2.1 Description of the Study Area
The study was conducted in Gomma District, Jimma zone, Oromia National Regional State, South-western Ethiopia. It is located at 395 km south west of Addis Ababa and about 45 km west of Jimma town. It is bordered on the south by Seka Chekorsa, on the southwest by Gera on the northwest by Setama on the north by the Didessa, on the northeast by Limmu Kosa and on the east by Manna. The district lies approximately between 7o 40’00’’ − 8o 03’30’’ N latitudes and 36o 25’00’’ − 36o 45’00’’E longitudes (Fig. 3.1); altitudinal ranges from 1,380 to 2870 meters above sea level. The mean annual rainfall is about 1524 mm with bi-modal distribution.
2.2 Sampling Technique and Sample Size
The study was conducted among from 36 rural kebeles where plantations/stands of the selected three exotic tree species (Eucalyptus species, Cupressus lustannica and Grevillea robusta) exist. The preliminary assessment showed that the three selected exotic tree species plantations/stands exist in 15 kebeles mostly on individual farmers’ land (Fig. 2.1).
From the total 15 kebeles (colored kebeles in Fig. 2.1), in which the plantations/stands of Eucalyptus species, Cupressus lustannica and Grevillea robusta exist, three sites/kebeles were purposively selected based on agroecology, availability of natural forests adjacent to each stand, availability of the plantations/stands more or less around the same site, accessibility of road, etc (Fig. 2.1). Accordingly, the three sites/kebeles were selected from three agroecological zones; one site/kebele from each agroecological zone. The three sites/kebeles selected were Yachiaurache, Aomafantule and Aomobeko from Kola/lowland, Woina Dega/midland and Dega/highland agroecological zones respectively. From each site/kebele, three plantations/stands (one plantation/stand for each selected exotic tree) and adjacent natural forest stand/patch were identified. Then, the total stands were 12; nine exotic tree plantations and three natural forest patches. The plantations/stands in each site/kebele were adjacent to one another.
Following a reconnaissance survey, actual sampling of vegetation was done focusing on appropriate sampling methods. Hence, systematic sampling design was employed. For each plantation/stand and forest patch, quadrats with 10mx10m were laid at 50m intervals using line transect survey method considering edge effect as described by Bullock (1996). Four square plots (quadrats of size 10mx10m) were laid in each stand. Within each main plot, five 1m x 1m subplots were laid to collect data regarding seedlings and saplings. A total of 48 plots (36 plots for exotic tree plantations and 12 plots for natural forest patches), 16 plots at each site/kebele were laid. GPS Garmin 60 was used to take site characteristics and quadrats points.
2.3 Method of Data Collection
From a total of 48 plots, all of the naturally regenerated woody species were identified, measured and counted. The indigenous woody plant species were categorized into three based on their height as seedlings (height ≤ 1m), saplings (height > 1m and ≤ 2m) and matured shrub/tree (height > 2m). The diameter at breast height (DBH) of all matured trees and shrubs (height > 2m and/or ≥ 2.5 cm DBH) within the main plot was measured using measuring tape and height was measured using centimetre marked stick, measuring tape and ruler with help of trigonometric calculation.
The height and collar diameter (diameter at the ground level) of seedlings and saplings within the subplot were measured using a centimetre marked stick and measuring tape, respectively. The identification of all indigenous woody plant species was carried out in the field with help of local elders who are experienced in knowing more plant names in local language. Then, the identified local names were crosschecked and matched using identification keys of Useful trees and Shrubs of Ethiopia: identification, propagation, and management for 17 agro climatic zones [12].
2.4 Method of Data Analysis
The data obtained from the three exotic tree species of nine plantations/stands and three adjacent natural forest patches about diversity and regeneration status of indigenous woody species were merged into four stands as Eucalyptus, Grevillea, Cupressus and Natural forest stands for analysis and interpretation.
The diversity, density, evenness, and richness of indigenous woody species were analysed using commonly used biodiversity measuring formulas and methods. Shannon diversity index (H′) and Shannon evenness index (J’) were used to measure diversity of naturally regenerated indigenous woody species in the different plantations and adjacent natural forests [13, 14]. These help to measure diversity of naturally regenerated indigenous woody species in the different plantations. And the similarity index of understory regenerated native woody species in the different stands was calculated using Jaccard’s Coefficient of Similarity (JCS) [13, 14].
The Species richness is measured to know how much number of different kinds of woody species presents in each stand; and from the results to compare between stands species composition. Species richness computed simply by counting the number of types of indigenous woody species (taxa) regenerated in a stand (S’). And the Margalef index, was used to compute the species richness for each stand. This index takes the total observed individuals ‘N’ as one factor in the computation. And the formula is as follows:
R=(S’-1)/lnN “Eq. 1”
Where, R is Margalef index of species richness; S’ is number of taxa or species; N is number of individuals.
Thus, the higher the R index value, the richer is the stand in species.
The diversity of species is computed to know how diverse (in item and abundance) the species present in each stand. Or to know how much is the average degree of uncertainty in predicting to what species an individual chosen at random is from the population of a collection of species. Species diversity is analyzed by using the most popular of metrics Shannon-Wiener Diversity Index (H’) and Simpson’s Index, D [13, 14, 15].
Shannon diversity index (H’) is a non-parametric index based on the proportional abundance of species taking into account both species richness and evenness [14]. It can be determined using the equation:
𝐻′ = − ∑ 𝑝𝑖 𝑙𝑛𝑝𝑖 “Eq.2”
Where 𝑝𝑖 represents the proportion of the population that belonging to the ith species
Evenness is a measure of how similar species are in their abundances. And as a heterogeneity measure, it describes the equitability of species abundance in the community [14]. Evenness is calculated using Shannon Evenness index (E’) [13]. And the formula of evenness index is described as follows:
\({E}^{{\prime }}= \frac{H{\prime }}{H{\prime }max}\) “Eq. 3”
Where, H’ is observed diversity (Shannon diversity index); H′max is equal to natural logarithm of richness, ln S.
Thus, an assemblage in which most species are equally abundant is one that has high evenness. And it is conventional to equate high diversity with high evenness (equivalent to low dominance of one or few species) [14].
The Density of species expresses how many members of that particular specie are found in the forest. Or it expresses the abundance of the specie in that forest. And it is computed as the division of the number of plants of a certain species by the area sampled expressed in hectare.
Basal area is the cross-sectional area of tree stems at breast height. It is measured through diameter, usually at breast height that is 1.3 m above ground level. It measures the relative dominance (the degree of coverage of a species as an expression of the space it occupies) of a species in a forest [16]. It is calculated as:
𝐵𝐴 = 𝜋 𝑥 (𝐷𝐵𝐻)2 “Eq.4”
Where, BA = basal area (m2), DBH = diameter at breast height (cm); π = 3.14
Dominance refers to the degree of coverage of a species as an expression of the space it occupied in a given area. Usually, dominance is expressed in terms of basal area of the species [15]. Two set of dominance: dominance (the sum of basal areas of the individuals in m2/ha), and relative dominance, which is the percentage of the total basal area of a given species out of the total measured stem basal areas of all species.
\(Dominance= \frac{Total basal area}{Area sampled}\) “Eq. 5”
\(Relative dominance= \frac{Dominance of species A}{Total dominance of all species}x100\) “Eq. 6”
Frequency is defined as the probability or chance of finding a species in a given sample area or quadrant [18]. Thus, it shows the presence or absence of a given species within each sample plot. Frequency will be computed for each woody species encountered within the study plots:
\(Frequency of species = \frac{Number of plots in which that species occurs}{Total number of plots}\) “Eq. 7”
\(Relative frequency= \frac{Frequency of species A}{Total frequency of all species} x 100\) “Eq. 8”
The Similarity analysis is used to identify the highly similar and or dissimilar stands in their understory plant composition. The similarity is analysed using a statistical measure of similarity Jaccard’s Similarity Coefficient (JSC) [13, 14]. The Jaccard index for two sets, set A and set B, is defined as the cardinality of their intersection divided by the cardinality of their union. Mathematically, this can be is described as follows:
J(A,B)=[AnB]/[AUB] “Eq. 9”
Where, A and B are assumed as stand 1 and stand 2.
Or,
Cj=j/(a + b-j) “Eq. 10”
Where, j = the number of species found in both sites and a = the number of species in stand 1 with b the number of species in stand 2.
The status of natural regeneration of indigenous woody species in a stand or in a forest as a whole can be determined from population structures. Population structure refers to the distribution of individuals in arbitrarily defined by diameter or height classes. In this study, the height class category of regeneration was followed based on seedlings, saplings and mature trees [18], as shown in the following manners:
- Seedling > sapling > tree/shrub state, pattern represents good regeneration
- Seedling outnumbers sapling and tree/shrub state but sapling less than tree/shrub state pattern fair regeneration
- Seedling < sapling < tree/shrub state, this pattern shows poor reproduction and hampered regeneration
- With no individual in seedling and sapling stages but relatively many individuals in matured tree/shrub stage pattern shows poor reproduction and hampered regeneration
Then the values evaluated by using height class histograms, (Height class vs. % proportion of density). And from the population structure one of the two types of regeneration status will be determined: good regeneration status or poor or hampered regeneration status in determining regenerating ability of both a stand and a species. Those Species which possess high number of individuals in the lower height category, particularly in the first two categories, are considered to have good regeneration potential; whereas, other species which possess either no or few number of individuals are in poor regeneration status. From this analysis, the species which show good regeneration or hampered regeneration status in each stand or for the forest as a whole were described.
A computer program, Microsoft Excel, and R Statistical software were employed for data analysis. To make computation and analysis for diversity, evenness, abundance and status of growth and to draw graphs, curves and tables, a computer program, Microsoft Excel was employed while R Statistical software was employed for all statistical analyses.
Density and species richness as well as co-variables affecting density and species richness of IWS were analysed using a generalized linear model (GLM) specifying negative binomial distribution with log link function (using the R MASS package). Chi-square and correlation tests were used to test the significance of distribution of density and species richness of different stands among agro-ecologies/altitudinal ranges and to test the interdependence or relationship between density and species richness of IWS and other co-variables respectively.
For basal area (BA), Shannon diversity index, Richness index and evenness, one-way ANOVA (with log transformation to account for the large variation in the standard deviations) was used. Significant results (alpha ≤ 0.05) were compared using the glht function of the R multcomp package for GLM and ANOVA models.