The Effects of Landscape Change on Plant Diversity and Structure in the Bale Mountains National Park, Southeastern Ethiopia

Background: This study aimed to determine the effects of landscape change on oristic composition, diversity, and structure in the Bale mountains national park, which faces a critical challenge from anthropogenic factors. The park is one of the 34 International Biodiversity Hotspots that comprise a variety of life forms. The vegetation data were collected systematically from 96 sample plots laid along 24 line transects. Vegetation composition and landscape structural analysis were made using R software version 3.5.2 and FRAGSTATS version 4.2.1, respectively. Results: Patch number was strong and positively affected species richness (r = -0.90, p < 0.05) and Shannon diversity index (r = -0.96, p < 0.01), and basal area (r = -0.96, p < 0.001), whereas mean patch size was strong and positively inuenced species richness (r = 0.95, p < 0.05), diversity (r = 0.87, p < 0.05), and basal area (r = 0.82, p < 0.05). The overall species richness, Shannon diversity index, and Margalef index were signicantly higher in the edge habitat than the interior at p < 0.05. However, the mean basal area of woody species in the interior habitat (11.16 ± 1.82 m 2 ha -1 ) was signicantly higher than the edge (3.99 ± 0.54 m 2 ha -1 ) at p < 0.05. Conclusion: This study revealed that the park was a biologically diverse and ecologically signicant area that provides a variety of ecological and economic benets to the surrounding communities. Though its habitats are changing alarmingly and urgent restoration and conservation action need to be taken to reverse this situation.


Introduction
Landscapes all over the world are alarmingly changed and fragmented due to anthropogenic factors such as urbanization, agricultural expansion, forest re, and climate change (Lindenmayer and Fischer 2013;Mullu 2016). Most of the global changes responsible for the reduction of population and biodiversity are exacerbated by fragmentation (Gibson et al. 2013;Krosby et al. 2010). The primary causes of global biodiversity reduction are the destruction and degradation of natural ecosystems (Pereira et al. 2010). Predominantly, habitat loss and fragmentation are presently the main threats to terrestrial biodiversity (Rogan and Lacher 2018). Moreover, habitat fragmentation can affect the species interactions and community composition, as invasive or pest species and may substitute the original species pool, increase the transmission and prevalence of the disease in small fragments (Haddad et al. 2015). Moreover, the species richness and abundance usually decrease with reduced patch size (Fahrig 2003). As landscapes become more fragmented, patch diversity increases with subsequent increases of the edge, exotic, and generalist species and ultimately lead to the reduction of landscape quality as habitat for species (Saunders et al. 1991). Accordingly, species richness in interior habitat, as well as indigenous and specialist species, tends to decrease (Bierregaard et al. 1992). The number of species existing in a patch tends to rise with patch size up to a certain limit, and the types of species found also tend to vary in size (Fahrig 2003). Size and shape interact to in uence the amount of interior area remaining in a particular habitat fragment (Mullu 2016).
Tropical montane ecosystems are one of the hot spot ecosystems on earth that comprises more than 200,000 species of owering plants (Aliyi et al. 2015;Vergara-Rodríguez et al. 2017). The Ethiopian highland, which is located in the tropical region, encompasses over 50% of the Afromontane vegetation in African (Ahmedin and Elias 2020). A suitable geographical position, a wide range of altitude, a high amount of rainfall, and a wide range of temperature variations equip the area with huge ecological diversity and a wealth of biological resources (Yimer 2007). However, severe deforestation coupled with the cultivation of steep marginal lands, overgrazing, and socio-political uncertainty, has resulted in rigorous land degradation over large areas of the country (WBISPP 2004). The overdependence of the country's economy on agricultural production and the existence of 80% of the population in the highlands mainly contribute to the degradation of ecological resources and biodiversity loss (Hurni 1998 Ethiopia and established in 1969 to preserve the endemic and indigenous oras and faunas in the area (Mekonnen et al. 2010;Stephens et al. 2001). It is one of the 34 International Biodiversity Hotspots and meets the requirements for the World Heritage Site and Biosphere Reserve Listing (Tesfaye and Bires 2015). However, the park is facing a critical challenge from the illegal settlement and overgrazing and that leads to the change in its landscape structure and function. As a result, the habitats in the park are changing and the provision of ecological services from it is substantially reduced. Consequently, no research provides detailed information about the landscape structure and its potential impact on vegetation composition and structure in the park. Therefore, this research was aimed to analyze the potential impact of landscape change on oristic composition, diversity, and structure in the BMNP.
Particularly, a comparative analysis was made among the edge and interior habitats of the park.

Materials And Methods
Description of the study site BMNP is located within the geographic bounds of 6º53'08" N latitude and 39º44'03" E longitude and 400km southeast of Addis Ababa, Ethiopia (Fig. 1). It encompasses a broad range of habitats between 1,500m and 4,377m altitude. The park holds the largest area of Afro-alpine habitat (about 1,000 km 2 ) above 3000 m asl in Africa and the second largest stand of moist tropical forest (Williams-Linera 2002). It is one of the 34 International Biodiversity Hotspots and also quali es for World Heritage Site and Biosphere Reserve Listing (OARDB 2007). It received rainfall that ranged from 1000 to 2400 mm annually, and the distributional pattern is bimodal with heavy rains from July to October, (highest peak in August) and small rains from March to June (highest peak in April). The mean monthly minimum and maximum temperatures are 5.6 o c and 21.4 o c, respectively. Its soil is fertile silty loams of reddish-brown to black clay soils dominated by Vertic Cambisols and Leptosols (Miehe and Miehe 1994).

Vegetation sampling design
A total of 96 sample plots (20 x 20 m) was systematically laid along 24 line transects in eight directions along three altitudinal gradients at 100 m elevational differences as it maximizes the distance between plots and minimizes spatial correlation among the observations (Barry 2008). An equal number of sample plots were laid on the edge and interior habitats to make a comparison between their vegetation data following (Daye 2012) Floristic composition and structure The most commonly used diversity indices of Species richness (S), Simpson index (D), Shannon-Wiener index (H'), Pielou's evenness index (J'), Whittaker β-diversity (βw), Margalef index (D M ), and Berger-Parker index (d) was computed to analyze the patterns of plant diversity in the edge and interior habitats following (Magurran 2013;Økland 1990). The woody species density, frequency, dominance, and their relative values in the interior and edge habitats were computed to obtain the important value index and describe the woody species structure following (Ellenberg and Mueller-Dombois 1974;Martin 2010).
Moreover, DBH, tree height, and basal area were analyzed to determine the population structure following (Kitessa et al. 2007;Van der Maarel 1979).

Measurement of landscape structure
Landsat images of the years 1985, 1995, 2005, and 2017 were processed using ArcGIS version 10.3 to produce time-series datasets of land use/land cover. Then, eight landscape indices were analyzed using the processed land use/land cover data following (McGarigal et al. 2012;Smiraglia et al. 2015). The indices include patch number (PN), mean patch size (AREA_MN), total core area (TCA), edge density (ED), area-weighted mean shape index (SHAPE_AM), mean Euclidean nearest neighbor distance (ENN_MN), and interspersion and juxtaposition index (IJI). FRAGSTATS software version 4.2.1 was used to compute the landscape patterns in each land cover class and the entire landscape (McGarigal 2002). The two-way Analysis of Variance (two-way ANOVA) and linear regression analysis was made to test signi cant differences between fragmentation indices and species composition and structure parameters following the post hoc Tukey's Highly Signi cance Difference (Tukey's HSD) test at 5% signi cance level using PAST software version 4.02 (Hammer et al. 2001).

Results And Discussion
The extent of landscape change in the class and landscape level The analysis of landscape structure in this study revealed that the habitats in the BMNP are progressively transformed. The area has shown an increase in PN by 40.2% and a decrease in AREA_MN by 28.7% from 1985 to 2017. According to (Oertli et al. 2002) the high number of separated patches in a habitat indicates a high level of fragmentation. Across the entire study period SHAPE_AM, which indicates the complexity of patch shape, increased by 18.8%. A higher perimeter-area relationship characterizes the rapid rate of fragmentation in the landscape (Flowers et al. 2020). Moreover, there was inconsistency in the values of ED, however, it was increased by 22.3% over the study period. As it was emphasized by (McGarigal 2002), the oscillation of ED indicated a major reduction in the spatial heterogeneity of the landscape. Conversely, the study area has shown a declining trend in TCA by 10.6 % from 1985 to 2017. This was due to the escalated level of disturbances in the study area. As it was reported in the study by (Kidane et al. 2012), the most dominant practices in the Bale mountains, especially after 1995, were the upward expansion of agriculture and enrichment plantation.
The isolation of patches within the landscape of the study area was increased from 105.22 m to 111.94 m over time (Table 1). This result is in agreement with the result reported by (Tolessa, Senbeta, and Kidane 2017) in the central highlands of Ethiopia and (Daye 2012) in southwest Ethiopia. Conversely, the intermixing of patches in the study area shown an overall declining trend from 95.38 to 86.77 over the study period. This result showed that the BMNP constitutes more scattered patches compared to other similar areas studied by (Posada Posada 2012;Tolessa et al. 2017). Overall oristic composition and structure A total of 205 species that belongs to 153 genera and 71 families were identi ed in the BMNP. In terms of life form, 114 were woody species (50 trees, 52 shrubs, and 12 lianas) and 91 were herbaceous species. Asteraceae was the richest family with 31 species and the most species-rich genus was Helichrysum with 9 species. Twenty endemic species (about 9.7% of the total) including Alchemilla haumanii Rothm, Erythrina brucei Schweinf, and Knipho a isoetifolia Hochst, were identi ed. The overall Shannon diversity and evenness index of the study area were 4.34 and 0.81, respectively. This was higher relative to other similar areas such as Bonga forest (Senbeta et al. 2014), Agama forest (Dibaba et al. 2020), and Munessa forest (Ahmedin and Elias 2020). Conversely, the total density of seedlings, saplings, and mature trees in the BMNP were 8751, 4413, and 1567 individuals ha -1 , respectively. This was less than other similar areas such as Kuandisha forest (Berhanu et al. 2017) and Wof-Washa forest (Fisaha et al. 2013). The ratios of seedling to mature tree, sapling to mature trees, and seedling to sapling were 5.58, 2.82, and 1.98 respectively. This shows the recruitment potential of the forest is relatively higher (Ayalew 2018).
Floristic composition and structure in the edge and interior habitats A total of 136 species that belongs to 111 genera and 59 families was identi ed in the edge habitats of the sampled patches, whereas 117 species that belong to 84 genera and 40 families were recorded in the interior habitats. From the identi ed life forms 19 species were trees, 22 species were shrubs, 86 species were herbs, and 7 species were lianas in the edge habitats, whereas 28 species were trees, 21 species were shrubs, 57 species were herbs, and 11 species were lianas in the interior habitats. The overall means (± SE) of species richness (35 ± 4.2), Shannon diversity index (2.93 ± 0.17), and Margalef index (5.68 ± 0.69) of the edge habitat were signi cantly higher compared to the interior habitat at p < 0.05 (Table 2). These variations could be due to the differences in site productivity, habitat heterogeneity, and disturbance factors (Maestre 2004;Senbeta et al. 2014) or the invasion of exotic plant species (Ranney et al. 1981). However, the woody species richness in the interior habitat (28) was signi cantly higher than the edge (17). Moreover, the evenness index in the interior habitats (0.83 ± 0.04) was higher but not signi cant than the edge habitat (0.79 ± 0.05). This result was in agreement with the nding by (Kacholi 2014). Abiotic factors, seed predation, loss of pollinators and seed dispersers, and tree mortality were reported as the common causes for the differences in woody species composition between the edge and interior habitats (Kacholi 2014;Laurance et al. 1998). The computed Sorensen's similarity index depicted that the number of species in the edge habitats was 45% similar to the species in the interior habitats. This value indicated that the similarity between the edge and interior habitat was weak (Ahmedin and Elias 2020). The mean density of seedling (995.42 ± 19.27 individuals ha -1 ), sapling (509.29 ± 9.06 individuals ha -1 ), and mature trees (187.60 ± 4.70 individuals ha -1 ) in the interior habitat was signi cantly higher compared to the edge. This indicates that the recruitment potential of the interior forest was signi cantly higher compared to the edge habitat (Ayalew 2018). This could be due to the increase of seedling, sapling, and mature tree mortality rate in the edge (Benitez-Malvido 1998; Kacholi 2014). The mean woody species density in the interior habitat (85 ± 22.17 stems ha -1 ) was signi cantly higher compared with the edge habitat (70 ± 16.53 stems ha -1 ) at p < 0.05 (Fig. 2a). This could be due to the selective cutting of trees for timber production, house construction, and rewood in the edge habitats, which ultimately leads to a reduced density of large trees and greater canopy openness (Laurance 2004).
Moreover, the seedlings are most affected by edge effects due to their sensitivity to environmental changes and biotic interactions (Murcia 1995;Saunders et al. 1991). Conversely, the mean basal area of woody species in the interior habitat (11.16 ± 1.82 m 2 ha -1 ) was signi cantly higher than the edge habitat (3.99 ± 0.54 m 2 ha -1 ) at p < 0.05 (Fig. 2b). This was due to the signi cantly higher mean DBH (78.62 ± 4.56 cm, p < 0.001) and height (33.63 ± 2.71 m, p < 0.05) of woody species (Yineger et al. 2008) in the interior habitat than the edge. There were 27.32 % of larger diameter individual tree species with DBH > 100 cm recorded in the interior habitat, whereas 4.09 % of individuals with DBH > 100 cm were identi ed in the edge habitat. Microenvironmental conditions such as high temperature, low relative humidity, high wind force, low soil nutrient, and litter moisture in the edge habitats may contribute to the changes in tree abundance and distribution in the forest (Newmark 2001;Yan et al. 2007).
Juniperus procera was the dominant woody species in the edge habitat with an IVI of 32.49, whereas Croton macrostachyus was dominant in the interior habitat with an IVI of 40.61 (Table xx). Accordingly, Juniperus procera, Hagenia abyssinica, and Syzygium guineense were identi ed as generalists that abundantly occurred in both edge and interior habitats, whereas Hypericum revolutum Vahl was identi ed as marginalized species that characteristically dominated the edge habitats (Lomolino and Weiser 2001;Pimm et al. 2014). However, no woody species were found as a specialist that typically occur in the interior habitats (Table 3). The impacts of landscape change on oristic composition and structure The computed regression analysis among the landscape indices and species composition and structural parameters in this study revealed that only PN and AREA_MN signi cantly affected both the species composition and structural properties of the study area. Accordingly, PN was strong and negatively affected the overall species richness (r = -0.90, p < 0.05) and Shannon diversity index (r = -0.96, p < 0.01) (Table 4). Conversely, the overall species richness (r = 0.95, p < 0.05) and Shannon diversity (r = 0.87, p < 0.05) shown strong and positively correlated with AREA_MN. This implies that, as the number of fragmented habitats increases, species richness and diversity, particularly interior dependent species, decreases. However, edge dependent species comfortably ourished. One of the consequences of habitat fragmentation is an increase in the proportional abundance of the edge-in uenced habitat and its adverse impacts on interior sensitive species (Robbins et al. 1989). Undoubtedly, while some species (e.g. Habitat specialists) suffer from fragmentation, others benefit from it (e.g. Generalists and edge species) (Henle et al. 2004). Consequently, PN were strong and negatively correlated with AREA_MN (r = -0.71, p < 0.001). This implies that as the PN increases the area of fragments decreases as a result small fragments contain a smaller species richness and lower species density than large fragments (Laurance and Vasconcelos 2009). Large areas of habitat tend to support more individuals, and hence, more species (Rosenzweig 1995). Besides modifying the spatial pattern of the landscape habitat size reduction and increase of isolation cause an alteration of the dispersal rate, affecting survival, and mortality of individuals (Fahrig 2003). Many population and community changes in habitat fragments were commonly attributed to edge effects (Laurance and Vasconcelos 2009). Interior species may be affected by the size decrease of their habitat, by edge effect, and by competition with generalists (Bolger et al. 2001;Schonewald-Cox and Buechner 1992). The most threatened endemic species due to edge effect in the BMNP were Helichrysum harennense Mes n, Knipho a insignis Rendle, Rubus erlangeri Engl., and Vepris dainellii Pichi Serm Kokwaro. Also, the most common invasive species in the study area favored by edge effect was Achyranthes aspera L, which is also common in the disturbed forests and forest edges of the dry Afromontane forests and moist Afromontane forests in Ethiopia (Friis et al. 2010). The gradual decline of the more sensitive species may induce a species turnover in fragments and cascade effects (Lomolino and Weiser 2001;Pimm et al. 2014).
Besides, among the landscape indices computed only PN and AREA_MN signi cantly affected some of the oristic structural properties assessed. Thus, the PN was strong and negatively affected the woody species density (r = -0.84, p < 0.05) and basal area (r = -0.96, p < 0.01) as well as AREA_MN was strong and positively affected the density (r = 0.71, p < 0.05) and basal area (r = 0.82, p < 0.05) of woody species. Habitat destruction, isolation, and transformation affect the structure and dynamics of populations, communities, and ecosystems, as well as ecological processes (Soulé and Orians 2001).
Generally, as AREA_MN and COA of patches increases, species richness, diversity, evenness, woody species density, basal area, DBH, and height also increase. Whereas, as PN, SHAPE_MN, ED, ENN_MN, and IJI of patches increases, oristic composition, and structural variables decreases. This implies that the landscape composition and con guration change may potentially affect the vegetation composition and structure of a particular area.

Conclusions
This study revealed that BMNP was one of the richest ecological areas that comprise of endemic and indigenous plant species in Ethiopia. However, anthropogenic factors, such as settlement expansion, overgrazing, and recurrent res, are strongly impaired the plant species diversity and structure as well as the overall ecological integrity of the landscape. These activities are responsible for the occurrence of habitat loss and fragmentation in the park. As a result, the species richness and the Shannon diversity index were signi cantly lower in the interior habitat compared with the edge habitat. This was due to the dominance of generalist species in the edge habitat, which can easily ourish in the limited resources, and specialists in the interior habitat, which are sensitive to the limited resources and restricted on a certain environmental condition. Moreover, the species in the edge habitat were unevenly distributed and the basal areas of woody species were relatively lower than the interior habitat. This was due to the presence of larger size trees in the interior habitat and high tree mortality in the edge habitat. Therefore, human activities in the park should be stopped, and settlements in the park should be relocated to other places to avoid their potential impacts on the oras and faunas of the park. Moreover, studies on microenvironmental factors such as light availability, air and soil temperature, humidity, and soil nutrients along the edge and interior gradient should be conducted to determine their effect on species richness, composition, and structure.

Declarations
Ethics approval and consent to participate Not applicable.

Consent for publication
Not applicable.

Availability of data and materials
The data used in this paper can be provided upon request.

Competing interests
The authors declare that they have no competing interests.

Funding
This research was funded by the Ethiopian ministry of science and higher education.
Authors' contributions AM conceived, designed, collected the data, analyzed, and wrote the manuscript. EE supervised the inception, design, and edited the manuscript. All authors read and approved the nal manuscript.  (2): 103-20. Figure 1 Location map of the study area. Note: The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors.