Predictive analysis method for spatio-temporal distribution of species structure

Abstract

The invention discloses a predictive analysis method for spatio-temporal distribution of a species structure, and belongs to the technical field of ecological data analysis. The method comprises the following steps: S1, monitoring species data and related environmental factor data in a specific space-time range; s2, calculating a species uniformity index representing community stability based on the species data; s3, performing quantitative evaluation on the monitored environmental factor data by adopting a preset index evaluation method; s4, establishing a functional quantitative relationship between the species uniformity index and one or more environmental factor state indexes; and S5, simulating and predicting the variation trend of the species uniformity index in the future. By focusing the quantitative correlation between the species uniformity and the environmental factors and constructing the prediction model based on the correlation, the community structure change can be more sensitively captured, and the potential influence of the environmental change on the community stability can be quantitatively predicted; and more accurate and more targeted scientific basis and decision support are provided for prospective biodiversity protection and ecological system management.

Description

Technical Field

[0001] The present invention relates to the technical field of ecological data analysis, and in particular to a method for predicting species community uniformity based on the association of environmental factors. Background Art

[0002] The spatial and temporal distribution of species community structure is the key to assessing the state and dynamic changes of ecosystems. Existing technologies often use species composition surveys and community diversity indices (such as the Shannon index) to analyze the spatial and temporal distribution of species. H' ) calculation, environmental factor monitoring, and community-environment relationship analysis. These methods provide a certain basis for understanding the ecological pattern.

[0003] However, existing technologies still have the following major shortcomings in terms of in-depth understanding of community dynamics and prediction of future changes: Insufficient depth of analysis of species evenness: Existing studies often calculate the Shannon index ( H' ), but less commonly species evenness (such as the Pielou evenness index J' ) as the core analysis object for in-depth research. Species evenness is an important sensitive indicator for measuring community stability, and its changes can often reflect environmental stress earlier. Existing technologies do not pay enough attention to and analyze this indicator, which affects the accuracy of early warning of deterioration of community health or reduced stability.

[0004] Lack of quantitative prediction capability of evenness based on environmental factors: Existing analysis methods mostly describe or explain the historical or current statistical relationship between species evenness and environmental factors, and it is difficult to establish a clear and robust quantitative functional relationship and build an ecological model with predictive function based on it. Therefore, it is difficult to quantitatively predict how the species evenness index will respond to future environmental changes.

[0005] Lack of an integrated predictive analysis framework: Existing technologies such as species monitoring, index calculation, environmental assessment and relationship analysis often fail to be organically integrated into a systematic analysis process to identify key environmental drivers, establish quantitative predictive relationships between environmental factors and species uniformity, and apply models to predict future scenarios, which limits the transformation from basic data to effective management decisions.

[0006] In summary, existing technologies have obvious shortcomings in using species evenness to predict community stability, especially in establishing quantitative prediction models between species evenness and environmental factors. It is urgent to develop new technical methods to achieve scientific predictions of the dynamics of community stability under future environmental changes, and provide more accurate technical support for biodiversity conservation and ecosystem management. Summary of the invention

[0007] The present invention aims to solve the following major technical problems existing in the prior art in analyzing the spatiotemporal distribution of species structure: Insufficient depth of analysis of key indicators of community stability (species evenness): Existing technologies often focus on species richness or comprehensive diversity indices, and fail to fully utilize species evenness as a sensitive indicator reflecting community stability and early stress response, resulting in the assessment of changes in community health status may not be timely and accurate enough.

[0008] Lack of quantitative prediction capabilities based on environmental driving factors: Existing methods mostly describe or explain the historical or current relationship between species evenness and environmental factors. It is difficult to establish a clear quantitative prediction model, and it is impossible to quantitatively predict future changes in species evenness based on future changes in environmental conditions (such as pollution levels, climate change, etc.), which limits the foresight of ecological management.

[0009] Lack of an integrated predictive analysis framework: The data monitoring, index calculation, environmental assessment and relationship analysis links in existing technologies are often not organically integrated into a systematic predictive analysis process with "environmental factors-quantitative relationship of uniformity-prediction model-management application" as the main line, making it difficult to efficiently transform basic data into a scientific basis for supporting proactive management decisions.

[0010] Therefore, the present invention is committed to providing a new analytical method to overcome the above-mentioned defects, especially by focusing on species evenness and establishing a functional quantitative predictive relationship between it and environmental factors, so as to achieve forward-looking prediction of community stability dynamics and provide more effective technical support for biodiversity conservation and ecosystem management.

[0011] In order to solve the above technical problems, the present invention provides a predictive analysis method for the spatiotemporal distribution of species structure, which is characterized by comprising the following core steps:

[0012] Data monitoring and collection (S1): Within the preset research time and space scope (for example, a specific geographical area, a specific time point or a continuous time series), systematically monitor and collect species data of the target species community. The species data should at least contain the information required to calculate the community structure index (especially the evenness index), that is, the species list and the number of individuals of each species (which can be used to calculate density); preferably, species biomass information can also be included for a more comprehensive assessment. At the same time, synchronously monitor the environmental factor data closely related to the living environment of the species community. The selected environmental factors should reflect the key natural or human factors that may affect the community structure (for example, physical and chemical indicators of water or soil, pollutant concentrations, etc.).

[0013] Calculation of community structure index (S2): Based on the species data obtained in the monitoring step, calculate the key index characterizing the species community structure. In particular, calculate the species evenness index (e.g., Pielou evenness index J ′), which is used to quantify the degree of balance in the distribution of the number of individuals of each species in the community. Preferably, the species diversity index (e.g., Shannon index H ′) can also be calculated at the same time to provide more comprehensive information on the community structure.

[0014] Quantitative evaluation of environmental factors (S3): The original data of each environmental factor monitored in the monitoring step are quantified using a preset, standardized index evaluation method and converted into a dimensionless environmental factor status index. This step aims to eliminate the influence of different environmental factor dimensions and units, so as to facilitate subsequent unified correlation analysis and model construction. For example, the single factor pollution index method, standard index method or specific deviation calculation formula can be used according to relevant environmental quality standards (such as seawater quality standards, sediment quality standards, etc.).

[0015] Establishing a quantitative relationship between evenness and environmental factors (S4): This is one of the core steps of the present invention. Perform a statistical correlation analysis on the species evenness index calculated in step (S2) and one or more key environmental factor status indexes obtained in step (S3). By using methods such as the Pearson correlation test, the correlation coefficient (r) and significance level (P) between them are calculated to identify which environmental factors have a statistically significant effect on species evenness, and determine whether the effect is positive or negative, as well as the intensity of the effect. Furthermore, for significantly correlated environmental factors, a quantitative functional relationship between the species evenness index and the status index of these environmental factors can be established by regression analysis (linear or nonlinear) and other methods (for example, J′=f(E 1 , E 2 , ...), where Ei is the i-th significantly correlated environmental factor state index).

[0016] Constructing and applying a predictive ecological model (S5): This is another core step and ultimate goal of the present invention. Based on the quantitative relationship between the species evenness index and one or more significantly correlated environmental factor state indices established in the relationship building step (e.g., a regression equation or correlation coefficient itself), a predictive ecological model is constructed. The core mechanism of the model is to use the known "environmental factor state-species evenness response" relationship to predict future species evenness. The specific form of the model can be selected according to data characteristics and research needs, such as a simple linear regression model, a multivariate linear regression model, a nonlinear regression model, or a regression model based on machine learning (such as random forest regression, support vector regression, etc.) can be selected for more complex relationships. After the model is constructed, by inputting the set values ​​of the future environmental factor state index (these set values ​​can be derived from environmental change scenario simulations, pollution control targets, expected effects of management intervention measures, etc.), the prediction model can be used to calculate and output the predicted value or change trend of the future species evenness index.

[0017] Compared with the existing technology, the predictive analysis method for the spatiotemporal distribution of species structure provided by the present invention has the following significant beneficial effects: It improves the monitoring sensitivity and early warning capability of changes in community stability: The present invention takes the species evenness index as the core analysis and prediction object. Since evenness is usually more sensitive to environmental stress than species richness, it can capture signals of community structure imbalance and decreased stability earlier, thereby gaining time for taking preventive measures.

[0018] The key environmental driving forces affecting community stability were revealed and quantified: by establishing a quantitative relationship between species evenness and environmental factor state index, the present invention can not only identify the main environmental pressure factors affecting community stability, but also accurately quantify the degree and direction of the impact of these factors, deepening the understanding of the dynamic response mechanism of the community.

[0019] Achieved quantitative prediction of the future stability state of the community: The core advantage of the present invention is that it has constructed a species evenness prediction model driven by environmental factors, which elevates ecological analysis from the traditional descriptive and explanatory level to the forward-looking prediction level, and can quantitatively predict the possible state of species community evenness under different future environmental scenarios.

[0020] Provide decision support for proactive and precise ecological management: Based on the prediction results of the model, it is possible to: (1) conduct ecological risk assessment and identify high-risk areas or time periods where community stability may decline significantly under future environmental changes; (2) scientifically formulate environmental management thresholds or pollution control targets, such as determining the level at which a certain (or certain) environmental factor needs to be controlled to maintain or restore the ideal community uniformity; (3) evaluate the potential effects of different management intervention measures (such as emission reduction, restoration projects) on improving community stability, thereby optimizing decisions and achieving more proactive, scientific, and cost-effective biodiversity conservation and ecosystem management. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 A flowchart of a predictive analysis method for the spatiotemporal distribution of species structure provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0022] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the following will be combined with the accompanying drawings and specific embodiments to further describe in detail a predictive analysis method for the spatiotemporal distribution of species structure provided by the present invention. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the scope of protection of the present invention. Any simple modifications, equivalent changes and modifications made to the following embodiments based on the technical essence of the present invention are still within the scope of protection of the technical solution of the present invention.

[0023] The core idea of ​​the embodiment of the present invention is: by monitoring species community data and related environmental factor data, calculating the key indicator characterizing community stability - species evenness index, quantifying the state of environmental factors, establishing a functional quantitative relationship between species evenness and the state of environmental factors, and constructing a prediction model based on this relationship, ultimately achieving the prediction of the changing trend of species evenness under future environmental changes, and providing a scientific basis for ecological management.

[0024] Reference Figure 1 The present invention provides a predictive analysis method for the spatiotemporal distribution of species structure, which mainly includes the following steps:

[0025] Step S1: Data monitoring and collection

[0026] This step aims to obtain the basic data needed for predictive analysis. First, it is necessary to determine the temporal and spatial scope of the study, such as a specific geographical area (such as a bay, lake, river section, forest plot, etc.) and the time scale of monitoring (such as seasonal, annual, or long-term positioning monitoring).

[0027] Species data monitoring: Based on the research object and ecosystem type, standard ecological survey methods are used to collect biological samples. For example, for aquatic benthic organisms, grab-type mud samplers, box samplers, etc. can be used; for plankton, plankton nets can be used to collect water samples; for terrestrial vegetation, sample plots can be set up for investigation. After collecting samples, species classification and identification are carried out in the laboratory, and the list of species appearing in each sample plot or station is accurately recorded, and the number of individuals of each class is counted (for calculating density, such as ind / m², cells / L, etc.). In a preferred embodiment, the biomass of each class can also be measured (such as g / m², mg / L, etc.) to more comprehensively evaluate the community structure.

[0028] Environmental factor data monitoring: While collecting biological samples or as synchronously as possible, collect environmental samples (such as water samples, sediment samples, soil samples, etc.) at the same or adjacent sampling points, and measure relevant environmental factors. The measured environmental factors should be key factors that may affect the community structure of the target species. For example, in aquatic ecosystems, they may include water temperature, salinity, pH, dissolved oxygen (DO), chemical oxygen demand (COD), biochemical oxygen demand (BOD), suspended matter, nutrients (such as inorganic nitrogen DIN, active phosphate DIP, total nitrogen TN, total phosphorus TP), chlorophyll a, petroleum, heavy metals, sulfide, total organic carbon (TOC), etc. The determination of environmental factors may include on-site rapid determination and laboratory standard analysis.

[0029] Step S2: Calculation of community structure index

[0030] Based on the species list and individual quantity data obtained in step S101, biological indices that can reflect the community structure characteristics, especially the species evenness index, are calculated.

[0031] Species evenness index (Evenness Index): The index that this invention focuses on. The commonly used calculation method is the Pielou evenness index (J ′), which is calculated as follows: , where H′ is the species diversity index (described below), S is the total number of species in the plot / site (species richness), and lnS is the theoretical maximum diversity index Hmax′ (i.e. when the number of individuals of all species is exactly the same). The value range of J′ is between 0 and 1. The closer the value is to 1, the more evenly the number of individuals of each species in the community is distributed, and the more stable the community structure is; the closer the value is to 0, the more the community is dominated by a few dominant species, the lower the uniformity, and the less stable it may be.

[0032] Diversity Index: The commonly used calculation method is the Shannon-Wiener index (H ′), and its calculation formula is: , where S is the total number of species, Pi is the ratio of the number of individuals of the i-th species ni to the total number of individuals N in the plot / site (Pi = ni / N). H′ combines species richness and evenness information.

[0033] By calculating these indices, the status and changes of species community structure at different times, locations or environmental conditions can be quantitatively assessed.

[0034] Step S3: Quantitative assessment of environmental factors

[0035] In order to facilitate subsequent analysis and model construction, the original environmental factor data with different dimensions and units monitored in step S1 need to be converted into a standardized, dimensionless environmental factor status index. The evaluation method and the standards based on it should be determined in advance, usually referring to the environmental quality standards issued by the state or local government.

[0036] Example 1: Single factor pollution index method (applicable to factors with positive correlation between concentration and pollution degree): For pollutants such as COD, petroleum, inorganic nitrogen, active phosphate, heavy metals, etc., their state index B can be calculated according to the following formula: B = GBI / X. Among them, X is the measured concentration value of the environmental factor, and GBI is the corresponding environmental quality standard limit (for example, the corresponding category limit specified in the "Seawater Quality Standard" GB 3097, the "Surface Water Environmental Quality Standard" GB 3838, and the "Marine Sediment Quality" GB 18668). B≤1 usually means that it is not polluted or meets the standard, B>1 means it is polluted, and the larger the value, the more serious the pollution.

[0037] Example 2: Specific index calculation method (applicable to special factors):

[0038] For dissolved oxygen (DO), the standard is usually the lower limit. Its state index B DO The calculation of B may require a specific formula, for example, it can be used to calculate its ratio to saturated dissolved oxygen, or to calculate its deviation from the standard limit. A possible calculation method (to be determined according to the specific standard and research purpose) is: B DO =(X DO -GBI DO ) / GBI DO (applicable to the lower limit standard), or B DO =(DO max -X DO ) / DO max -GBI DO (Assuming DO max is the saturation or optimal value, GBI DO is the minimum standard value).

[0039] For pH value, its standard is usually a range. Its state index B pH can calculate the deviation between the measured value and the optimum value (such as 8.15 for seawater) or the value in the standard range. For example: B pH = 2 (X pH -pH opt ) / (pH upper iimit-pH lower i imit).

[0040] Example 3: Standard index method (applicable to sediments, etc.): For certain factors in sediments (such as TOC, sulfide, and petroleum), the ratio of the measured value to the corresponding sediment quality standard limit can be directly used as its state index B=X / GBI sediment .

[0041] Through this step, each environmental factor obtained a standardized state index at each sampling point / time point, which facilitates comparison and analysis across factors and across time and space.

[0042] Step S4: Establishing the quantitative relationship between uniformity and environmental factors

[0043] This step is the basis for building a predictive model, which aims to identify the key environmental factors that affect species evenness and quantify the relationship between them.

[0044] (d1) Data collation and matching: The species evenness index J′ calculated in step S2 (for example, forming a time series or spatial distribution data set) is matched with the environmental factor state index (B 1 , B 2 , ..., Bk ) of the corresponding sampling point / time point calculated in step S103 to form a data set containing all variables.

[0045] (d2) Correlation analysis: Statistical methods are used to test the correlation between J′ and each environmental factor state index Bi. Commonly used methods are Pearson correlation analysis (suitable for testing linear relationships) or Spearman rank correlation analysis (suitable for testing monotonic relationships). Calculate the correlation coefficient r (range -1 to 1) and the significance level P value. The absolute value of r indicates the strength of the correlation, and the sign indicates the direction (positive or negative correlation). The P value is used to determine whether the correlation is statistically significant (usually the threshold is set at P<0.05 or P<0.01). Through correlation analysis, environmental factor state indices that have a significant statistical relationship with species evenness J′ can be screened out.

[0046] (d3) Establishment of quantitative relationship (regression analysis): For the significantly correlated environmental factors screened out, the quantitative functional relationship between J′ and these factors can be further established through regression analysis.

[0047] If there is only one significantly correlated environmental factor E (state index) and the relationship is approximately linear, a simple linear regression model can be established: J ′ = a×E+b.

[0048] If there are multiple significantly correlated environmental factors E 1 , E 2 , ..., E k , a multiple linear regression model can be established: J ′=a 0 +a 1 ×E 1 +a 2 ×E 2 +...+a k ×E k .

[0049] If the relationship between J ′ and environmental factors shows obvious nonlinear characteristics, a nonlinear regression model (such as exponential model, logarithmic model, polynomial model, etc.) or a more complex machine learning regression model (see step S5) can be used. Regression analysis not only gives the specific function form, but also provides the model's goodness of fit (such as R 2 The significance test results of the coefficients (values) and coefficients help evaluate the reliability of the quantitative relationship.

[0050] Step S5: Constructing and applying a predictive ecological model This step is the ultimate goal of the present invention, namely, using the established quantitative relationship to predict future species evenness.

[0051] Model construction: Based on the quantitative relationship (usually a regression equation) between the species evenness J ′ established in step S4 and one or more significantly related environmental factor state indices E, a predictive ecological model is constructed. The choice of model should correspond to step S4.

[0052] Statistical models: such as linear regression, multivariate linear regression, and nonlinear regression models. These models are relatively simple in structure and easy to explain.

[0053] Machine learning models: such as Random Forest Regression, Support Vector Regression (SVR), Gradient Boosting Machine (GBM), Neural Networks, etc. These models can handle more complex nonlinear relationships, high-dimensional data, and interactions between factors, and may have higher prediction accuracy, but their interpretability is relatively weak. Model construction usually requires dividing historical data into training sets and test sets, using the training set to train model parameters, and using the test set to evaluate model performance (such as prediction accuracy and generalization ability).

[0054] Model application and prediction:

[0055] Scenario setting: First, you need to set one or more possible environmental scenarios that may occur in the future. These scenarios can be based on:

[0056] Climate change predictions (e.g. the impact of future changes in temperature, precipitation, sea level rise, etc. on environmental factors).

[0057] Socio-economic development planning (e.g. impact of industrial emissions, agricultural activities, urban expansion, etc. on environmental factors).

[0058] Environmental management and governance measures (such as setting pollution reduction targets, implementing ecological restoration projects, etc.). For each scenario, it is necessary to determine the future predicted value of the state index of the relevant environmental factors under that scenario (E future ).

[0059] Prediction calculation: Set the future environmental factor state index value E future Input it into the constructed prediction model and calculate the corresponding predicted value of future species evenness index Predicted J ′. It can predict the value at a single time point or the change trend over a period of time.

[0060] Results interpretation and management applications:

[0061] Risk assessment: Compare the Predicted J ′ values ​​under different scenarios. One or more ecological health thresholds can be set (for example, the ideal range, warning line, critical value of J ′ can be determined based on historical data or expert knowledge). If the predicted J ′ value is lower than a certain threshold, it indicates that under this scenario, the species community in the area may face ecological risks of structural imbalance and decreased stability. The prediction results can be spatially visualized to generate a "species evenness risk assessment map" to intuitively show the risk level faced by different regions in the future.

[0062] Decision support: The prediction results can directly serve the decision-making of environmental management and biodiversity conservation:

[0063] Establish management objectives: For example, in order to maintain species evenness at a certain ideal level (such as J′>0.7), the model can reversely calculate which key environmental factor status indexes need to be controlled within what range, thereby providing a quantitative basis for establishing specific environmental quality standards or management objectives.

[0064] Evaluate intervention effects: Models can be used to simulate the impact of different management options (such as different emission reduction efforts, different remediation measures) on future J ′, compare the cost-effectiveness of different options, and select the optimal management strategy.

[0065] Optimizing the monitoring network: The model can help identify the environmental factors that have the greatest impact on community uniformity and the most sensitive areas, and guide the optimization of the environmental monitoring network layout.

[0066] Establish an early warning system: embed the model into the environmental monitoring system, and issue a warning in time when the real-time monitored environmental factor data indicates that J ′ may drop significantly in the future.

[0067] Example 1: Prediction and analysis of uniformity of large benthic communities in shallow sea areas

[0068] This embodiment takes a large benthic community in a certain coastal shallow sea area in China as an example and applies the method of the present invention.

[0069] (1) Data monitoring (S1):

[0070] Study area: A shallow bay affected by human activities was selected.

[0071] Monitoring period: For three consecutive years, sampling is conducted in four seasons each year: spring (May), summer (August), autumn (November), and winter (February).

[0072] Sampling stations: 20 fixed monitoring stations will be set up in the bay.

[0073] Species data: Three replicate samples were collected at each station using a 0.1 m² grab sampler. The samples were sieved through a 0.5 mm mesh, and the sieve was collected and fixed with 75% alcohol. Macrobenthic animals were identified, counted, and weighed (wet weight) in the laboratory. Species density (ind / m²) and biomass (g / m²) were calculated for each station in each season.

[0074] Environmental factor data: Surface seawater and surface (0-5 cm) sediment samples were collected simultaneously.

[0075] Seawater: On-site measurement of water temperature, salinity, pH, and DO. Laboratory measurement of COD (alkaline potassium permanganate method), DIN (nutrient salt automatic analyzer), DIP (phosphomolybdenum blue method), and petroleum (ultraviolet spectrophotometry).

[0076] Sediment: Determination of particle size (laser particle size analyzer), TOC (element analyzer), sulfide (iodine titration), petroleum (gas chromatography), heavy metals (such as Cu, Zn, Pb, Cd, Cr, As, determined using ICP-MS).

[0077] (2) Calculation of community index (S2): Based on the species list and individual number data of each station in each quarter, the Shannon index H ′ and Pielou evenness index J ′ were calculated.

[0078] ;

[0079] J ′ =H ′ / ln S;

[0080] (3) Quantitative assessment of environmental factors (S3):

[0081] According to the Seawater Quality Standard (GB 3097-1997) and Marine Sediment Quality (GB 18668-2002), it is assumed that the functional positioning of this area is a Class II area.

[0082] Seawater factor: Calculate the single factor pollution index of COD, DIN, DIP, and petroleum B=X / GBI II (GBI II is the Class II standard limit). Calculate B DO and B pH (Using a specific deviation formula).

[0083] Sediment factor: Calculate the single factor standard index B=X / GBI for TOC, sulfide, petroleum, and heavy metals sediment,I (Assuming that a type of standard is used as the ideal background value).

[0084] (4) Establishing the quantitative relationship between uniformity and environmental factors (S4):

[0085] The J′ values ​​of all stations and all seasons and the corresponding B values ​​of each environmental factor state index are organized into a data matrix.

[0086] Pearson correlation analysis was performed. The results of the hypothesis analysis showed that J ′ and the sediment TOC state index B TOC (r= -0.58,P < 0.01)、sulfide state index B Sulfide (r = -0.45,P < 0.05) was significantly negatively correlated with the dissolved oxygen status index B DO (r = 0.35, P < 0.05) showed a significant positive correlation (hypothesis B DO The smaller the index, the better the DO condition).

[0087] Based on the significantly correlated factors, a multivariate linear regression model was constructed: J ′ =β 0 +β 1 ×B TOC +β 2 ×B Sulfide +β 3 ×B DO +ϵEstimate the model coefficients β0, β by the least squares method 1 , β 2 , β 3, and test the significance of the model (F test) and the significance of the coefficients (t test). Assume that a significant regression equation is obtained, for example: J ′ =0.85−0.12×B TOC −0.08×B Sulfide +0.05×B DO (The coefficients are examples).

[0088] (5) Construction and application of prediction models (S5):

[0089] Model construction: The above multivariate linear regression equation was used as the prediction model.

[0090] Model Application:

[0091] Scenario setting: Scenario A (maintaining the status quo): Use the current average value of each environmental factor index.

[0092] Scenario B (intensified organic pollution): Assuming that due to increased pollution, TOC and B Sulfide Both increased by 50%, while B DO constant.

[0093] Scenario C (comprehensive governance): Assuming that through governance, future B TOC and B Sulfide Both decreased by 30%, and B DO Improvement (index value decreases by 20%).

[0094] Prediction calculation: B under each scenario TOC，future , B Sulfide，future , B DO，future Substitute the value into the regression equation and calculate Predicted J ′.

[0095] Predicted J′ A =...(current predicted value);

[0096] Predicted J′ B =... (the predicted value of pollution increase is expected to be significantly lower than the current situation);

[0097] Predicted J′ C =... (predicted value of governance improvement, expected to be higher than the status quo);

[0098] Management implications: The prediction results show that controlling sediment organic pollution and sulfide pollution, while improving the dissolved oxygen conditions of seawater, is the key to maintaining and improving the uniformity of the large benthic community in the bay. TOC and B Sulfide The upper limit of the management control target, and B DOFor areas with low predicted J′ values, pollution source control and sediment environment remediation measures should be implemented as a priority.

[0099] Example 2: Prediction and analysis of the uniformity of phytoplankton communities in freshwater lakes (using machine learning models)

[0100] This embodiment applies the method of the present invention to the phytoplankton community in eutrophic lakes and uses a machine learning model to handle more complex nonlinear relationships.

[0101] (1) Data monitoring (S1): Monitor the phytoplankton species, density (cells / L) and related environmental factors (water temperature T, pH, DO, transparency SD, total nitrogen TN, total phosphorus TP, chlorophyll a Chl-a) in different areas and seasons of the lake.

[0102] (2) Calculation of community indices (S2): Calculate the Shannon index H′ and Pielou evenness index J′ of the phytoplankton community.

[0103] (3) Quantitative assessment of environmental factors (S3): TN and TP are indexed according to the Surface Water Environmental Quality Standard (GB 3838) (for example, their ratio relative to the Class III water standard is calculated as the state index B TN , B TP Other factors such as T, pH, DO, SD, and Chl-a can be used directly with their original values ​​or standardized (such as Z-score standardization).

[0104] (4) Establishing the quantitative relationship between uniformity and environmental factors (S4):

[0105] Because the responses of phytoplankton communities to environmental factors may involve complex nonlinear relationships and interactions, preliminary correlation analysis may show that multiple factors are associated with J ′.

[0106] Choose to build a Random Forest Regression model. TN , B TP , Chl-a value) is divided into a training set (e.g. 70%) and a test set (e.g. 30%).

[0107] (5) Construction and application of prediction models (S5):

[0108] Model construction: Use the training set to train the random forest regression model, with the target variable J ′ and the predictor variables being all the selected environmental factors. Use cross-validation and other methods to adjust the model hyperparameters (such as the number of trees, the maximum depth of each tree, the minimum number of samples for node splitting, etc.) and select the model with the best performance. After the model training is completed, evaluate its prediction performance (such as root mean square error RMSE, determination coefficient R 2 ). The random forest model can also output the relative importance ranking of each environmental factor for predicting J ′.

[0109] Model Application:

[0110] Scenario setting: Scenario D (nutrient load increase): Predict future TN and TP concentration increases leading to B TN , B TP Increase.

[0111] Scenario E (climate warming): The average summer water temperature T is predicted to increase in the future.

[0112] Scenario F (Phosphorus Control Measures): Prediction of the implementation of strict phosphorus control measures TP decline.

[0113] Prediction calculation: Input the future values ​​of environmental factors under each scenario into the trained random forest model to predict Predicted J ′.

[0114] Management implications: Random forest models can reveal the key drivers of phytoplankton uniformity and their relative importance (for example, TP and water temperature may be found to be the most important factors). The prediction results can be used to evaluate the effects of different nutrient management strategies (nitrogen control vs phosphorus control) and climate change adaptation measures on preventing harmful algal blooms (usually accompanied by low uniformity), and provide decision support for lake eutrophication control and algal bloom warning.

[0115] The above embodiments are only exemplary descriptions of the technical solutions of the present invention. Those skilled in the art can select or adjust specific monitoring indicators, index calculation methods, environmental factor quantification methods, correlation analysis techniques, and types of prediction models within the framework of the method of the present invention according to specific application scenarios (different ecosystems, species communities, and types of environmental pressure) and data characteristics. These variations and modifications should all fall within the scope of protection of the present invention.

Claims

1. A predictive analysis method for the spatiotemporal distribution of species structure, characterized in that: The following steps are involved: S1 Monitoring: Monitoring the species data of the target species community within a preset time and space range, wherein the species data at least includes information on species type and individual quantity; and simultaneously monitoring the environmental factor data related to the species community; S2 calculation: based on the species data obtained in the monitoring step, calculating the species evenness index characterizing the species community structure; S3 quantification: using a preset index evaluation method to quantitatively evaluate the environmental factor data obtained in the monitoring step to obtain a state index of each environmental factor; S4 establishing a relationship: determining the functional quantitative relationship between the species evenness index calculated in the calculation step and the one or more environmental factor state indexes obtained in the quantification step through correlation analysis or regression analysis; S5 Construction and prediction: Based on the functional quantitative relationship determined in the relationship establishment step, a predictive ecological model is constructed; and using the predictive ecological model, the set value of the future environmental factor state index is input to predict the future change trend of the species evenness index.

2. The predictive analysis method for the spatiotemporal distribution of species structure according to claim 1, characterized in that: In the monitoring step, the species data also includes species biomass information.

3. The predictive analysis method for the spatiotemporal distribution of species structure according to claim 1, characterized in that: In the monitoring step, the target species community is a marine benthic community, including large benthic animals or small benthic animals.

4. The predictive analysis method for the spatiotemporal distribution of species structure according to claim 1, characterized in that: In the S2 calculation step, the species diversity index is also calculated; the species diversity index is the Shannon index H', which is calculated according to the following expression: ； Among them, S is the number of species, and Pi is the ratio of the number of individuals of the i-th species to the total number of individuals in the sample.

5. The predictive analysis method for the spatiotemporal distribution of species structure according to claim 1, characterized in that: In the S2 calculation step, the species evenness index is the Pielou evenness index J', which is calculated according to the following expression: ； Wherein, H ′ is the Shannon index calculated according to claim 4, and S is the number of species.

6. The predictive analysis method for the spatiotemporal distribution of species structure according to claim 1, characterized in that: The environmental factor data in the S1 monitoring step includes seawater environmental factors and sediment environmental factors; The seawater environmental factor is selected from at least one of pH, dissolved oxygen DO, chemical oxygen demand COD, petroleum, inorganic nitrogen DIN, and active phosphate DIP; The sediment environmental factor is selected from at least one of petroleum, sulfide, and total organic carbon (TOC).

7. The predictive analysis method for the spatiotemporal distribution of species structure according to claim 1, characterized in that: In the quantification step S3, the quantitative assessment of seawater environmental factors adopts a single factor pollution index method or a specific index calculation method; For seawater environmental factors whose concentration is positively correlated with the degree of pollution, the single factor pollution index method is used to calculate the state index B, which is calculated based on the following expression: , where X is the measured concentration value and GBI is the corresponding seawater quality standard limit; For dissolved oxygen DO, the state index B is calculated using a preset specific index formula. DO ; For pH value, the state index B is calculated by calculating the deviation between the measured pH value and the optimum value or the median value of the standard range. pH .

8. The predictive analysis method for the spatiotemporal distribution of species structure according to claim 1, characterized in that: In the S3 quantification step, the quantitative evaluation of sediment environmental factors adopts the standard index method, which is calculated according to the following expression: ； Where X is the measured content of environmental factors in sediment, and GBI sediment is the corresponding marine sediment quality standard limit.

9. The predictive analysis method for the spatiotemporal distribution of species structure according to claim 1, characterized in that: The correlation analysis in the S4 relationship establishment step adopts the Pearson correlation test method to calculate the correlation coefficient r and its significance level P between the species evenness index and the state index of each environmental factor; the functional quantitative relationship is determined based on the correlation coefficient r to determine the influencing factor or the functional relationship obtained by regression analysis.

10. The predictive analysis method for the spatiotemporal distribution of species structure according to claim 1, characterized in that: In the S5 construction and prediction step, the predictive ecological model is a statistical model selected from a linear regression model, a multivariate linear regression model, a nonlinear regression model or a regression model based on machine learning; the predicted change trend of the future species evenness index is used to generate a species community stability risk assessment map, or to provide quantitative decision support for formulating environmental management thresholds, pollution control targets or ecological restoration priorities.

Images