A method for delineating restoration zones based on biodiversity and habitat stress

Abstract

This invention discloses a method for delineating restoration areas based on biodiversity and habitat stress in the field of natural resource management technology. The method includes acquiring species distribution records and environmental variable data for each species in a target spatial area, and then filtering the environmental variable data to obtain filtered environmental variable data. This invention processes the input data using multiple models in an integrated species distribution model to calculate species richness, functional diversity, and phylogenetic diversity as representative indicators. This overcomes the limitations of previous methods that only focused on a single dimension of biodiversity information, providing a more comprehensive representation of biodiversity levels. By constructing a comprehensive biodiversity index and a comprehensive habitat stress index to comprehensively identify priority restoration areas, this method avoids the limitations and shortcomings of previous methods that only focused on a single dimension of biodiversity, and fully considers the dynamic coupling relationship between biodiversity and habitat conditions.

Description

Technical Field

[0001] This invention relates to the field of natural resource management technology, specifically a method for delineating restoration areas based on biodiversity and habitat stress. Background Technology

[0002] The sharp decline in biodiversity is one of the most severe environmental challenges facing the world today. According to the Millennium Ecosystem Assessment, approximately 12% of birds, 23% of mammals, and 32% of amphibians are threatened with extinction. Habitat change caused by climate and human activities is the main driver of altered species community composition. Biodiversity loss not only weakens various ecosystem services, such as pest control, pollination, and carbon sequestration, but also poses a direct threat to human health and well-being. Therefore, restoring habitats to maintain biodiversity has become a core objective in promoting regional sustainable development and ecological restoration of national land space. To maximize the resource benefits of ecological restoration, scientifically identifying priority areas for habitat restoration is crucial.

[0003] However, existing methods for determining priority areas for habitat restoration only consider the single dimension of species distribution or habitat stress, failing to fully take into account the dynamic coupling relationship between biodiversity and habitat conditions. When delineating priority areas, most methods focus only on the single indicator of species richness, neglecting the comprehensive assessment of the multidimensional value of biodiversity, such as functional diversity and phylogenetic diversity. Traditional biodiversity modeling often relies on a single model algorithm, and the differences between the prediction results of different models may increase the uncertainty of the prediction of the distribution of some species, thus affecting the accuracy of priority area delineation. Summary of the Invention

[0004] The purpose of this invention is to provide a method for delineating restoration areas based on biodiversity and habitat stress, in order to solve the problem that the above-mentioned approach only considers the single dimension of species distribution or habitat stress, fails to fully take into account the dynamic coupling relationship between biodiversity and habitat conditions, and neglects the comprehensive assessment of the multidimensional value of biodiversity such as functional diversity and phylogenetic diversity.

[0005] To achieve the above objectives, the present invention provides the following technical solution: A method for delineating restoration areas based on biodiversity and habitat stress includes the following steps: S1: Obtain species distribution records and environmental variable data for each species in the target spatial area, and filter the environmental variable data to obtain the filtered environmental variable data; S2: Construct an integrated species distribution model by inputting the species distribution records and the filtered environmental variable data into the integrated species distribution model to generate predicted comprehensive species distribution probability raster and binary distribution raster data; S3: Based on the binary distribution raster data of each species, calculate species richness, functional diversity and phylogenetic diversity, and obtain the comprehensive biodiversity index by weighting using the entropy weight method; S4: Obtain data on the economic development index, population index, and human footprint index, and obtain the comprehensive habitat stress index by weighting using the entropy weighting method; S5: Perform bivariate local spatial autocorrelation analysis on the comprehensive biodiversity index and the comprehensive habitat stress index to obtain spatial clustering results of the preset spatial combination type, and delineate priority areas for biodiversity restoration in the target spatial area based on the spatial clustering results.

[0006] As a further aspect of the present invention: in S1, the latitude and longitude coordinates of species occurrence records are obtained based on the species distribution database, and a species distribution record is formed; The environmental variable data include climate variables and landscape variables. The climate variables include annual average temperature, average daily temperature range, isotherm, seasonal temperature variation, maximum temperature of the hottest month, minimum temperature of the coldest month, annual temperature variation range, average temperature of the wettest quarter, average temperature of the driest quarter, average temperature of the warmest quarter, average temperature of the coldest quarter, annual average precipitation, precipitation of the wettest month, precipitation of the driest month, coefficient of variation of precipitation, precipitation of the wettest quarter, precipitation of the driest quarter, precipitation of the warmest quarter, precipitation of the coldest quarter, and wind speed. The landscape variables include normalized vegetation index, Shannon diversity index, patch density, separation index, fractal dimension, and patch proximity distance.

[0007] As a further aspect of the present invention: in S1, the environmental variables are screened using Spearman correlation analysis to obtain screened environmental variable data, wherein the screening condition for the Spearman correlation analysis is that the absolute value of the correlation coefficient |R| is less than 0.7.

[0008] As a further aspect of the present invention: in S2, the integrated species distribution model includes a generalized linear model, a generalized additive model, an artificial neural network model, and a decision tree model; The integrated species distribution model divides the target spatial area into a spatial grid with grid units of a preset resolution; The species distribution records and the filtered environmental variable data are input into the generalized linear model to obtain linear predicted distribution probability data. The linear predicted distribution probability data are coupled with the spatial raster to obtain the first species distribution probability raster. The species distribution records and the filtered environmental variable data are input into the generalized additive model to obtain the additive predicted distribution probability data. The additive predicted distribution probability data is coupled with the spatial raster to obtain the second species distribution probability raster. The species distribution records and screened environmental variable data are input into the artificial neural network model to obtain artificially predicted distribution probability data. The artificially predicted distribution probability data are coupled with the spatial raster to obtain the third species distribution probability raster. The species distribution records and the filtered environmental variable data are input into the decision tree model to obtain the decision prediction distribution probability data. The decision prediction distribution probability data is coupled with the spatial raster to obtain the fourth species distribution probability raster. The first species distribution probability grid, the second species distribution probability grid, the third species distribution probability grid, and the fourth species distribution probability grid are weighted and integrated according to a preset method to obtain a comprehensive species distribution probability grid. The comprehensive species distribution probability raster is predicted based on the true skill statistics assessment model to obtain the final distribution probability corresponding to the spatial raster. The final distribution probability is then binarized to obtain binary distribution raster data of species presence or absence.

[0009] As a further aspect of the present invention: In S2, the generalized linear model, the generalized additive model, the artificial neural network model, and the decision tree model repeatedly predict the environmental variable data after species distribution records and screening a preset number of times. The average of the data repeatedly predicted by the generalized linear model is used to obtain the linear prediction distribution probability data, the average of the data repeatedly predicted by the generalized additive model is used to obtain the additive prediction distribution probability data, the average of the data repeatedly predicted by the artificial neural network model is used to obtain the artificial prediction distribution probability data, and the average of the data repeatedly predicted by the decision tree model is used to obtain the decision prediction distribution probability data.

[0010] As a further aspect of the present invention: the preset resolution is 1km-10km, and the preset number of times is 2-4 times.

[0011] As a further aspect of the present invention: in S2, the binary distribution raster data of each species are accumulated to obtain the species richness; Obtain the species traits corresponding to the species, quantify the volume occupied by the species traits in the target spatial region based on the binary distribution raster data corresponding to the species, and calculate the functional diversity. Obtain the species phylogenetic tree dataset, calculate the total length of species branches based on the binary distribution raster data corresponding to the species, and obtain the phylogenetic diversity; The species richness, functional diversity, and phylogenetic diversity constitute three diversity indicators; The three diversity indicators were weighted using the entropy weighting method to obtain the comprehensive biodiversity index. Normalize the three diversity indicators; In the formula, For the first The object in the first The normalized values ​​of the diversity indicators range from [0, 1]. For the first The grid cell in the first... Original observations on each diversity index; Secondly, calculate the first Under the diversity index, the first The proportion of each grid cell : In the formula, For the first Under the first indicator, the first The feature weight of each object; Calculate the first Information entropy of a diversity indicator The calculation formula is as follows: In the formula, For the first The information entropy value of each diversity indicator is between [0, 1]. Calculate the first Information utility value of each diversity indicator With weight : In the formula, For the first The information utility value of each indicator; For the first The final weight of each indicator.

[0012] In the formula, As a comprehensive index of biodiversity; This represents the normalized species richness. Functional diversity after normalization; For normalized phylogenetic diversity; The weights for normalized species richness; The weights for functional diversity after normalization; This represents the weights of phylogenetic diversity after normalization.

[0013] As a further aspect of the present invention: In S2, economic development index, population index and human footprint index data are obtained based on GDP, population and human footprint datasets, and habitat stress comprehensive index is obtained by weighting using entropy weighting method; In the formula, The comprehensive index of habitat stress; This is the normalized economic development index; The normalized population index; The normalized human footprint index; The weights of the normalized economic development index; The weights of the normalized population index; The weights of the normalized Human Footprint Index.

[0014] As a further aspect of the present invention: in S5, the preset spatial combination type includes high biodiversity-high human stress, high biodiversity-low human stress, low biodiversity-high human stress, and low biodiversity-low human stress; The spatial clustering relationship between the comprehensive biodiversity index and the habitat stress index was quantified using the bivariate local Moran index, and spatial clustering results were obtained for four spatial combination types, including high biodiversity-high human stress, high biodiversity-low human stress, low biodiversity-high human stress, and low biodiversity-low human stress. The bivariate local Moran index is calculated as follows: ; in, = ; = ; In the formula, Here, x represents the bivariate local Moran index value at spatial unit i, and y represents the Biodiversity Index (DIV) and Habitat Stress Index (HSI), respectively. The comprehensive biodiversity index value of the i-th grid cell. i, The value of the habitat stress composite index at neighboring unit j , and σ and y are the standardized values ​​of variables x and y, respectively, and μ and σ are the mean and standard deviation of the corresponding variables, respectively; This is a spatial weight matrix that quantifies the spatial proximity relationship between positions i and j. Based on the calculation results of the bivariate local Moran index, each grid cell is classified into spatial clustering results of a preset spatial combination type; Based on the preset restoration priority, and according to the spatial clustering results, priority areas for biodiversity restoration in the target spatial area are delineated.

[0015] As a further aspect of the present invention, the preset repair priority is as follows: High priority: Areas with high biodiversity and high habitat stress; Higher priority: Areas with high biodiversity and low habitat stress; Lower priority: Areas with low biodiversity and high habitat stress; Low priority: Low biodiversity - low habitat stress areas.

[0016] Compared with the prior art, the beneficial effects of the present invention are: 1. This invention processes input data through multiple models in an integrated species distribution model to obtain comprehensive species distribution probability raster and binary distribution raster data. It then calculates species richness, functional diversity, and phylogenetic diversity as representative indicators, overcoming the limitations of previous approaches that focused only on a single dimension of biodiversity information. This provides a more comprehensive and complete representation of biodiversity levels. Regarding the delineation of priority areas for spatial restoration, this invention constructs a comprehensive biodiversity index and a comprehensive habitat stress index to determine priority restoration areas. This avoids the limitations and shortcomings of previous approaches that focused only on a single dimension of biodiversity, fully considering the dynamic coupling relationship between biodiversity and habitat conditions. This allows limited resources to be invested more effectively in restoration practices, improving the effectiveness of national land space ecological restoration.

[0017] 2. This invention uses a generalized linear model, a generalized additive model, an artificial neural network model, and a decision tree model in an integrated species distribution model to predict environmental variable data after species distribution records and screening. This results in predicted distribution probability data from multiple models, avoiding the differences caused by a single model algorithm that could lead to inaccurate predictions of some species distributions, and ensuring the accuracy of the final priority area delineation. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the method flow of the present invention; Figure 2 Spearman correlation analysis and variable selection of environmental variables for Implementation Example 1 of this invention; Figure 3 This invention provides a ranking of the importance of environmental variables in integrated species distribution modeling for Implementation Example 1. Figure 4 This invention provides a graph showing the results of species richness, functional diversity, and phylogenetic diversity in Implementation Example 1. Figure 5 This invention provides a comprehensive biodiversity index result chart for Implementation Example 1; Figure 6 This invention provides a graph showing the economic level index, population size index, and human footprint index results of Implementation Example 1. Figure 7 This invention provides a comprehensive habitat stress index result diagram for Implementation Example 1; Figure 8 This invention provides a spatial repair priority area identification result diagram for Implementation Example 1. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Please see Figure 1 This embodiment provides a method for delineating restoration areas based on biodiversity and habitat stress, including the following steps: S1: Obtain species distribution records and environmental variable data for each species in the target spatial area, and filter the environmental variable data to obtain the filtered environmental variable data; S2: Construct an integrated species distribution model. Input species distribution records and filtered environmental variable data into the integrated species distribution model to generate predicted integrated species distribution probability raster and binary distribution raster data. S3: Based on the binary distribution raster data of each species, calculate species richness, functional diversity and phylogenetic diversity, and obtain the comprehensive biodiversity index by weighting using the entropy weight method; S4: Obtain data on the economic development index, population index, and human footprint index, and obtain the comprehensive habitat stress index by weighting using the entropy weighting method; S5: Perform bivariate local spatial autocorrelation analysis on the comprehensive biodiversity index and the comprehensive habitat stress index to obtain spatial clustering results for the preset spatial combination type, and delineate priority areas for biodiversity restoration in the target spatial area based on the spatial clustering results.

[0021] Specifically, this invention establishes an integrated species distribution model. Multiple models within this model process the input data to obtain comprehensive species distribution probability raster and binary distribution raster data. Species richness, functional diversity, and phylogenetic diversity are then calculated as representative indicators, overcoming the limitations of previous methods that focused only on a single dimension of biodiversity information. This approach provides a more comprehensive and complete representation of biodiversity levels. Regarding the delineation of priority areas for spatial restoration, a comprehensive biodiversity index and a comprehensive habitat stress index are constructed to determine priority areas. This avoids the limitations and shortcomings of previous methods that focused only on a single dimension of biodiversity. It fully considers the dynamic coupling relationship between biodiversity and habitat conditions. Multiple models within the integrated species distribution model obtain prediction results, which are then integrated to obtain the final prediction result. This avoids discrepancies caused by single-model algorithms, preventing inaccurate predictions of some species distributions and ensuring the accuracy of the final priority area delineation. This method allows limited resources to be invested more effectively in restoration practices, improving the effectiveness of national land space ecological restoration.

[0022] In this embodiment, in S1, the latitude and longitude coordinates of species occurrence records are obtained based on the species distribution database, and a species distribution record is formed; Environmental variable data include climate variables and landscape variables. Climate variables include annual average temperature, average daily temperature range, isotherm, seasonal temperature variation, maximum temperature of the hottest month, minimum temperature of the coldest month, annual temperature range, average temperature of the wettest quarter, average temperature of the driest quarter, average temperature of the warmest quarter, average temperature of the coldest quarter, annual average precipitation, precipitation of the wettest month, precipitation of the driest month, coefficient of variation of precipitation, precipitation of the wettest quarter, precipitation of the driest quarter, precipitation of the warmest quarter, precipitation of the coldest quarter, and wind speed. Landscape variables included normalized vegetation index, Shannon diversity index, patch density, separation index, fractal dimension, and patch proximity distance.

[0023] Specifically, the acquisition and processing of environmental variable data are achieved through multiple means, including but not limited to remote sensing image interpretation, meteorological station observation data interpolation, and geographic information system analysis. After the data is standardized, the dimensional differences between different variables can be effectively reduced, thereby improving the accuracy of model predictions.

[0024] In this embodiment, in S1, the environmental variables are screened using Spearman correlation analysis to obtain the screened environmental variable data. The screening condition for Spearman correlation analysis is that the absolute value of the correlation coefficient |R| is less than 0.7.

[0025] Specifically, Spearman correlation analysis can effectively identify and eliminate factors with high collinearity in environmental variables, thereby avoiding unstable predictions due to multicollinearity. The filtered environmental variable data can not only more accurately reflect the actual state of the ecosystem, but also significantly improve the efficiency and reliability of subsequent modeling.

[0026] In this embodiment, in S2, the integrated species distribution model includes a generalized linear model, a generalized additive model, an artificial neural network model, and a decision tree model; The integrated species distribution model divides the target spatial area into a spatial grid with grid cells of preset resolution; The species distribution records and the filtered environmental variable data are input into the generalized linear model to obtain linear predicted distribution probability data. The linear predicted distribution probability data are coupled with the spatial raster to obtain the first species distribution probability raster. The species distribution records and the filtered environmental variable data are input into the generalized additive model to obtain the additive predicted distribution probability data. The additive predicted distribution probability data is coupled with the spatial raster to obtain the second species distribution probability raster. The species distribution records and screened environmental variable data are input into the artificial neural network model to obtain artificially predicted distribution probability data. The artificially predicted distribution probability data are coupled with the spatial raster to obtain the third species distribution probability raster. The species distribution records and the filtered environmental variable data are input into the decision tree model to obtain the decision prediction distribution probability data. The decision prediction distribution probability data is coupled with the spatial raster to obtain the fourth species distribution probability raster. The first species distribution probability grid, the second species distribution probability grid, the third species distribution probability grid, and the fourth species distribution probability grid are weighted and integrated according to a preset method to obtain a comprehensive species distribution probability grid. The model uses a true skill statistic to predict the overall species distribution probability grid, obtaining the final distribution probability corresponding to the spatial grid. The final distribution probability is then binarized to obtain binary distribution grid data indicating whether a species exists or not. The true skill statistic assessment model uses the TSS method to predict the overall species distribution probability grid. The TSS method is a commonly used method for determining the binarization threshold in species distribution models, used to convert continuous distribution probabilities into binary distributions (suitable / unsuitable). The true skill statistic assessment model can effectively measure the consistency between the predicted results and the actual observation data. Its core lies in calculating the assessment index by comparing the difference between the actual distribution and the model-predicted distribution.

[0027] Specifically, by using generalized linear models, generalized additive models, artificial neural network models, and decision tree models in an integrated species distribution model, the environmental variable data after species distribution records and screening are processed for prediction, resulting in predicted distribution probability data from multiple models. This avoids the situation where a single model is inaccurate in predicting the distribution of some species. These data are coupled with spatial grids to generate corresponding species distribution probability grids. Through weighted integration, the results of each model are combined to form a more accurate comprehensive species distribution probability grid. True skill statistics are used to evaluate and optimize the comprehensive results to ensure the accuracy of the prediction. By binarizing the distribution probability, the presence or absence status of species in different regions is clearly identified.

[0028] In this embodiment, in S2, the generalized linear model, the generalized additive model, the artificial neural network model, and the decision tree model repeatedly predict the environmental variable data after species distribution records and screening a preset number of times. The average of the data repeatedly predicted by the generalized linear model is used to obtain the linear prediction distribution probability data, the average of the data repeatedly predicted by the generalized additive model is used to obtain the additive prediction distribution probability data, the average of the data repeatedly predicted by the artificial neural network model is used to obtain the artificial prediction distribution probability data, and the average of the data repeatedly predicted by the decision tree model is used to obtain the decision prediction distribution probability data.

[0029] Specifically, by calculating the mean of prediction results from multiple repetitions of generalized linear models, generalized additive models, artificial neural network models, and decision tree models, the random errors that may be caused by predictions from a single model can be effectively reduced, thereby improving the stability and reliability of the prediction results.

[0030] In this embodiment, the preset resolution is 1km-10km, and the preset number of times is 2-4.

[0031] Specifically, choosing an appropriate resolution and number of predictions is crucial for improving model accuracy. Setting the resolution range to 1km to 10km effectively balances computational efficiency with the need to represent spatial details. Meanwhile, limiting the number of predictions to between 2 and 4 reduces potential random fluctuations caused by single predictions and avoids resource waste due to excessive repetitive calculations. This ensures that the final generated raster data of species distribution probabilities more closely reflects actual distributions, thus providing a reliable basis for subsequent ecological analysis and decision-making.

[0032] In this embodiment, in S2, the binary distribution raster data of each species are summed to obtain the species richness. Obtain the species traits corresponding to the species, quantify the volume occupied by the species traits in the target spatial region based on the binary distribution raster data corresponding to the species, and calculate the functional diversity. Obtain the species phylogenetic tree dataset, calculate the total length of species branches based on the binary distribution raster data corresponding to the species, and obtain the phylogenetic diversity; Species richness, functional diversity, and phylogenetic diversity are the three diversity indicators; The three diversity indicators were weighted using the entropy weighting method to obtain the comprehensive biodiversity index. Normalize the three diversity indicators; In the formula, For the first The object in the first The normalized values ​​of the diversity indicators range from [0, 1]. For the first The grid cell in the first... Original observations on each diversity index; Secondly, calculate the first Under the diversity index, the first The proportion of each grid cell : In the formula, For the first Under the first indicator, the first The feature weight of each object; Calculate the first Information entropy of a diversity indicator The calculation formula is as follows: In the formula, For the first The information entropy value of each diversity indicator is between [0, 1]. Calculate the first Information utility value of each diversity indicator With weight : In the formula, For the first The information utility value of each indicator; For the first The final weight of each indicator.

[0033] In the formula, As a comprehensive index of biodiversity; This represents the normalized species richness. Functional diversity after normalization; For normalized phylogenetic diversity; The weights for normalized species richness; The weights for functional diversity after normalization; This represents the weights of phylogenetic diversity after normalization.

[0034] In this embodiment, in S2, economic development index, population index and human footprint index data are obtained based on GDP, population and human footprint datasets, and habitat stress comprehensive index is obtained by weighting using entropy weight method; In the formula, The comprehensive index of habitat stress; This is the normalized economic development index; The normalized population index; The normalized human footprint index; The weights of the normalized economic development index; The weights of the normalized population index; The weights of the normalized Human Footprint Index.

[0035] In this embodiment, in S5, the preset spatial combination types include high biodiversity-high human stress, high biodiversity-low human stress, low biodiversity-high human stress, and low biodiversity-low human stress. The spatial clustering relationship between the comprehensive biodiversity index and the habitat stress index was quantified using the bivariate local Moran index, and spatial clustering results were obtained for four spatial combination types, including high biodiversity-high human stress, high biodiversity-low human stress, low biodiversity-high human stress, and low biodiversity-low human stress. The bivariate local Moran index is calculated as follows: In the formula, Here, x represents the bivariate local Moran index value at spatial unit i, and y represents the Biodiversity Index (DIV) and Habitat Stress Index (HSI), respectively. The comprehensive biodiversity index value of the i-th grid cell. i, The value of the habitat stress composite index at neighboring unit j , and σ and y are the standardized values ​​of variables x and y, respectively, and μ and σ are the mean and standard deviation of the corresponding variables, respectively; This is a spatial weight matrix that quantifies the spatial proximity relationship between positions i and j. Based on the calculation results of the bivariate local Moran index, each grid cell is classified into spatial clustering results of a preset spatial combination type, and each cell is classified into the following spatial association type: high biodiversity-high habitat stress area ( >0, and both DIV and HSI are high; high biodiversity - low habitat stress areas ( <0, and DIV is high; HIS is low); low biodiversity - high habitat stress areas ( <0, and DIV is low; HIS is high); high biodiversity - high habitat stress areas ( <0, and both DIV and HSI are low); Based on the preset restoration priority, and according to the spatial clustering results, priority areas for biodiversity restoration in the target spatial area are delineated.

[0036] In this embodiment, the preset repair priority is as follows: High priority: Areas with high biodiversity and high habitat stress; Higher priority: Areas with high biodiversity and low habitat stress; Lower priority: Areas with low biodiversity and high habitat stress; Low priority: Low biodiversity - low habitat stress areas.

[0037] Specifically, restoration priorities are set based on spatial clustering results of biodiversity and habitat stress, and are dynamically adjusted according to actual ecological protection needs. In high-priority areas, ecological restoration projects are implemented to mitigate the strong impact of human activities on biodiversity; in higher-priority areas, preventative protection measures are emphasized to ensure that existing biodiversity levels are not destroyed; in lower-priority areas, the specific causes of habitat stress need to be assessed, and targeted local improvement measures are taken; low-priority areas can serve as control areas for ecological monitoring, used for long-term trend analysis and ecological research. Through hierarchical management, resource allocation is optimized, and the efficiency and effectiveness of restoration work are improved.

[0038] Example 1: Taking a certain area as an example, the specific steps for delineating the repair area are as follows: S1: Obtain environmental variable data for species distribution records and predictions in national land space; observation records of 777 bird species were obtained from the species distribution database; 26 environmental variables from two dimensions, climate and landscape, were selected as predictors (Table 1); all variable raster data were resampled to the same 2 km resolution and standardized; variables with correlation coefficients |R| < 0.7 (e.g., ...) were screened using Spearman correlation analysis. Figure 2As shown), nine variables, including average annual temperature, average diurnal temperature range, annual temperature variation range, driest quarter precipitation, wind speed, normalized difference vegetation index, Shannon diversity index, fractal dimension, and distance to neighboring patches, were included in subsequent integrated species distribution modeling (e.g., annual average temperature, average diurnal temperature range, annual temperature variation range, driest quarter precipitation, wind speed, normalized difference vegetation index, Shannon diversity index, fractal dimension, and distance to neighboring patches). Figure 3 (As shown).

[0039] Table 1 S2: Input species distribution records and environmental variable data into an integrated species distribution model, using four species distribution models: generalized linear model, generalized additive model, artificial neural network and decision tree model; generate a comprehensive species distribution probability raster through an integrated approach, and use the TSS method to obtain binary raster data of species distribution; S3: Based on the binarized results of each species distribution generated in step S2, calculate three biodiversity characterization indicators (such as species richness, functional diversity, and phylogenetic diversity). Figure 4 As shown), the comprehensive biodiversity index is then obtained by weighted summation using the entropy weight method. Figure 5 ); S31: The species richness index is obtained by summing the binary distribution data of species obtained from S2. Figure 4 a) S32: Acquire 11 species traits in birds, including bill length, anterior nostril beak length, beak width, beak depth, tarsus length, wing length, Kipling distance, first secondary flight feather length, hand-wing index, tail length, and body weight. The binary distribution data of species obtained from S2 quantifies the volume occupied by the community in the functional trait space, thereby obtaining the functional diversity index. Figure 4 b) S33: Collect the species phylogenetic tree dataset, calculate the total length of species branches in the community from the species binarized distribution data obtained in S2, and obtain the phylogenetic diversity index ( Figure 4 c) S34: The indices of species richness, functional diversity, and phylogenetic diversity are normalized, and the calculation formula is as follows: In the formula, For the first The object in the first The normalized values ​​of each indicator range from [0, 1]. For the first The grid cell in the first... The original observations on each indicator; S35: The entropy weighting method is used to assign weights to the normalized species richness, functional diversity, and phylogenetic diversity indicators. The calculation formula is as follows: In the formula, For the first Under the first indicator, the first The feature weight of each object; For the first The information entropy value of each indicator ranges from [0, 1]. For the first The information utility value of each indicator; For the first The final weight of each indicator; Table 2. Weighting of Indicators for Calculating the Biodiversity Composite Index S36: The comprehensive biodiversity index is obtained by weighted summation of species richness, functional diversity, and phylogenetic diversity. The formula for its calculation is as follows: In the formula, As a comprehensive index of biodiversity; This represents the normalized species richness. Functional diversity after normalization; This represents the normalized phylogenetic diversity.

[0040] S4: Obtain GDP, population size, and human footprint index to characterize the degree of human disturbance stress on habitats ( Figure 6 The habitat stress comprehensive index is obtained by weighting and summing the economic development index, population index, and human footprint index using the entropy weight method. Figure 7 The calculation formula is as follows: In the formula, The comprehensive index of habitat stress; This is the normalized economic development index; The normalized population index; The normalized human footprint index.

[0041] Table 3. Weighting of Indicators for Calculating the Comprehensive Habitat Stress Index S5: Perform bivariate local spatial autocorrelation analysis on the comprehensive biodiversity index and the comprehensive habitat stress index obtained in steps S3 and S4. Figure 8The study identified four spatial combinations: high biodiversity with high human stress, high biodiversity with low human stress, low biodiversity with high human stress, and low biodiversity with low human stress. Restoration priorities were then assigned in the following order: 1) High priority: High biodiversity with high habitat stress. These areas possess high levels of biodiversity but also face significant habitat disturbance, making them priority areas for habitat restoration. 2) High priority: High biodiversity with low habitat stress. These areas have relatively high biodiversity but low habitat stress caused by human disturbance, making them secondary areas for restoration. 3) Low priority: Low biodiversity with high habitat stress. These areas experience high habitat stress but very low biodiversity levels, resulting in a lower priority for restoration. 4) Low priority: Low biodiversity with low habitat stress. These areas have both low biodiversity value and low levels of human disturbance, making conservation resource investment less necessary. The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for delineating restoration areas based on biodiversity and habitat stress, characterized in that, Includes the following steps: S1: Obtain species distribution records and environmental variable data for each species in the target spatial area, and filter the environmental variable data to obtain the filtered environmental variable data; S2: Construct an integrated species distribution model by inputting the species distribution records and the filtered environmental variable data into the integrated species distribution model to generate predicted comprehensive species distribution probability raster and binary distribution raster data; S3: Based on the binary distribution raster data of each species, calculate species richness, functional diversity and phylogenetic diversity, and obtain the comprehensive biodiversity index by weighting using the entropy weight method; S4: Obtain data on the economic development index, population index, and human footprint index, and obtain the comprehensive habitat stress index by weighting using the entropy weighting method; S5: Perform bivariate local spatial autocorrelation analysis on the comprehensive biodiversity index and the comprehensive habitat stress index to obtain spatial clustering results of the preset spatial combination type, and delineate priority areas for biodiversity restoration in the target spatial area based on the spatial clustering results.

2. The method for delineating restoration areas based on biodiversity and habitat stress according to claim 1, characterized in that, In S1, the latitude and longitude coordinates of species occurrence records are obtained based on the species distribution database, and species distribution records are formed; The environmental variable data include climate variables and landscape variables. The climate variables include annual average temperature, average daily temperature range, isotherm, seasonal temperature variation, maximum temperature of the hottest month, minimum temperature of the coldest month, annual temperature variation range, average temperature of the wettest quarter, average temperature of the driest quarter, average temperature of the warmest quarter, average temperature of the coldest quarter, annual average precipitation, precipitation of the wettest month, precipitation of the driest month, coefficient of variation of precipitation, precipitation of the wettest quarter, precipitation of the driest quarter, precipitation of the warmest quarter, precipitation of the coldest quarter, and wind speed. The landscape variables include normalized vegetation index, Shannon diversity index, patch density, separation index, fractal dimension, and patch proximity distance.

3. The method for delineating restoration areas based on biodiversity and habitat stress according to claim 2, characterized in that, In S1, the environmental variables are filtered using Spearman correlation analysis to obtain filtered environmental variable data. The filtering condition for Spearman correlation analysis is that the absolute value of the correlation coefficient |R| is less than 0.

7.

4. The method for delineating restoration areas based on biodiversity and habitat stress according to claim 3, characterized in that, In S2, the integrated species distribution model includes a generalized linear model, a generalized additive model, an artificial neural network model, and a decision tree model; The integrated species distribution model divides the target spatial area into a spatial grid with grid units of a preset resolution; The species distribution records and the filtered environmental variable data are input into the generalized linear model to obtain linear predicted distribution probability data. The linear predicted distribution probability data are coupled with the spatial raster to obtain the first species distribution probability raster. The species distribution records and the filtered environmental variable data are input into the generalized additive model to obtain the additive predicted distribution probability data. The additive predicted distribution probability data is coupled with the spatial raster to obtain the second species distribution probability raster. The species distribution records and screened environmental variable data are input into the artificial neural network model to obtain artificially predicted distribution probability data. The artificially predicted distribution probability data are coupled with the spatial raster to obtain the third species distribution probability raster. The species distribution records and the filtered environmental variable data are input into the decision tree model to obtain the decision prediction distribution probability data. The decision prediction distribution probability data is coupled with the spatial raster to obtain the fourth species distribution probability raster. The first species distribution probability grid, the second species distribution probability grid, the third species distribution probability grid, and the fourth species distribution probability grid are weighted and integrated according to a preset method to obtain a comprehensive species distribution probability grid. The comprehensive species distribution probability raster is predicted based on the true skill statistics assessment model to obtain the final distribution probability corresponding to the spatial raster. The final distribution probability is then binarized to obtain binary distribution raster data of species presence or absence.

5. The method for delineating restoration areas based on biodiversity and habitat stress according to claim 4, characterized in that, In S2, the generalized linear model, the generalized additive model, the artificial neural network model, and the decision tree model repeatedly predict the environmental variable data after species distribution records and screening a preset number of times. The average of the data repeatedly predicted by the generalized linear model is used to obtain the linear prediction distribution probability data, the average of the data repeatedly predicted by the generalized additive model is used to obtain the additive prediction distribution probability data, the average of the data repeatedly predicted by the artificial neural network model is used to obtain the artificial prediction distribution probability data, and the average of the data repeatedly predicted by the decision tree model is used to obtain the decision prediction distribution probability data.

6. The method for delineating restoration areas based on biodiversity and habitat stress according to claim 5, characterized in that, The preset resolution is 1km-10km, and the preset number of times is 2-4 times.

7. The method for delineating restoration areas based on biodiversity and habitat stress according to claim 6, characterized in that, In S2, the species richness is obtained by summing the binary distribution raster data of each species. Obtain the species traits corresponding to the species, quantify the volume occupied by the species traits in the target spatial region based on the binary distribution raster data corresponding to the species, and calculate the functional diversity. Obtain the species phylogenetic tree dataset, calculate the total length of species branches based on the binary distribution raster data corresponding to the species, and obtain the phylogenetic diversity; The species richness, functional diversity, and phylogenetic diversity constitute three diversity indicators; The three diversity indicators were weighted using the entropy weighting method to obtain the comprehensive biodiversity index. Normalize the three diversity indicators; In the formula, For the first The object in the first The normalized values ​​of the diversity indicators range from [0, 1]. For the first The grid cell in the first... Original observations on each diversity index; Secondly, calculate the first Under the diversity index, the first The proportion of each grid cell : In the formula, For the first Under the first indicator, the first The feature weight of each object; Calculate the first Information entropy of a diversity indicator The calculation formula is as follows: In the formula, For the first The information entropy value of each diversity indicator is between [0, 1]. Calculate the first Information utility value of each diversity indicator With weight : In the formula, For the first The information utility value of each indicator; For the first The final weight of each indicator; In the formula, As a comprehensive index of biodiversity; This represents the normalized species richness. Functional diversity after normalization; For normalized phylogenetic diversity; The weights for normalized species richness; The weights for functional diversity after normalization; This represents the weights of phylogenetic diversity after normalization.

8. The method for delineating restoration areas based on biodiversity and habitat stress according to claim 7, characterized in that, In S2, economic development index, population index and human footprint index data are obtained based on GDP, population and human footprint datasets, and habitat stress comprehensive index is obtained by weighting using entropy weight method; In the formula, The comprehensive index of habitat stress; This is the normalized economic development index; The normalized population index; The normalized human footprint index; The weights of the normalized economic development index; The weights of the normalized population index; The weights of the normalized Human Footprint Index.

9. The method for delineating restoration areas based on biodiversity and habitat stress according to claim 8, characterized in that, In S5, the preset spatial combination types include high biodiversity-high human stress, high biodiversity-low human stress, low biodiversity-high human stress, and low biodiversity-low human stress. The spatial clustering relationship between the comprehensive biodiversity index and the habitat stress index was quantified using the bivariate local Moran index, and spatial clustering results were obtained for four spatial combination types, including high biodiversity-high human stress, high biodiversity-low human stress, low biodiversity-high human stress, and low biodiversity-low human stress. The bivariate local Moran index is calculated as follows: ； in, = ; = ； In the formula, Here, x represents the bivariate local Moran index value at spatial unit i, and y represents the Biodiversity Index (DIV) and Habitat Stress Index (HSI), respectively. The comprehensive biodiversity index value of the i-th grid cell. i, The value of the habitat stress composite index at neighboring unit j , and σ and y are the standardized values ​​of variables x and y, respectively, and μ and σ are the mean and standard deviation of the corresponding variables, respectively; This is a spatial weight matrix that quantifies the spatial proximity relationship between positions i and j. Based on the calculation results of the bivariate local Moran index, each grid cell is classified into spatial clustering results of a preset spatial combination type; Based on the preset restoration priority, and according to the spatial clustering results, priority areas for biodiversity restoration in the target spatial area are delineated; Based on the preset restoration priority, and according to the spatial clustering results, priority areas for biodiversity restoration in the target spatial area are delineated.

10. The method for delineating restoration areas based on biodiversity and habitat stress according to claim 9, characterized in that, The preset repair priorities are as follows: High priority: Areas with high biodiversity and high habitat stress; Higher priority: Areas with high biodiversity and low habitat stress; Lower priority: Areas with low biodiversity and high habitat stress; Low priority: Low biodiversity - low habitat stress areas.

Images