Intraspecific Morphological Variations among the Populations of Milicia excelsa , Pouteria adolfi-friedericii, and Prunus africana in Different Natural Forests of Southwest Ethiopia

Plants have the ability to change their morphological and physiological traits in response to environmental variations. Te objective of this study was to determine the intraspecifc morphological variations among the populations of M. excelsa , P. adolf-friedericii , and P. africana in southwest Ethiopia. Representative forests were systematically selected, and a total of ten transects of 160 m length were randomly laid at 100m intervals, and 30 quadrats (20m by 20 m) were laid along each transect line at 50m intervals. Stem height, DBH, and bole length of trees for each species were measured in each quadrat. Te intraspecifc morphological variations among populations of each species were computed using hierarchical clustering and principal component analysis (PCA) with R.4.1.3. A total of 55 trees for M. excelsa in four forests, 232 trees for P. adolf-friedericii in eight forests, and 184 trees for P. africana in fve forests were measured. Accordingly, three, fve, and three population clusters were identifed for M. excelsa , P. adolf-friedericii, and P. africana , respectively. Te analysis of similarity (ANOSIM) indicated the presence of considerable dissimilarity among population clusters for M. excelsa and P. africana but was not signifcant at p ≤ 0 . 05 ( R =0.9, p � 0 . 17). However, ANOSIM indicated the presence of considerable dissimilarity among population clusters of P. adolf-freidericii, which was signifcant at p ≤ 0 . 05 ( R =0.9, p � 0 . 03). Overall, there was a visible morphological variability among the populations of M. excelsa , P. adolf-friedericii, and P. africana each at the diferent sites. Terefore, it is important to look for conservation strategies, such as domestication, to maintain and improve the variability and genetic quality among the populations in a wider scale of the ecological and social environment.


Introduction
Presently, habitat destruction and fragmentation are the major causes of species extinction [1]. Te increasing human populations and attendant land-use intensifcation (e.g., cultivation, grazing, and urban development) resulted in the loss and subdivision of native habitats, increasing species extinction rates, and lowered species diversity within managed ecosystems. Habitat loss has caused a proportional loss of individuals from the landscape resulting from changes in confguration of habitat such as reduction of habitat patch size and isolation of patches [2,3].
Natural forests in southwestern Ethiopia are the major sources of livelihood (timber and nontimber forest products) for the communities in the surrounding area [4]. Tese forests are also the potential sources of economically important timber species such as Prunus africana, Milicia excelsa, and Pouteria adolf-friedericii. However, in recent years, the conversion of the forest ecosystem into agriculture lands for cofee and tea plantations is increasing. As a result, species such as Milicia excelsa, Pouteria adolf-friedericii, and Prunus africana are under high pressure due to random cutting and deforestation [5,6].
Plants have the ability to change their morphological and physiological traits in response to environmental variations [7][8][9]. As a result, plants tend to adjust the expression of traits such as morphology, growth, structure, function, and metabolism to persevere their adaptability in diverse environments [9,10]. Hence, plant populations respond to variable environments by becoming phenotypically plastic and genetically variable [11], provided that phenotypic plasticity itself could be under genetic control and subjected to selective pressure [12]. Te adaptation of a species to these variations may produce diferent morphological and physiological characteristics, resulting in the development of ecotypes.
Milicia excelsa (Welw.) C. Berg is a deciduous tree with a height ranging from 30-50 m and a straight clear bole. Te seeds can be stored in airtight containers in a cool dry place for a period of up to 2 years with no signifcant loss of viability [13,14]. Prunus africana (Hook.f.) Kalkm is an evergreen tree, 10-24 (36 max.) m in height, with a stem diameter of 1 m; bark blackish-brown and rugged; branchlets dotted with breathing spots, brown and corky; twigs that are knobbly [14]. Pouteria adolf-friedericii (Engl.) Baehni is a very tall tree, to 50 m, with a clear straight bole to about 16 m, topped by a rather small dense crown, mature trees buttressed at the base [13,14]. All the three tree species are socioeconomically important ones and are used for timber, frewood, medicine, etc. [13,14]. Terefore, the aim of this study was to determine the intramorphological variation (stem height, diameter the breast height (DBH), and bole length) of individual trees and populations of M. excelsa, P. adolf-friedericii, and P. africana in southwest Ethiopia.

Te Study Area.
Te study area includes four zones in southwest Ethiopia: Ilu Ababora, Shaka, Bench Maji, and Kafa ( Figure 1). Te descriptions of the study area given below are based on WorldClim version 2.1 climate data [15]. Te mapping and extraction of bioclimatic data for the study area was made using ArcMap10.8. Te study area covers about 61,591 km 2 . Te area is located between 5.325-8.967°N and 34.183-36.825°E. Te elevation ranges from 407 m to 3308 m above sea level. Te mean annual temperature of the study area is 21.1°C, while the minimum annual temperature is 10.7°C and the maximum annual temperature is 28.4°C. Te mean annual precipitation is 1624.7 mm, while the minimum annual precipitation is 1096 mm and the maximum annual precipitation is 1968 mm.

Population Sampling and Measuring Morphometric
Parameters. A systematic random quadrat sampling technique was employed, i.e., representative forest habitat areas were identifed and systematically selected for sampling purposes ( Figure 2). Te data were collected between June and August 2019. A total of 10 parallel transect lines (160 m) were laid down across a representative forest habitat area, and 30 quadrats (20 m by 20 m) were laid randomly at 50 m intervals along each transect line (modifed from [16][17][18]). Te delimitation of the quadrant numbers to 30 was to satisfy the minimum empirical assumption to run statistical analysis. Ten, matured tree population, with DBH ≥2.5 cm and height ≥1.5 m [19], was counted, and three important morphometric parameters (i.e., stem height in meters, diameter at the breast height (DBH) in centimeters, and bole length in meters) were measured. Hence, for trees with DBH ≥2.5 cm and height ≥1.5 m [19], stem height, DBH (at 1.3 m above the ground), and bole length were measured. Moreover, the altitude of each sampling site was recorded. Te heights were measured using a clinometer and the DBH was measured using a standard caliper. Te altitudes of the sampling sites were recorded using a Garmin eTrex 20 GPS device.

Data
Analysis. Te morphological dissimilarities (or diversity) among the diferent individual trees and populations of the P. africana, M. excelsa, and P. adolf-friedericii were analyzed using a multivariate data procedure using distance-based agglomerative hierarchical clustering using R.4.1.3. Agglomerative hierarchical clustering is a "bottomup" approach where each observation starts in its own cluster, and pairs of clusters are merged as one moves up the hierarchy. Moreover, principal component analysis (PCA) ordination was computed using R.4.1.3 based on a correlation matrix to determine the most important morphometric parameters contributing for the variations among the populations of each species among the diferent forest sites. Data analyses were based on untransformed measurement data, and clustering was done based on the mean value of each morphometric parameter. As stated by Chahal et al. [20], morphometric parameters with high component values on axis 1 (close to 1 or −1) have a high contribution to clustering. Overall, PC1 and PC2 explain greater than 90% of the variance. Te greater the value of the main component on the frst axis (PC1), the greater the efect of this parameter in clustering. Te length of the vectors represents the magnitude of the representation of each variable for each component and the angles between the variables indicate the correlation between them.

Results
In this study, a total of 13 forest areas were surveyed for this study (Table 1). M. excelsa was recorded in four forest sites, P. adolf-friedericii in eight forest sites, and P. africana in fve forest sites. Hence, a total of 55 tree stems for M. excelsa, 232 tree stems for P. adolf-friedericii, and 184 tree stems for P. africana were recorded and measured.
In this study, a total of 13 forest areas were surveyed of which M. excelsa was recorded in four forest sites, P. adolffriedericii in eight forest sites, and P. africana in fve forest sites. Accordingly, a total of 55 trees for M. excelsa, 232 trees for P. adolf-friedericii, and 184 trees for P. africana were recorded and measured. Accordingly, three, fve, and three population clusters were identifed for M. excelsa, P. adolffriedericii, and P. africana, respectively (  clusters of P. adolf-friedericii (R � 0.9, p � 0.03) which was signifcant at p ≤ 0.05. R value closer to 1 normally indicated that the presence of high dissimilarity among cluster groups, while when closer to 0 implied an even distribution of high and low ranks (or ranges) within and between groups. Analysis of the principal component analysis (PCA) to determine the efect of morphometric parameters on morphological variations among the populations of M. excelsa indicated that DBH has a higher coefcient value (−0.99) in PCA1, and bole length and stem height have coefcient values of 0.80 and 0.58 in PCA2, respectively ( Table 2). An ANOVA test of the signifcance of the efect of parameters showed all morphometric parameters, i.e., DBH (F � 156.6, p ≤ 2e − 16), bole length (F � 13.34, p � 1.82e − 07) and stem height (F � 19.27, p � 1.2e − 09) have a highly signifcant efect on the variation of the populations of M. excelsa at p ≤ 0.001 * * * .
On the other hand, PCA to determine the efect of morphometric parameters on morphological variations among the populations of P. adolf-friedericii revealed that stem height and bole length have higher coefcient values of 0.73 and 0.67 in PCA1, respectively, and DBH has a coefcient value of 0.88 in PCA2 (Table 2). However, an ANOVA test of the signifcance of the efect of parameters showed that bole length (F � 109.2, p � 7.5e − 05) has a highly signifcant efect on the variation of the populations of P. adolf-friedericii at p ≤ 0.001 * * * . Moreover, PCA to determine the efect of morphometric parameters on morphological variations among the populations of P. africana indicated that bole length had a higher coeffcient value of −0.99 in PCA1, and stem length with 0.79 coefcient value in PCA2 (Table 2). However, an ANOVA test of the signifcance of the efect of parameters showed that bole length (F � 118.7, p � 0.008) has a very signifcant efect on the variation of the populations of P. africana at p ≤ 0.01 * * .
Moreover, analysis of the bivariate correlations between the morphometric parameters indicated that the stem height and bole length were positively correlated, while DBH was negatively correlated with both the stem height and DBH for both the populations of M. excelsa and P. africana. On the contrary, DBH was positively correlated with stem height while negatively correlated with bole length in the population of P. adolf-friedericii ( Figure 6).

Discussion
Tis pilot survey analyzed the populations of M. excelsa, P. adolf-friedericii, and P. africana in 13 natural forests in southwest Ethiopia. A total of 55 trees of M. excelsa from four forest sites, 232 trees of P. adolf-friedericii from eight forest sites, and 184 trees of P. africana from fve forest sites were randomly sampled and measured for their stem height, DBH, and bole length ( Table 1). Analysis of the AHC of the International Journal of Forestry Research population of M. excelsa showed three possible population clusters ( Figure 3). Te analysis of similarity (ANOSIM) indicated the presence of considerable dissimilarity among population clusters for M. excelsa, but was not signifcant at p ≤ 0.05 (R � 0.9, p � 0.17). Te R value of 1 showed that more of the populations within clusters were similar to each other and dissimilar to the populations in other clusters [21]. A study by Ouinsavi and Sokpon [22] also identifed four distinct population clusters of M. excelsa across the diferent bio-geographical zones in Benin based on eight morphometric parameters. Nevertheless, the type and the number of parameters included in a study determine the number of possible population clusters in a given biogeographical area. Moreover, the number of populations of a given species depends on the heterogeneity of biogeography where the scope of the study is focused. On the other hand, ANOVA tests of signifcance for the relative contributions of morphometric parameters in clustering the population of M. excelsa showed that DBH, stem height, and bole length was all important with highly signifcant efects at p ≤ 0.001 * * * . In other study, Ouinsavi and Sokpon [22] reported that stem height was found to show signifcant morphological variation in the Milicia excelsa populations across the diferent biogeographical zones in Benin. In this study, analysis of the patterns of morphological variation among the populations of M. excelsa along the altitudinal gradient indicated that stem height and DBH tends to show contrasting patterns of dynamics along the altitudinal gradient (Figure 7). In this study, larger sizes of DBH and shorter stem heights were recorded between 1050 m and 1210 m above sea level. Moreover, shorter stem heights were also recorded above 1310 m above sea level. In contrast, longer stem heights were observed in the midaltitude between 1210-1290 m above sea level. Overall, both parameters have appeared to have nearly a contrasting distribution along the altitudinal gradient. Overall, the stem height smoothly increased with increasing altitude and later decreased with increasing altitude making a unimodal normal distribution. However, Christelle et al. [23] also reported that tree height decreased signifcantly with increasing altitude in their study focusing on tree species diversity on a small montane of Atlantic Central Africa. Furthermore, our results also indicated that DBH tends to decrease with increasing altitude. In the contrary, Pokhrel, and Sherpa [24] reported that DBH showed a small increase with increasing elevation in their study of the tree species diversity on the Chitwan-Annapurna landscape in Nepal. Terefore, our study revealed that DBH and stem height were negatively correlated in the populations of M. excelsa ( Figure 6). In the contrary, Buba [25] reported that DBH was positively correlated with the tree height while Sumida et al. [26] reported that DBH was negatively correlated with the stem height. Tese fndings suggested that the variability of correlations between DBH and stem height was likely species and habitat-specifc. Overall, many studies have indicated that morphological variation is apparently the result of an adaptive response to the environment. For instance, variations in growth and phonological traits are associated with altitudinal ranges [27,28] or with contrasting climatic conditions [29].
Similarly, AHC of the population of P. adolf-friedericii showed that fve population clusters were possible (Figure 4). However, ANOSIM indicated the presence of considerable dissimilarity among population clusters of P. adolf-friedericii, which was signifcant at p ≤ 0.05 (R � 0.9, p � 0.03). On the other hand, an ANOVA test for the relative contributions of parameters in clustering the population of P. adolf-friedericii showed that the efect of bole length is highly signifcant on the clustering of populations at p ≤ 0.001 * * * . Moreover, the analysis of the patterns of morphological variation among the populations of P. adolf-friedericii along an altitudinal gradient indicated that the stem height and DBH appeared to show similar patterns of dynamics along altitudinal gradient with bimodal distributions (Figure 8). Here, the maximum measurement scores in both parameters were recorded around an altitude of 1250 m above sea level, while the minimum measurement scores were observed around 1120 m above sea level. A study by Christelle et al. [23] on tree species diversity on a small montane of Atlantic Central Africa showed that the tree height decreased signifcantly with increasing altitude. However, in this study, International Journal of Forestry Research the relationship between morphological parameter and altitude is not linear but rather observed to bimodal. Generally, our study showed that DBH and stem height were positively correlated in the population of P. adolffriedericii ( Figure 6). Similarly, Buba [25] also reported that DBH was positively correlated with the tree height, while Sumida et al. [26] reported that DBH was negatively correlated with the stem height.
Furthermore, AHC of the population of P. africana showed three possible population clusters ( Figure 5). ANOSIM indicated the presence of considerable dissimilarity among population clusters for P. africana, but was not signifcant at p ≤ 0.05 (R � 0.9, p � 0.17). A study by Kadu et al. [30] on the multivariate relationships of 25 populations of P. africana in the African highlands using nuclear DNA has also identifed three distinct (distant) clusters and two overlapping clusters. On the other hand, the analysis of the relative contributions of morphometric parameters in clustering the population of P. africana showed that bole length and stem height are important parameters. However, an ANOVA test of the signifcance of the efect of morphometric parameters showed that bole length has a very signifcant efect on the clustering of populations at p ≤ 0.01 * * . Te patterns of morphological variation among the populations of P. africana along the      International Journal of Forestry Research altitudinal gradient showed that stem height and DBH showed contrasting patterns of dynamics along the altitudinal gradient (Figure 9). Generally, our study revealed that the DBH and stem height were negatively correlated in the populations of P. africana ( Figure 6). In the contrary, Buba [25] reported that DBH was positively correlated with tree height, while Sumida et al. [26] reported that DBH was negatively correlated with stem height. Similar to the other species in this study, the relationship between morphological parameters and altitude is not linear when it comes to P. africana as well. For stem height, the maximum measurement score was observed at around 2500 m above sea level.

Conclusions
Tis study analyzed the intraspecifc morphological variation among the diferent populations of M. excelsa, P. adolf-friederici, and P. africana in diferent natural forests of southwest Ethiopia. Overall, there was a visible morphological variability among the populations of M. excelsa, P. adolf-friederici, and P. africana each at the diferent forest sites. Te morphological variability among the populations of M. excelsa, P. adolf-friederici, and P. africana across the diferent forest locations has great implications for selection and breeding strategies in the future. Terefore, it is important to look for conservation strategies, such as domestication, to maintain and improve the variability and genetic quality among the populations in a wider scale of the ecological and social environment. Tese vital indigenous tree species are naturally confned in southwest natural forest in Ethiopia. Te species are also found as a remnant of isolated and scattered populations, which are still under severe pressure from human impacts. Hence, the development of management strategies to protect and conserve the remaining wild populations of this valuable species should be efected. In other words, conservation eforts should focus on maintaining large populations to counteract the potential negative efects of drift and promote the maintenance of genetic diversity in these populations.

Data Availability
Te data used to support the fndings of this study will be made available on reasonable request to the corresponding author.

Conflicts of Interest
Te authors declare that they have no conficts of interest.