These floral resources found in different vegetation types comprising in forests, woodlands and bush lands and others. The flora of Ethiopia have been investigated by several scholars since the beginning of the 19^th C to the mid of the 20^th C. Some studies have provided substantial contribution to describe vegetation types of Ethiopia. Pichi-Sermolli (1957); Chaffey (1979); Friis (1992); Sebsebe et al. (1996);  Zerihun (1999); Friis and Sebsebe (2001) and Sebsebe et al. (2011).These studies were carried out  in different parts of the Ethiopia and contributed in generating valuable botanical information for the Ethiopian flora.

The vegetation composition and structure of forests in southern and southwestern part of the country was also studied Lisane work Negatu L., (1987); Kumlachew and Tamrat (2002); Tadesse (2003); Feyera (2006); Ensermu and Teshome (2008). These studies have described floristic composition and analyzed plant communities and their relation with environmental factors. According to above studies several plant communities were identified, with characteristic plant species of Ilex mitis, Syzygium guineense, Pouteria adolfi-friederici, Olea welwitschii, Psychotria orophila, and Schefflera abyssinica in southwest Ethiopia. These species are also reported by (Friis, 1992) as the characteristic species of Afromontane rainforest in southwest Ethiopia. The vegetation of southwest Ethiopia varies with altitude and affecting the diversity of plant species. The study of variation of modern pollen rain (Bonnefille et al., 1993) along the ecological gradients containing range of vegetation types in southwest Ethiopia. This indicated that the vegetation types vary with altitude and altitudinal variation is an important environmental factor contributing for diversity of vegetation communities. Afromontane forests are the place of origin of the Coffea arabica and encompass a variety of commercially valuable spices and honey from wild bee. Furthermore the forests also play a pivotal role in providing water resources for the flow of the Baro-Akobo river system which is an important tributary of the Nile and it accounts for  42%  of  the  water  in  the  White Nile (NTFP, 2006). It is also important for carbon sequestration which has implications for climate change management.

 In spite of the ecological and economic role of Ethiopian forests, the forest cover of Ethiopia has declined by human impact. About 35%, of the country’s area was once covered by natural high forests, (EFAP, 1994). By the early 1950s, high forests were reduced to 16% and the country’s forests have declined at fast rate and reached 3.6% by 1980, 2.6% by 1987 (IUCN, 1990), 2.4% in 1992, (Sayer et al., 1992), and were finally reduced to 2.3% in 2003 (Shibru, 2003). Because of this shrinkage of the forest resources, most of the remaining forests of Ethiopia are restricted to the south and southwest parts of Ethiopia, which are less accessible, and less populated (Kumelachew and Tamrat, 2002). These forests are continuously threatened by human activities such as clearing forest for coffee and tea plantation, subsistence farming and periodic movements of immigrants from northern and southern parts of Ethiopia looking for fertile land, resulting in the loss of forest cover in the region.

Recognizing the above mentioned threats to forest biodiversity of afromontane rain forests, the government designed different strategies to conserve the remaining forest resources in the region. Participatory forest management (PFM) was one of the solutions to solve the problem of open access to forest resources and to promote sustainable forest management. For instance farm-Africa introduced the Participatory Forest Management approach in southwest Ethiopia particularly in Bonga and the implementation process has been developed since 1996. Agama forest is part of Bonga forest delineated for PFM since 1996 (Aklilu et al., 2014). Though the forest has been under protection since its demarcation, still it has been continuously exploited for agricultural land expansion, timber harvesting, firewood collection and charcoal production due to  lack of awareness on principles of PFM. Botanical assessment such as inventory of floristic composition, providing information on species diversity and community structure is necessary for the forest management and sustainable resource utilization by the community members. Therefore the current study aimed to assess the floristic composition and diversity of an Afromontane forests (Agama forest), to analyze community types of the forest and to evaluate the ecological relationships between plant communities and environmental parameters and to analyze phytogeographical relationship of Agama forests with other similar Afromontane forests types in Ethiopia and finally to recommend conservation action for protecting the forest biodiversity of the region.

The vegetation of the area belongs to Afromonatin rain forest and transitional rain forests (Friis et al., 2011).The forest was stratified into four different layers, namely, upper canopy, sub-canopy, shrub layer and the ground layer and Pouteria adolfi-friederici occupied emergent trees of the upper canopy. The area receives very high annual rainfall reaching up to 1830 mm in some peak years. The rainfall pattern shows low rainfall in January and February, gradually increasing to the peak period in July and then decreasing in November and December. Maximum and minimum monthly mean temperatures of the area are 26.6 and 9.5°C respectively.

Data collection

A systematic sampling design was used to collect vegetation and environmental data, following Kent and Cooker (1992). Vegetation data were collected using quadrates of 25 m x 25 m (625 m²), for woody plant species within which a 2 m x 2 m sub-plot was used for recording herbaceous species and soil sample collection. Ten line transect were established and the distance between each transect was 1km.The vegetation was sampled following the transect and located at 1 km apart and were laid systematically at every 500 m. The number of plots per transect is vary depending on length of the transect and accessibility of the sample plots.  All level of   altitudinal ranges starting from 1700 m up to 2300 m were covered during the sampling. A total of 60 releves were sampled and all vascular plant species found in each relevé were recorded and identified. The cover of all the vascular plant species found in each relevé was estimated and rated according to modified 1-9 Braun Blanquet approach (Vander, 1979). The specimens were collected and identified at the National Herbarium (ETH), Addis Ababa University using the published volumes of Flora Ethiopia and Eritrea by comparing with the authentic specimens in the National Herbarium.

Environmental data and soil analysis

The environmental parameters recorded in this study were altitude, slope, aspect, disturbance, soil pH and soil texture (sand, silt, clay), Altitude was measured by Garmin GPS and slope and aspect were measured using Silva Clinometer and 15T Silva Ranger Compass respectively. Soil analysis was performed in the soil  laboratories  of Addis Ababa University, following the procedures, (Sahlemedhin and Taye, 2000). The soil samples were analyzed for texture, using Hydrometer method of mechanical analysis and Sodium hexametaphosphate were used as dispersing agent. The pH measure was taken using Digital pH meter and it was standardized using buffer solutions of pH 4.0 and 9.2. Disturbance was determined on the basis a five point scale following Anderson and Currier (1973). Disturbance scores were based on visible signs of, tree cutting, grazing and presence of beehives. The points of scale were 0–5, with 0= (No disturbance), 1= (0-20% of the quadrat disturbed), 2= (21-40% of the quadrat disturbed), 3= (41-60% of the quadrat disturbed), 4= (61- 80% of the quadrat disturbed), 5 = (81-100% of the quadrat disturbed).

Phytogeographical comparison

Agama forest was compared with five other afromontane rain forests in Ethiopia for the purpose of investigating forest similarity and differences among the afromontane rainforests of Ethiopia.  These included Godre,Gera, Masha, Harenna, Jibbat and Yayu. Godre and Masha forests are close to Agama forest found in Bench Maji and Sheka zones in Southern Ethiopia Nations and Nationalities People’s Regional State. Jibbat forest is a transitional forest between dry evergreen afromontane and moist evergreen afromontane forest and found in western Shewa in Oromia region. Harenna Forest is located in the Bale Mountain National Park in south eastern Ethiopia.

Data analysis

An agglomerative Hierarchical Cluster Analysis was performed using Similarity Ratio (S.R), using R software version 2.1.5.2 (Venables et al., 2012). The cut point of the Clusters was decided after visual inspection of the level of aggregation/homogeneity of relevé. Plant community types were further refined in a Synoptic table. The resulting groups were recognized as community types and the species occurrences were summarized by synoptic cover abundance values. The community types were named based on the tree and shrub with high synoptic value.

Ordination was computed using Detrended Correspondence Analysis (DCA) is a technique that is used to display sample plots (sites). In DCA similar samples are plotted close together and dissimilar samples are placed far apart (Hill et al., 1980) (Figure 4).

Richness was calculated and Shannon and Wiener (1949) index was applied to quantify species diversity. Shannon diversity index (H’) was based on frequency of species    Where, “H” = Shannon and Wiener diversity index, S= number of the species Pi=the proportion of individuals or the abundance of species i^{th =} the proportion of total cover in the sample and ln = natural logarithms. The Pielou's (1966) J' evenness index (J) was calculated using the formula:     where J = evenness H’ = Shannon–Wiener Diversity Index S= total number of species in the sample and ln = natural logarithms.

A similarity analysis was carried out to evaluate the relationship between forests based on presence of trees, shrubs and herbs. Evaluation was conducted using Sorensen’s similarity index. It is described using the following formula (Kent and Coker, 1992). Ss = 2a/ (2a+b+c), Where, Ss = Sorensen’s similarity coefficient a = Number of species common to both samples; b= Number of species in sample 1; c = Number of species in sample 2. All the environmental parameters such as altitude, slope, aspect, disturbance and soil  data  were analyzed as follows: aspect was codified according to Zerihun Woldu et al. (1989): N = 0; NE = 1; E = 2; SE = 3; S = 4; SW =3.3, W= 2.5; NW = 1.3; Ridge top = 4. In order   to   examine   the   significant   differences and similarities between the community types identified, Tukey’s multiple tests were performed to detect significant differences between the community types for different environmental parameters (altitude, slope, aspect, and pH and soil texture). Pearson’s correlation was calculated to evaluate the relationship between the environmental parameters, anthropogenic disturbances (timber, charcoal and encroachment and grazing).

Species diversity, richness and equitability

From analysis of vegetation data using the Shannon-Wiener diversity index, community II had the highest species diversity (4.18), followed by communities IV (4.08) and III (3.96) and I (3.94). Community II had the highest number of species (110) followed by communities III (103) and IV (103), and I (79). Community I had thehighest evenness value (0.902) followed by communities II (0.89), Community IV (0.88) and III (0.85).

Plant community-environment relationship

The community types identified from cluster analysis showed significant difference with respect to altitude, slope, soil texture and number of species (Table 4). Community I was significantly different from the rest of the communities in terms of altitude and slope. All the communities identified in this study occur in slightly acidic soils (pH 4 and 5). There is no significant difference between plant communities for soil pH and sand content.

Community types II differed in clay and silt content from community type 1, 3 and 4.

The result of Pearson’s correlation of the environmental parameters shows that some of the environmental parameters were correlated (Table 5). Altitude is positively correlated with sand with R² = 0.64 and negatively correlated with clay and silt (-0.732 and -0.47) respectively.

Phytogeographical comparison

The Agama forests in southwestern Ethiopia are floristically related more to the Masha and Godre forest since they are situated in the same climatic region and geographical location. Accordingly  Masha and Godere forests share  similar  Sorensen similarity Index of (0.59) and (0.46) respectively (Table 6). Furthermore Harenna and Jibbat forests are less similar with lower similarity index (0.38 and 0.32) respectively.

Agama forest is one of the moist Afromontane rainforest and rich in plant biodiversity. It comprises an economically important plants used for coffee, spices and honey. These plant species are Coffee arabica, Aframomum corrorima, Piper capense and Schefflera abyssinica are the most frequent species in almost in all sample plots. Even though the size of the study area was limited; it had high number of plant species which is more or less comparable with that reported for the Afromontane and transitional rainforest vegetation in southwestern Ethiopia (Kumelachew and Tamrat, 2002) and Gara Ades forest in southeastern Ethiopia (Uhligsit and Uhlig, 1990). Agama forest contains a number of flowering plant species that are endemic to Ethiopia. The endemic plant species identified (9 species)  in this  study  is in agreement with similar studies with Abreham (2009) for Masha Andracha Forest, Derje (2002) for Gura Ferdea forest and Ensermu and Teshome (2008) for Bonga forest. Some of this species are in the IUCN Red Data list, were identified in Agama Forest. The number of endemic plant species recorded for the study area is small compared to dry afromontane forests, since the southwest moist montane forests are poor in trees/shrubs endemicity (Kumelachew and Simon, 2002).

Species diversity and evenness are used to interpret the relative variation among the communities. Lower evenness in Community III indicates the dominance of a few species such as Pouteria adolfi-friedericii (Engl.) Baehni, Schefflera abyssinica (Hochst. ex A. Rich.) Harms, Vepris dainellii (Pichi-serm.) Kokwaro, Dracaena afromontana Mildbr, and Acanthopale aethiogermanica Ensermu in the community and similar observation was reported by Abraham (2009) for Masha forest. On the other hand, high evenness in community I, II and IV indicates   little   dominance   by  any  single  species  but repeated coexistence of species over all plots in a community (Vepris dainellii). The highest species richness and diversity were observed in community II. The possible reason may be the altitude since it is found at mid altitudes which provides with optimal conditions of environmental factors that favor vegetation growth (Rosenzwieg, 1995). Some species may exclusive to live in mid, low and high altitudes depending on their physiological need and adaption (Austrheim, 2002). Community type I showed the lowest species richness and diversity. This difference is the result of differences in site productivity, habitat heterogeneity and anthropogenic influences such as selective removal of economically important trees and grazing by livestock. Local people reported  that  this  community  was  used  as  settlement area for indigenous people before introduction of PFM.  The altitude was relatively the important environmental factor that separates the four plant communities studied. Community I was significantly different from the rest of the communities since it is found at higher altitude (2343 m) and the communities II, III and IV are found at intermediate altitudes. This study is in agreement with Bonnefille et al. (1993) reported that the presence of altitudinal Zonation is delimiting vegetation types in southwestern Ethiopia and affecting atmospheric pressure, moisture and temperature which have again a strong influence on the growth and development of plants and the distribution of vegetation. Studies by Herdberg in 1951 also confirmed the altitude effect on vegetation in eastern African mountain.

The acidity of the soil in southwest Ethiopia is relatively higher as compare to other parts of the country due to intense breakdown of organic matter and leaching of the soil by heavy rainfall. This results in leaching of appreciable amounts of exchangeable basic ions like calcium (Ca), magnesium (Mg), sodium (Na) and potassium (K) from the surface of soil (Achalu et al., 2012). The soil pH decreases with increasing altitude and this could affect the chemical reaction between plant roots and nutrients, the availability of nutrients in the soil for plant use and microbial activity (Donahue et al., 1983). These could be the possible reasons for the decline of species richness and diversity with altitude. The differences in soil texture among the communities are not strongly significant due to limited size of the study area and the altitude ranges are not strongly significant to show variation among the plant communities.

The Agama forest was more floristically related to the Masha and Godre forests due to geographical proximity and located in similar climatic zone. Proximity of the areas favors seed dispersal and migration which result in a high floristic similarity. On the other Jibbat and Harenna forests are found in south eastern parts of the country and it has low floristic similarity due to variation in altitude soil, and climatic factors (rainfall, temperature). On top of this human influence on the forest resource also causes variation in floristic diversity. Bonnefille and Hamilton (1986), reported that the destruction of montane forest in southeastern Ethiopia as far back as ca. 2000 years and these historical factors may have led to the variation in floristic diversity between the southwest (Agama) and southeastern (Harenna).

The Agama forest, in southwest Ethiopia, has high floristic richness and diversity. Four community types were identified at an altitude between 1800 and 2371 m. The communities at the bottom and middle of the altitudinal gradient were richest while the community restricted to the top had less species. The Soil of Agama forest is acidic (with low pH) caused by excessive rainfall. The Agama forest is rich in plant biodiversity as compare to its limited size thus conservation of forest through, strengthening of existing PFM and provision of environmental education for forest user groups are highly recommended. 

The authors have not declared any conflict of interests.

The authors highly acknowledged for the financial support provided by thematic research division of Addis Ababa University and Biovision  foundation.  We  also  thank  the National Herbarium, Addis Ababa University, for letting to utilize the available facilities including flora books and specimens for plant identification.