinal range and extend of forest is fairly small therefore segregation 955 due to elevation is unlikely. Possible explanations include: (1) ecological differences such as edaphic factors and the presence of micro-habitats; some of the species are highly specialized for certain niches such as riverine; (2) some preferred species could have been selectively harvested in one sub-community; (3) the influence of topography and site exposure might bear on species assemblages and distribution. This phenomenon indicates that coastal forest vegetation changes in species composition and association in fairly short time intervals, and this could be a feature that has contributed to making coastal forests biologically diverse with high species endemism. 4.2. Species diversity and richness In terms of species diversity and richness, the Asteranthe and Bridelia communities varied significantly, a scenario best explained by variability in abiotic factors as well as type, frequency and intensity of human perturbation both in the recent and the distant past. The general tree layer species diversity differed significantly between Asteranthe and Bridelia communities. The wider life forms of the Asteranthe community presuppose the presence of diverse ecological niches to support high diversity of species. The primary species germinates below closed canopies and get established into seedling banks to await favorable conditions to be ‘released’, while gaps are invaded by pioneers and woody climbers (Whitmore, 1991). The Asteranthe community is comprised of tall forest with interlocking canopies, with occasional tree gaps created by natural tree fall. The gaps together with forest edges provided ecotone microclimates suitable for both forest dependent and forest margin species to thrive. The two closed forest sub-communities (Manilkara and Scorodophloeos) differed in species diversity and richness. Species richness of woody climbers and understory specialists declined from Manilkara to Scorodophloeos. The comparatively higher tree species diversity in Manilkara than Scorodophloeos probably has something to do with past human perturbation that targeted the latter sub-community. Past logging was evident in Scorodophloeos sub-community where old stumps were recorded in several plots. Increased light penetration to ground level may have led to the development of a dense understorey, consisting of light demanding Acalypha fruticosa and M. carprinifolia (Hawthorne, 1984) at the expense of shade tolerant species. Most of the revered cultural sites and burial grounds located within the Manilkara subcommunity and the fear of the unknown for transgressing against the spirits kept potential culprits at bay (Nyamweru et al., 2008). The high number of species (57 species) in one plot within the Catunaregam sub-community occurred close to the forest edge, showing that light demanding, forest margin and a few forest dependant species were beginning to expand into the secondary vegetation in response to conservation measures currently in place. 4.3. Recruitment and regeneration Broadly, the seedlings of canopy species were few in number compared to those of shrub layer species. This could be attributed to the phenomenon where seedlings of several canopy species in Asteranthe were only confined to close forest. These closed forests occur on only a third of the total area sampled. Secondly, some species showed rather clumped distribution and thus those with large seeds may have had all of their seedlings concentrated underneath mother plants. This was the case with Caesalpinioids; Julbernadia magnistipulata, C. webberi and C. suaheliensis. In terms of conservation these species are seriously threatened by habitat destruction as their seedlings prefer closed forest to thrive. Furthermore, if such species are commonly preferred for use by the local people, they are prone to over exploitation as all the popula- 956 S. Kibet / Forest Ecology and Management 261 (2011) 949–957 tion are concentrated in a few areas thus making them easy targets for harvesting and transporting. Species whose seedlings grow both in the closed and open vegetation have higher chances of establishing themselves, and are likely to take over degraded habitat if human perturbation is removed. A. quanzensis, C. schumannii and H. verrucosa could easily constitute the first climax species to re-establish themselves in disturbed sites if parent plants are in close proximity to supply the germplasm. Species whose seedlings are exclusively confined to the open shrub land (S. madagascariensis, P. chrysoclada and U. lucida) are likely to gradually decline as forest species re-invade adjacent open shrub lands. The opposite is the case if wood extraction and opening up of forest recur. The seedlings of Parkia filicoidea and H. verrucosa consistently occurred underneath a number of A. occidentale individuals that were closer to forest edge. The agricultural tree crops formed the majority of the huge evergreen individuals in open shrub land, making them obvious stopover points for birds and mammals. The fruits of A. occidentale attract frugivores and small mammals like monkeys and bats, all of whom are major seed dispersers. This scenario needs further investigation as it may provide insight on the role of nurse plants and/or forest relics in degraded ecosystems. The forest gaps in the Asteranthe community were dominated by light requiring species that could have been from ‘seed-rain’, trans-located by mammals, borne by birds or wind, or from seed banks, thus increasing its diversity. Some species identified during the inventory work are of conservation concern. Bauhinia mombassae, Combretum tenuipetiolatum, Holarrhena pubescens and Euphorbia wakefieldii are classified as critically endangered in 1997 IUCN Red List of Threatened plants (Walter and Gillet, 1998) though they are currently undergoing review by plant experts with experience in the region. One such review took place in April 2010 at IUCN Eastern Africa Regional Office in Nairobi. All the species seems to have a specialised habitat. B. mombassae and C. tenuipetiolatum were common along Kombeni River while H. pubescens and E. wakefieldii were common on rocky areas and on forest edges. Due to their specificity to certain ecological niches, these species are in danger of becoming extinct if habitat destruction is not stopped. Other species requiring close monitoring include Encephalartos hildebrandtii and V. kirkii both recorded in Appendix 1 in the IUCN Red List data although in this current inventory their frequencies were fairly high compared to other species. Similarly Milicia excelsa, Terminalia sambesiaca and Brachylaena huillensis occurred in extremely low numbers. Single resprouting individuals of each species were noted and may need special attention. 5. Conclusions Physiognomic similarities among the coastal forests of eastern Africa do not necessarily mean similarity in plant communities and/or species diversity and richness. The findings of this study indicate that it is easy to overlook the uniqueness of plan t communities and presence of exceptional species that may require specialized attention. The anthropogenic factor is the single most important influence in shaping plant communities, species composition, recruitment/regeneration and grand pattern, reflecting of peoples’ culture and historical influence within the study site. The high species diversity within areas having cultural sites supports the need to promote and strengthen traditional management system within natural and cultural landscapes, so as to take advantage of cultural values of biodiversity and the local ecological knowledge. Recruitment and regeneration of some commonly harvested species is affected by their spatial distribution pattern, habitat suitability, and availability of mature individuals to provide the needed germplasm. Human input in the restoration could enhance the process if sound interventions are put in place. Introducing a few relatively fast growing non-invasive tree species (such as fruit trees) in strategic locations within degraded sites may hasten recolonisation by attracting pollinators and/or seed dispersal agents. Acknowledgements This study was part of an MSc research financially supported by UNESCO – People and Plants Initiative for which I will forever be indebted. I sincerely appreciate support from my supervisors; Dr. Robert Hoft, Dr. Enoch Mrabu and Dr. Elizabeth Omino. This work would not have been possible without the commitment of Drs. Robert Hoft and Martina Hoft who squeeze their free time to assist with data analysis. The Coastal Forest Conservation Unit project personnel help with field logistics for which I am truly grateful. I appreciate the assistance I received from Quentin Luke with species identification and John Charo who helped me during my fieldwork. I thank Celia Nyamweru for editing the paper. Lastly, I am indebted to the two reviewers for their edits, comments and critiques on the manuscript. References Beentje, H.J., 1994. Kenya Trees Shrubs and Lianas. National Museums of Kenya, Nairobi. Boxem, H.W., Meester, T., Smaling, E.M.A., 1986. Soils of the Kilifi Area. Reconnaissance Soil Survey Report No. R111987. Kenya Soil Survey. Ministry of Agriculture Kenya. Burgess, N.D., Clarke, G.P., Madgwick, J., Robertson, S.A., Dickinson, A., 2000. Distribution and status. In: Burgess, N.D., Clarke, G.P. (Eds.), Coastal Forests of Eastern Africa. IUCN, Gland, Switzerland and Cambridge, UK. Clarke, G.P., Robertson, S.A., 2000. Vegetation communities. In: Burgess, N.D., Clarke, G.P. (Eds.), Coastal Forests of Eastern Africa. IUCN, Gland, Switzerland and Cambridge, UK. Hall, P., Bawa, K., 1993. Methods to assess the impact of extraction of non-timber tropical forest products on plant population. Economic Botany 47 (3), 234–247. Hawthorne, W.D., 1984. Ecological and biogeographical patterns in the coastal forest of East Africa, DPhil Thesis, University of Oxford. Hill, M.O., 1979. TWINSPAN, a FORTRAN programs for arranging multivariate data in an ordered two-way table by classification of individuals and attributes. Ecology and Systematics Cornell University, Ithaca, NY. Janzen, D.H., 1988. Management of Habitat Fragments in Tropical Dry Forests, Growth. Annals, Missouri Botanical Garden, USA. Kibet, S., 2002. Human Disturbance and its Impact on Vegetation Structure, Composition and Regeneration of Kenya Coastal Forests (a case study of Kaya Mudzimuvya Forest). Unpublished MSc Thesis submitted to Jomo Kenyatta University of Agriculture and Technology, Nairobi, Kenya. Kibet, S., Nyamweru, C., 2008. Cultural and biological heritage at risk; the case of the Rabai Kaya forests in Coastal Kenya. Journal of Human Ecology 24 (4), 287–295. McCune, B., Mefford, M.J., 1999. PC-ORD multivariate analysis of ecological data, version 4. MJM software designs. Gleneden Beach, Oregon, USA. Misra, K.C., 1989. Manual of Plant Ecology, third ed. Oxford and IBH, New Delhi, India. Moomaw, J.C., 1960. A study of the plant Ecology of the Coast Region of Kenya, East Africa, Nairobi. Government Printer. Nyamweru, C., Kibet, S., Pakia, M., Cooke, A.J., 2008. The Kaya forest of Coastal Kenya; remnant patches or dynamic entities? In: Sheridan, M.J., Nyamweru, C. (Eds.), African Sacred Groves: Ecological Dynamic & Social Change. James Currey, UK, pp. 62–86. Pakia, M., 2000. Plant ecology and ethno botany of two sacred forests (Kayas) of Kenyan Coast. Unpublished MSc thesis for School of Life and Environmental Science, University of Natal, South Africa. Republic of Kenya, 2000. Population Census – Central Bureau of Statistics. Ministry of Planning and National Development, Nairobi. Robertson, S.A., 1987. Preliminary floristic survey of Kaya forests of coastal Kenya Unpublished reports to the Director NMK and to WWF, Nairobi, Kenya. Robertson, S.A., Luke, W.R.Q., 1993. The Vegetation and Conservation Status of Kaya Coastal Forests in Kenya. WWF, Nairobi. Schmidt, R., 1991. Ecology of a Tropical Lowland Rain Forest plant communities: soil characteristics and nutrients relations of the forests of the Shimba Hills National Reserve, Kenya, Band 179. Dissertationes Botanicae, Berlin-Stuttgart. Spear, T.T., 1978. The Kaya Complex. Kenya Literature Bureau, Nairobi, Kenya. SPSS Inc., 1998. Systat Software. Statistical Package for Social Sciences, USA. Tengeza, A.H., 1999. The Mijikenda Grave markers: A report of a study on timber and live tree grave markers and associated traditions in Kaya forests. In: Jianchu, X. (Ed.) Links Between Cultures and Biodiversity: Proceedings of the Cultures and Biodiversity Congress, 2000, Yunnan, China. UNESCO, 2008. http://whc.unesco.org/en/list/1231. S. Kibet / Forest Ecology and Management 261 (2011) 949–957 Walter, K.S., Gillet, H.J., 1998. 1997 IUCN Red List of Threatened Plants. Compiled by the World Conservation Monitoring Centre. IUCN – The World Conservation Union, Gland, Switzerland and Cambridge, UK. Waiyaki, E.M., 1995. Effects of forest fragmentation, isolation and structure, on the richness and abundance of bird communities in major coastal forests of south coast, Kenya. Unpublished MSc Thesis in Conservation 957 Biology. Durrell Institute of Conservation and Ecology, University of Kent, UK. Whitmore, T.C., 1991. Tropical rain forest dynamics and its implications for management. In: Gomez-Pompa, A., Whitmore, T.C. (Eds.), Rainforest Regeneration and Management MAB Series, vol. 6. UNESCO, Paris. White, F., 1983. The Vegetation of Africa. UNESCO, Paris.