Forest Landscape Restoration and Ecosystem Services in A Luoi District, Thua Thien Hue Province, Vietnam

Abstract The Government of Vietnam has invested efforts to increase the forest cover, and to conserve biodiversity through different forest development projects and programs. Losing natural forests and landscapes in the context of the “exhaust” of ecosystem services has been seen as burden in many mountainous areas. The Decision No.16 on ecosystem restoration, which was adopted by the Conference of the Parties to the Convention on Biological Diversity (CBD) at the 11th meeting (December 5th, 2012) stated that ecosystem restoration requires the application of suitable technologies and the fully-effective participation of local entities. This serves to identify obstacles while attempting to restore, regenerate ecosystem services and biodiversity, which have been degraded and lost in the recent decades. Furthermore, Vietnam’s National Forest Development Strategy targeted to achieve a forest area of 16.2 million hectares by the year 2020. Local people living adjacent to forests depend on the forest ecosystem services supplied from various natural forest landscapes in the area. This holds true especially for the people of Central Vietnam where the terrestrial area is narrow due to the country shape. In this area, agriculture practices play an essential role although the agricultural land is very limited due to the topographic conditions. The distinct land-uses reflect the natural distribution of plant and animal species as well as human interventions. In Vietnam, the forest ecosystems have been classified into three categories according to their main functions: special-use forest for nature conservation; protection forest for the watershed and protective measures; and production forest for commercial operations. This study was conducted in the A Luoi District, Thua Thien Hue Province. Ground truth samples were inventoried in three forest types from 150 m to 1162 m above sea level (a.s.l.) and steep slopes from 5 to 48 degrees. The elevation range was divided into the lower elevation level H1 ranging from 150 m – 699 m and into the higher elevation level H2 from 700 m-1162 m a.s.l.. The slopes were stratified into level S1 from 5-20 degrees, and into S2 from 21-48 degrees. The forest cover was classified into the types: undisturbed forest (UF), low disturbed forest (LF), and heavily disturbed forest (DF). To strengthen the classification of forest types, a t-test of extracted vegetation indices between ground truth plots and training sample plots was done. Up to date, no remote sensing-based work on ecological stratification of the natural forest landscapes has been conducted. Finding the tree species distribution, species diversity, and species composition over the sub-stratification of the elevations, slopes, and the forest types - by applying remote sensing - are necessary to classify the land-use types and to map out the availability of natural resources, especially the ecosystem services supply and demand of local people. Land-use and forest type classification may contribute remarkably to long-term planning, which has been assigned to local authorities, and which should include local communities. The entire study consists of four main parts. The first part aimed at evaluating the influence of topography on tree species diversity, distribution, and composition of the forests in Central Vietnam. A significant difference of species richness and species diversity was found in shallower and steeper slopes (p < 0.05) and a relatively high correlation of the species distribution, the number of stems, and the number of tree families with the elevation factor was found. The lower elevation and shallower slope showed higher species richness (p < 0.05) but not a significant difference between the number of families and the evenness. The dominance and the abundance of tree species among the topographic attributes were significantly different (p < 0.05). Lower elevation and shallower slope showed higher species richness and species diversity than the higher elevation and steeper slope. The most dominant and abundant tree families from different elevations and slopes included the Myrtaceae, Dipterocarpaceae, Burseraceae, Fagaceae, Moraceae, Cornaceae, Apocynaceae, Sapindaceae, Cannabaceae, Juglandaceae, Lauraceae, Myristicaeae, Annonaceae, Ebenaceae, Meliaceae, Rubiaceae, and the Rosaceae. The second part aimed at assessing the soil qualities, which belong to the most essential elements for land-use planning and agricultural production. 155 soil samples from different land-use types and topographic aspects were collected in order to compare information on soil organic carbon (SOC), soil total nitrogen (STN), and soil acidity (pH) at two soil depths. The SOC of arable land and forest plantation land was found to be higher than those of grassland and of natural forests (p < 0.05). The total nitrogen in the natural forests was significantly less, compared to the other land-use types. No significant differences in the total nitrogen content (p < 0.05) were found among arable land, plantation forest, and grassland. The soil organic carbon and the total nitrogen were high in the upper soil and less downwards, within all land-use types. The soil pH in the plantation forest and the arable land-use types showed no significant change among soil depth categories. Significant differences were not found in topographic aspects and the soil organic carbon content; however, differing trends of soil organic carbon and land-use types and aspects were found. The impact of the slope, elevation, farming system and soil texture accounted for the main differences of soil indicators under varying land-use types in the A Luoi District. The third part of this study was designed to apply remote sensing data from Landsat-8 and Sentinel-2 sources in order to classify land-cover and land-use classes (including three forest types UF, LF, and DF) in the study area by using machine learning algorithms. Further, vegetation indices were applied to find possible correlations and regressions of both, vertical and horizontal structures of the dominant forest tree species within different forest types. It was found that the vegetation indices between the ground-truth plots and the training sample plots were significantly different (p