The study was conducted in Dellomena and Harenna Buluk districts of Bale Eco-region in Southeast Ethiopia (Figure 1). The study area is characterized by bimodal rainfall with mean annual rainfall of 850 mm and mean annual temperature of 18˚C (Gole & Senbeta, 2008). It encompasses flatlands; moderately steep rolling hills with valley bottoms and waterways. Harenna forest is the largest montane forest in the study eco-region (Senbeta & Denich, 2006; Gole & Senbeta, 2008). It supports over 300 plant species with many endemic plant species (Senbeta, 2006). The unique floristic composition includes Podocarpus falcatus, Ocotea kenyensis, Filicium decipiens and Warburgia ugandensis. Forest coffee occurs between 1300 and 1850 meter above sea level of Harenna forest and dominated by woody species and the understory coffee plants. Majority of the populations in the two districts are settled in the lower southern edging of the

forest and some even in the forest. All people in the study area have coffee plots and beehive, and graze their livestock’s in the forest. Their livelihoods mainly depend on forest exploitation, crop and livestock production, and collection of non-timber forest products (Gole & Senbeta, 2008).

First a reconnaissance survey was made to collect baseline information and to observe vegetation distribution of forests coffee and adjacent natural forest in Harenna Buluk and Dellomena districts. Two sites that encompass both forest coffee and adjacent natural forests were selected purposely from each district. Then, 16 transect lines with 1200 m average length were laid out in both forest coffee and adjacent natural forest. Five main plots of 20 m × 20 m were established on each transect line at interval of 200 m. Forty main plots with a total area of 1.6 ha were established in each forest types to identify woody species. This method was developed in line with Gole (2003). Saplings and seedlings were identified in five subplots of 5 m × 5 m within each main plot for each forest type. Accordingly, 80 main plots (20 m × 20 m) and 400 subplots (5 m × 5 m) were laid in the two forest types with the total area of 3.2 ha. From reconnaissance survey, the total sampled area was assumed to be representative considering that both forest types are under uniform topographical conditions, altitudinal range and woody species distribution.

Individuals of woody plant species with diameter at breast height (DBH) of larger than 5 cm were identified and counted in the main plots. Saplings (individuals with DBH from 5 to 2.5 cm and height greater than 1 m), and seedlings (individuals with height ≤ 1 m and DBH less than 2.5 cm) were identified and counted within the subplots. Diameter measurements for trees and sapling/seedlings were taken at breast height (1.3 m) and at root collar by using caliper and diameter tape, respectively following (Senbeta & Teketay, 2001). For trees branched below 1.3 m, DBH measurement of each branch was taken independently. Hypsometer and graduated stick were used for height measurements. Geographic location and elevation of each plots was taken using GPS. All woody species in the sample plots were identified by scientific name using the modern Flora of Ethiopiaand Eritrea (Friis, 2009). Moreover, forest coffee management practices were identified through transect walk and key informants’ interviews. For key informants’ interview 12 elderly people who are knowledgeable about the forest resource in the study site were involved.

Shannon-Wiener Diversity Index (H’) was used to determine floristic composition of forest coffee and adjacent natural forest (Shannon & Weaver, 1948). Evenness or Equitability index (E) was determined to measure of evenness (Krebs, 1999). Species richness was determined by converting the average number of species in forest coffee and adjacent natural forest into hectare bases, and expressed as number of species per hectare. Simpson’s diversity index was determined (Magurran, 1988). Sørensen’s Similarity Index (β) was used to evaluate floristic similarity between forest coffee and natural forest (Sørensen, 1948). Density of woody plants in each forest types was determined as number of individual per hectare. Then, the variation in woody species richness, dominance, and Simpson, Shannon and Equitability indices between forest coffee and adjacent natural forest was tested by using independent t-test at 95% significance level by using SPSS statistical Software version 20.0.

Key informants reported that they have collected non-timber forest products like honey, medicinal plants and coffee from the natural forest. However, they have only the right to sell pole of felled trees to Bale Forest and Wildlife Enterprise and utilize for household consumption. As a result, dwellers mainly managed forest coffee areas to increase productivity of coffee plants through slashing of bushes and herbs, thinning of trees through cutting/debarking of stems and cultivation of the land. They also confirmed that lians, herbs, shrubs and trees were cleared from forest coffee areas to reduce the competition of these vegetation with wild coffee plants and to ease management and harvesting of coffee. Similarly, Gole and Senbeta (2008) reported that during opening up phase; small trees, shrubs, and herbaceous vegetation competing with coffee are totally cleared without preferential even for endemic or threatened species. KIs also declared that they slash herbs, tree/shrub seedlings to reduce their competition with understory coffee before the commencement of rainy season and coffee harvesting operation. Moreover, KIs indicated that the productivity of coffee increases as the density of shade trees decrease. This result agrees with Beer et al. (1998), and Faminow and Rodriguez (2001) who revealed that un-shaded systems could produce greater coffee yields. Key informants also reported that they practiced thinning of large trees to allow exposure of coffee plants to solar radiation. However, the canopies of the remaining trees expand, and gradually close up after some years to the level that can highly reduce coffee production. Reduction of shade through killing shade trees is common practice in the forest coffee areas even if it is not allowed legally. Key informants clearly depicted that they gradually cleared other woody plant species and retain only those tree species that they believe to increase coffee productivity. Cordia africana, Croton macrostachyus, Millettia ferruginea, Ekebergia capensis, Podocarpus falcatus, Pouteria adolfi-friederici, Diospyros abyssinica, Olea capensis, and Olea welwitschii are trees species preferred by farmers for coffee shade (Gole & Senbeta, 2008).

It was observed that larger trees were killed furtively as cutting trees is prohibited in forest coffee areas. Key informants declared that most of forest dwellers have killed large trees through debarking of their stem and cutting of their main roots below or at ground level. A key informant amazingly expressed how secretly he killed large trees as “I dug an elbow pit around a large tree, cut its main roots, removed barks of the stem, returned back the soil into the pit and cover it with partially decomposed organic matter. How one can identify what I did? Thus, the trees seam naturally dead or felled.”. Similarly, Mengist et al. (2013) revealed that cutting tree and tree ringing were practiced in forest coffee of Belete Gera Forest. Hoeing/cultivation is an emerging practice in the forest coffee system to facilitate good rooting condition for coffee plants and to avoid competing understory vegetation. Key informants declared that hoeing was first practiced by illegal settlers in the forest and gradually adopted by others. Key informants also indicated that coffee management activities like hoeing and slashing have negative effects on diversity of woody and non-woody plant species.

There are 30 families identified in natural forest and forest coffee (Table 1 and

Table 2), respectively. Of which Oleaceae, Rutaceae and Rubiaceae were the plant families that contain large number of woody species in the natural forest, each represented by four species. The Ebenaceae family was represented by three woody species in natural forest. In forest coffee, Rutaceae and Rubiaceae were the dominant families in terms of number of woody species in which each were represented by four species. Oleaceae and Sapindaceae were the next dominant families with three species each in forest coffee. Apocyanceae, Capparidaceae, Celastraceae, Myrsinaceae, Rhamnaceae and Salvadoraceae were families with shrub life forms which were not found in forest coffee areas (Table 1). This indicates that shrub species were cleared from forest coffee to reduce competition with coffee plants. Similarly, clearing of shrubs in forest coffee areas was reported in Harenna forest and in Belete Gera Forest (Gole & Senbeta, 2008; Mengist et al., 2013).

Forty-seven woody plant species belonging to 29 families were identified in natural forest (TableA1), whereas, 39 woody plant species belonging to 24 families are identified in forest coffee (TableA2). Celtis africana, Diospyros mespiliformis, Podocarpus falcatus, Strychnos mitis, Diospyros abyssinica, Filicium decipiens, Teclea nobilis, Cassipourea malosana, Galiniera saxifraga and Coffea arabicawere frequently observed in the natural forest. In forest coffee, Coffea arabica, Celtis africana, Podocarpus falcatus, Diospyros mespiliformis, Strychnos miti and Diospyros abyssinica were frequently observed woody species. The absolute frequency of Filicium decipiens, Teclea nobilis and Olea welwitschii

recorded in the adjacent natural forest were reduced from 75% to 25%, 70% to 25% and 65% to 5% in the forest coffee. However, its value increased in the case of Coffea arabica and Croton macrostachyus from 60% to 100% and from 25 to 50% respectively. This implies that abundance of coffee and some shade tree species were increased by reducing other woody species that thereof affect woody species composition in forest coffee. Similarly, retention or planting of preferable shade tree species and clearance of other species were reported in coffee forest and plantations in different part of the world (Ambinakudige & Sathish, 2009; Mengist et al., 2013; Likassa, 2014).

The species richness of natural forest and forest coffee were 19 ± 0.9 and 12.55 ± 0.73 respectively (Table 1). It showed significant difference (P = 0.001) between forest coffee and adjacent natural forest. The above results indicate that more woody species were found in adjacent natural forest as compared to the forest coffee. Similarly, research findings of Ambinakudige and Sathish (2009), Hylander and Sileshi (2009) and Likassa (2014) revealed higher tree species diversity in adjacent natural forests than in shade coffee farms. The effect of coffee management practices has affected negatively woody species diversity and richness in the forest coffee. Moreover, local people retained only selected shade trees as over storey trees in the system. Similarly, Wassie et al. (2009) also declared the negative effect human induced disturbance on species diversity in forest coffee.

Tree is the dominant life form of woody species in the natural forest and forest coffee. Tree constituted 63.5% and 71.4% of woody species in natural forest and forest coffee, respectively. Shrubs are the co-dominant life form of woody species, which contributes 11.91% and 11.54% in natural and forest coffee, respectively. This greater percentage of tree life form in forest coffee is due to retention of shade trees in the system.

The Simpson diversity index of 0.84 ± 0.012 and 0.74 ± 0.039, and Shannon diversity index of 2.15 ± 0.072 and 1.74 ± 0.098 were obtained in natural forest and forest coffee respectively (Table 3). Similarly, Mengist et al. (2013) found higher Shannon diversity index in forest without coffee (2.98) than in the forest with coffee (2.13). Senbeta and Denich (2006) found higher Shannon diversity index of 2.82 and 2.6 at Bebeka forest coffee and Harenna forest coffee respectively. Both Simpson (P = 0.02) and Shannon (P = 0.00) have significant higher value in natural forest than forest coffee in this study. These indicate that coffee management practices resulted in lower woody species diversity in forest coffee. Similarly, Engida and Teshome (2012) revealed that forests with low levels of disturbance have high species diversity as compared to disturbed forest. The above results imply that management of coffee forests has affected and will continue to affect the diversity of woody species in the forests. Similarly, Kufa (2006) reported that as coffee management continues, the coffee forest could be changed to coffee farms with a few shade trees.

Dominance of 0.26 ± 0.039 and 0.16 ± 0.012 were recorded in the forest coffee and adjacent natural forest respectively. The significantly (P = 0.02) higher value of dominance in forest coffee indicated presence of large number of individuals of few species. This variation in dominance may come from different management intervention undertaken in forest coffee like slashing, hoeing, weeding and human interferences. Similarly, Wassie et al. (2009) and Mekuria and Aynekulu (2011) indicated that human induced disturbance have a strong negative effect on species composition, seed germination, seedling growth, and mortality of many of the plant communities and in turn results in less species richness. Schmitt et al. (2009) also indicated that management interventions in semi-forest coffee shown a strong impact on tree species composition. Moreover, Walters et al. (2006) indicated that land use change affects the composition, diversity and distribution pattern of vegetation. Species equitability/evenness indices of natural forest and coffee forest were 0.73 ± 0.02 and 0.69 ± 0.04 respectively (Table 3). This indicates that about 70% of woody species are equally

available and distributed in both natural forest and forest coffee. Wilson et al. (1996) also revealed that a higher value of evenness indicates that most species are present with relatively equal individuals in a community. Moreover, Magurran (2004) indicated that the evenness measure of 1 revealed being complete evenness in a community.

Sørensen’s similarity index of 0.814 was found for natural forest and forest coffee. This implies that more woody species overlap between the two ecosystems. Similarly, Sørensen (1948) revealed that Sørensen’s similarity index of 0 indicates there is no species overlap between the communities and a value of 1 indicates exactly the same species are found in both communities. The finding of this study also agrees with the findings of Hylander and Sileshi (2009) and contradicts with the finding of Mendez et al. (2007).

Of 51 woody species identified in both forest types, 35 species (68.6%) were common to both forest types. However, 12 (23.5%) and 4 (7.9%) were unique to the natural forest and forest coffee respectively. This implies natural forest contains comparatively more unique woody species than the adjacent forest coffee. This agrees with the results of Likassa (2014) which reported presence of more unique species in natural forest as compared to forest coffee.

Frequency of occurrence of the species across sample plots varies between natural forest and forest coffee (Figure 2 and Figure 3). Five species in the natural forest and two species in forest coffee had an absolute frequency greater than 90%. Those species in natural forest include Celtis africana (100%), Podocarpus

falcatus (95%), Strychnos mitis (95%), Diospyros mespiliformis (95%) and Diospyros abyssinica (90%) while those of forest coffee includes Celtis africana (95%) and Podocarpus falcatus (95%). Furthermore, 20 species in the natural forest and 11 in the forest coffee were occurred in more than half (50%) of the total plots investigated. This implies that large number of species were recorded in few quadrats.

Smaller number of tree species frequently observed in forest coffee as compared to the adjacent natural forest. This is due to, the traditional management of shade tree in Ethiopia is to reduce tree density and understory vegetation to improve the production of coffee while maximizing the use of selected tree species (Aerts et al. 2011). Farmers selectively remove some tree species through various management techniques, including selection of tree species with desirable properties (Asfaw, 2003; Abebaw, 2006). For this reason, only few shade tree species with a greater economic or ecological value (shade) or both dominated the coffee-based forest system in the Harenna forest.