| Forest fragmentation,defined as the process that original large and continuous forest patches are gradually converted into smaller and isolated forest parcels,which has currently been one of the major environmental problems that are of great concern to the world today.Forest fragmentation generally increases the mortality of trees,destroys the biogeographical surroundings that species need to survive and thrive,weakens the ability of ecosystems to resist external disturbances and adapt to environmental changes,and even endangers human security and socioeconomic progress.As the high biodiversity distribution centers in the world,the forest ecosystem is an important source of renewable energies.However,global climate change and accelerated social processes have led to dramatic changes to land use in forest areas in the past few decades.Human over-exploitation of natural resources,reclamation and urban expansion have greatly changed the spatial distribution pattern and composition structure of forests,or forest fragmentation patterns.To this end,how to accurately monitor the spatiotemporal dynamics of forest fragmentation,and develop a holistic forest resource regulation model by taking the impact mechanism of forest fragmentation into account has become a major concern that coordinates or balances forest landscape security and urban economic development.Based on MSPA and geographic detector models,the forest fragmentation characteristics and corresponding attribution analysis in Xinyu City,Jiangxi Province were done in this work.Specifically,four Landsat remote sensing images acquired in vegetation growing season in 1999,2006,2013 and 2020 were classified by using support vector machine classification algorithm to construct a high accuracy land cover dataset.Then,seven forest fragmentation components,including core,perforation,island,bridge,edge,branch and ring were classified by using MSPA model under a one-pixel boundary width window to establish the spatio-temporal pattern maps of forest fragmentation in the study area.After revealing the temporal and spatial evolution characteristics of forest fragmentation,the influencing mechanisms of fragmentation spatial differentiation were systematically excavated from the two aspects of natural and human interferences.Meanwhile,by considering the regional development characteristics and levels,the site-specific forest resources regulation strategies were formulated in an attempt to resolve the problems of accelerating fragmentation,and to achieve the overall objective of optimizing urban spatial layout,and coordinating ecology and economy.The main results were as follows:(1)Based on the optimized feature space,SVM classifier was used to map five typical land cover types including forests,farmlands,buildings,rivers and barelands in Xinyu City.According to an independent validation,the overall classification accuracies of land cover products in the four phases were greater than 89%,and the corresponding Kappa coefficients ranged from 0.86 to 0.91.In the east of the study area,forest shrinkage was observed due to the expansion of cropland and built-up,dropping from 1622.05 km~2in 1999 to 1372.05 km~2 in 2020.In the central region,the bareland and built-up expanded rapidly.The increased bareland,from 87.88 km~2 in 1999 to 122.70km~2 in 2020,was widely distributed in the whole city.The built-up increased from 210.27 km~2 in1999 to 376.52 km~2 in 2020,which showed a trend of agglomeration and radiated continuously.In the northwest,cropland expansion fluctuated from 1104.17 km~2 in 1999 to 1194.16 km~2 in 2020.(2)The outcome of MSPA showed that the core forest component dominated the seven fragmentation components,with an area share of 73.79%,edge ranked the second(14.77%),perforation(3.42%),branches(3.27%),rings(1.73%),islands(1.67%)and bridges(1.46%).The concentrated and contiguous large core forest patches were mainly distributed in the southwest and northwest regions,while some small core patches were located in the eastern and central regions.In the large core patches,some perforations were found.In the small and medium core patches,some connecting bridges and rings were observed.The islands were mainly distributed in the central and eastern parts with a small forest area.When examining the evolution dynamics in forest fragmentation over Xinyu City from 1999 to 2020,a gradually deteriorating forest fragmentation was witnessed,and this trend was proven by a decreased forest integrity and connectivity.The core area shrank from 73.8%in 1999 to 69.80%in 2020.The connecting bridges and rings also decreased.The edge effect increased,and the number of isolated islands increased.(3)The trend surface analysis and sensitivity analysis method were used to quantify the forest fragmentation intensity within a 3 km×3 km grid,and to characterize its spatial differentiation characteristics of forest fragmentation.In north-south direction,forest fragmentation decreased first then giving way to an increase,with the highest point occurring in the middle.In east-west direction,forest fragmentation continued to fall,showing a high east and low west trend.In both directions,there was a non-equilibrium transition in forest fragmentation.The prominent spatial differentiation of forest fragmentation was lower in the central and eastern regions,higher in the north,south and west regions.(4)The spatial differentiation mechanism of forest fragmentation was analyzed by using the geographic detector model.Both natural and human factors accelerated the fragmentation of forests.The factor’s contributions were identified as follows:average annual rainfall at 0.40,distance from rivers at 0.41,distances from roads at 0.63,distance from consumption centers at 0.40,land use diversity at 0.62),ratio of farmland area at0.59,and ratio of buildings area at 0.42.These factors were highly correlated with the spatial distribution pattern of forest fragmentation.In terms of indication function,the distance from the consumption center and the land use diversity factors showed a linear relationship with forest fragmentation intensity,while other factors showed high intensity values in certain specific intervals.In the comparison between single factor and double factors responsible for forest fragmentation differentiation,the combined effect of double factors on the spatial differentiation of forest fragmentation was stronger than that of a single factor,and the combination of human activity and other factors had a stronger driving effect on forest fragmentation.(5)Based on the forest fragmentation intensity value within the 3 km×3 km grid,the forest area of Xinyu City was divided into five categories by using the natural breaking points:high fragmentation area,second high fragmentation area,medium fragmentation area,second low fragmentation area,and low fragmentation area.By comprehensively considering the regional functions orientation,economic development level and ecological carrying capacity,the planning prerequisites of improving forest quality and strengthening ecological benefits,three differentiated forest resource regulation models including the intensive development mode,high-density development mode and conservation development mode were proposed for Xinyu City.Meanwhile,the control goals and specific measures of forest management in each mode were also put forward.These established models can provide technical references for the rational allocation and coordinated development of forest resources in forests. |