A method for screening environmental stress indicator species based on a multi-species occupation model

By using a multi-species occupancy model and environmental DNA technology, the problem of existing indicator species screening methods relying on expert experience has been solved, and the quantification and reliability of indicator species screening have been achieved, improving the objectivity and comparability of screening results.

CN123120040APending Publication Date: 2026-11-17NANJING UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611543357.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-09-30
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing methods for selecting indicator species rely too heavily on expert experience, resulting in poor repeatability and comparability of results, which makes it difficult to meet the needs of standardized monitoring.

Method used

A multi-species occupation model-based approach was adopted. This approach combines an occupation sub-model and a detection sub-model with a hierarchical Bayesian framework to obtain species abundance information using environmental DNA technology. The multi-species occupation model was then constructed, and the environmental stress effect was iteratively solved to screen out indicator species with significant responses.

Benefits of technology

This approach has enabled the shift from relying on expert experience to model-based quantitative methods in indicator species screening, improving the objectivity, repeatability, and comparability of screening results. The output environmental stress effect quantity accurately reflects the direction and intensity of species' response to environmental stress.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123120040A_ABST
    Figure CN123120040A_ABST
Patent Text Reader

Abstract

This application discloses a method for screening environmental stress indicator species based on a multi-species occupancy model, belonging to the field of aquatic ecological monitoring technology. The method includes: acquiring initial species abundance information and environmental data of multiple aquatic groups within a study area; determining target species abundance information from the initial species abundance information based on the target taxa corresponding to the target ecological zone; determining target environmental stress factors based on environmental data; iteratively solving the response coefficients of a pre-constructed multi-species occupancy model based on the target species abundance information and target environmental stress factors to obtain the environmental stress effect quantity corresponding to each species; and determining a list of target environmental stress indicator species corresponding to the target ecological zone based on the environmental stress effect quantity corresponding to each species. This application achieves objectivity, quantification, and reproducibility in indicator species screening, possesses multi-taxa compatibility and multi-stress source tracing capabilities, and is suitable for watershed ecological health diagnosis, environmental stress source tracing, and routine monitoring.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This application relates to the field of water ecological monitoring technology, and more specifically, to a method for screening environmental stress indicator species based on a multi-species occupancy model. Background Technology

[0002] Indicator species are species that exhibit a significant response to environmental changes in a specific ecological environment. Their distribution and abundance can reflect the trophic status, pollution level, and ecosystem health of water bodies. Therefore, indicator species can quickly and intuitively reflect the health status of ecosystems and the intensity of environmental stress, and are a core technical means for ecological risk early warning, biological integrity assessment, and environmental management. Standardized, quantitative, and scalable indicator species screening methods are needed to support comprehensive ecological monitoring and control.

[0003] According to relevant indicator species screening techniques, firstly, researchers select a few representative "groups of interest" (such as benthic animals or fish) within the study area based on limited historical data or personal experience for field sampling and morphological identification. Subsequently, by comparing the differences in species composition at different sampling points (such as clean sections and polluted sections), and combining a few routine physicochemical indicators (such as dissolved oxygen and total nitrogen), experts manually judge and select species whose abundance or frequency of occurrence changes with environmental gradients as candidate indicator species. Finally, univariate statistical tests (such as t-tests or ANOVA) are conducted on these candidate species to verify whether their differences between "clean" and "polluted" sites are significant, thereby ultimately determining the list of indicator species.

[0004] Therefore, the screening process of the above-mentioned indicator species screening technology relies heavily on qualitative judgment based on expert experience, resulting in poor repeatability and comparability of the results, which makes it difficult to meet the needs of standardized monitoring. Summary of the Invention

[0005] This application aims to overcome the problem that existing indicator species screening methods rely too much on expert experience and provides an objective and quantitative method for screening indicator species under environmental stress.

[0006] According to one aspect of this application, a method for screening environmental stress indicator species based on a multi-species occupation model is provided, comprising: acquiring initial species abundance information of multiple groups of aquatic organisms in a study area and environmental data of the study area; determining target species abundance information corresponding to the target ecological partition from the initial species abundance information based on the target group corresponding to the target ecological partition, wherein the target ecological partition is any one of multiple ecological partitions in the study area; determining target environmental stress factors related to the target group based on the environmental data; iteratively solving the response coefficients of a pre-constructed multi-species occupation model based on the target species abundance information and the target environmental stress factors to obtain the environmental stress effect quantity corresponding to each species in the target group, wherein the multi-species occupation model adopts a hierarchical Bayesian framework; the target species abundance information is used as the response variable of the multi-species occupation model, and the target environmental stress factor is used as the explanatory variable of the multi-species occupation model; and determining a list of target environmental stress indicator species corresponding to the target ecological partition based on the environmental stress effect quantity corresponding to each species in the target group.

[0007] According to the above aspects, in one embodiment, determining the target environmental stress indicator species list corresponding to the target ecological zone based on the environmental stress effect magnitude corresponding to each species in the target taxonomic group includes: selecting an initial environmental stress indicator species list from the target taxonomic group based on a preset species detection probability threshold and the result of sorting the environmental stress effect magnitude by absolute value; verifying the environmental stress response significance of the indicator species in the initial environmental stress indicator species list to obtain the response significance level of each indicator species; and selecting indicator species whose response significance level is less than a preset level threshold to obtain the target environmental stress indicator species list.

[0008] According to the above aspects, in one embodiment, the step of verifying the significance of the environmental stress response of the indicator species in the initial list of environmental stress indicator species to obtain the response significance level of each indicator species includes: performing multiple linear regression analysis on the indicator species in the initial list of environmental stress indicator species to obtain the regression coefficients and significance test P values ​​corresponding to each indicator species; and obtaining the response significance level of each indicator species based on the regression coefficients and significance test P values.

[0009] Based on the above aspects, in one embodiment, the multi-species occupancy model includes an occupancy sub-model and a detection sub-model, and the formula corresponding to the occupancy sub-model is as follows: , , ; in, This represents the actual state of existence of species i at location j; This represents the actual occupancy probability of species i at location j; This indicates that it follows a Bernoulli distribution; This represents a vector of environmental covariates related to the occupancy of species i. express transpose; This represents the response coefficient of species i to environmental stress; (·)express Join function; This represents the mean of the response coefficients for all species. express The covariance matrix, This indicates that the distribution follows a normal distribution; The formula corresponding to the detection sub-model is as follows: , , ; in, This indicates whether species i was detected in the k-th replicate at the j-th location; This represents the probability of species i being detected in the kth repeated instance at the j-th location; This represents a vector of covariates that influence the probability of detecting species i at the j-th location and the k-th repetition during the detection process. express transpose; The probability coefficient for detecting species i is represented by . This represents the mean of the detection probability coefficients for all species; express The covariance matrix.

[0010] According to the above aspects, in one embodiment, obtaining the initial species abundance information of multiple groups of aquatic organisms in the study area includes: collecting environmental DNA samples at various sampling points in the study area; performing PCR amplification on the environmental DNA samples using a multi-primer combination to obtain amplification products; performing high-throughput sequencing on the amplification products to obtain sequencing data; performing bioinformatics analysis on the sequencing data, comparing it with a reference database to complete species annotation, and obtaining annotation results; and calculating the abundance of each species at each sampling point based on the annotation results to obtain the initial species abundance information.

[0011] According to the above aspects, in one embodiment, the step of iteratively solving the response coefficients of a pre-constructed multi-species occupancy model based on the target species abundance information and the target environmental stress factor to obtain the environmental stress effect quantity corresponding to each species in the target group includes: standardizing the target species abundance information into a binary matrix, wherein when each element in the binary matrix is ​​a first value, it means that the species corresponding to that element is detected at the sampling point, and when each element in the binary matrix is ​​a second value, it means that the species corresponding to that element is not detected at the sampling point; and iteratively solving the response coefficients of the pre-constructed multi-species occupancy model based on the binary matrix and the target environmental stress factor to obtain the environmental stress effect quantity corresponding to each species in the target group.

[0012] In accordance with the above aspects, in one embodiment, the method further includes: dividing the study area into multiple ecological zones based on the geographical region and ecosystem type of the study area.

[0013] According to the above aspects, in one embodiment, determining the target environmental stress factor related to the target group based on the environmental data includes: extracting candidate environmental stress factors from the environmental data; performing multicollinearity diagnosis on the candidate environmental stress factors; and determining the environmental stress factor with independent influence as the target environmental stress factor.

[0014] According to the above aspects, in one embodiment, performing multicollinearity diagnosis on the candidate environmental stress factors and determining the environmental stress factors with independent influence as the target environmental stress factors includes: calculating the variance inflation factor value of each candidate environmental stress factor; and determining the candidate environmental stress factors whose variance inflation factor value is less than or equal to a preset factor threshold as the target environmental stress factors.

[0015] According to another aspect of this application, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and executable by the processor, wherein the processor implements the environmental stress indicator screening method of this application when executing the computer program.

[0016] Compared with existing technologies, the environmental stress indicator species screening method based on the multi-species occupation model of this application realizes a fundamental shift from "empirical qualitative" to "model quantitative" in indicator species screening: This application adopts a multi-species occupation model under the hierarchical Bayesian framework. Through the synergistic cooperation of the occupation sub-model and the detection sub-model, it directly estimates and outputs the response intensity (effect size) of species to environmental stress factors and its statistical significance based on separating the true distribution of species from the detection error. This replaces the traditional screening method that relies on expert experience or subjective thresholds, and significantly improves the objectivity, repeatability and comparability of the screening results. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating an environmental stress indicator species screening method based on a multi-species occupation model according to an embodiment of this application.

[0018] Figure 2 The goodness-of-fit test results of a multi-species occupancy model according to an embodiment of this application are shown.

[0019] Figure 3 A box plot of the effect sizes of environmental stress factors at the fish community level is shown according to one embodiment of this application.

[0020] Figure 4 This is a scatter plot ranking the effects of fish species on river connectivity stress according to one embodiment of this application.

[0021] Figure 5 This is a fitting verification diagram of the occupancy probability of a single species (taking the blunt snout barb as an example) as a function of river fragmentation, according to an embodiment of this application.

[0022] Figure 6 This is a scatter plot of the response relationship of sensitive indicator species abundance under a river fragmentation stress gradient, according to an embodiment of this application.

[0023] Figure 7 This is a verification diagram of an environmental stress indicator species for eutrophic phytoplankton sensitive species, according to an embodiment of this application.

[0024] Figure 8 This is a verification diagram of an environmental stress indicator species that indicates eutrophic phytoplankton tolerance according to an embodiment of this application.

[0025] Figure 9 This is a schematic diagram of the structure of an electronic device according to an embodiment of this application. Detailed Implementation

[0026] The present application will now be described in further detail with reference to the accompanying drawings. Figure 1 A flowchart illustrating an environmental stress indicator species screening method based on a multi-species occupation model is shown in one embodiment of this application.

[0027] like Figure 1As shown, in one embodiment, a method for screening environmental stress indicator species based on a multi-species occupancy model includes the following steps S1 to S5: S1, obtaining initial species abundance information of multiple groups of aquatic organisms in the study area and environmental data of the study area; S2, based on the target group corresponding to the target ecological partition, determining the target species abundance information corresponding to the target ecological partition from the initial species abundance information, wherein the target ecological partition is any one of multiple ecological partitions in the study area; S3, based on the environmental data, determining the target environmental stress factors related to the target group. S4. Based on the target species abundance information and the target environmental stress factor, iteratively solve the response coefficients of the pre-constructed multi-species occupation model to obtain the environmental stress effect quantity corresponding to each species in the target group. The multi-species occupation model adopts a hierarchical Bayesian framework, the target species abundance information is used as the response variable of the multi-species occupation model, and the target environmental stress factor is used as the explanatory variable of the multi-species occupation model. S5. Based on the environmental stress effect quantity corresponding to each species in the target group, determine the target environmental stress indicator species list corresponding to the target ecological zone.

[0028] This embodiment achieves a fundamental shift in indicator species screening from qualitative judgment based on expert experience to quantitative screening based on models by organically combining multi-group aquatic organism monitoring with a multi-species occupation model. By separating the true occupation probability and detection probability of a species within the model framework, estimation bias caused by missed detections in sampling is effectively eliminated, and the output environmental stress effect quantity can accurately reflect the true response direction and intensity of each species to environmental stress. Based on this, the final list of target environmental stress indicator species has the technical advantages of being objective, repeatable, and quantifiable, and can be directly used for watershed ecological health diagnosis and environmental stress tracing.

[0029] In one embodiment, step S5 specifically includes the following steps A1 to A3: A1. Based on a preset species detection probability threshold and the results of sorting environmental stress effect magnitudes by absolute value, an initial list of environmental stress indicator species is obtained from the target group; A2. The environmental stress response significance of the indicator species in the initial list of environmental stress indicator species is verified to obtain the response significance level of each indicator species; A3. Indicator species with response significance levels less than a preset threshold are selected to obtain the target list of environmental stress indicator species.

[0030] This embodiment employs a two-stage screening mechanism of preliminary selection and verification. While ensuring that candidate indicator species have sufficient regional representativeness and response intensity, it further eliminates species with unstable responses through significance verification, ensuring that each indicator species in the final list has a statistically significant stress response relationship. This effectively improves the reliability and monitoring accuracy of indicator species screening results.

[0031] In one embodiment, step A2 specifically includes the following steps A21 and A22: A21, performing multiple linear regression analysis on the indicator species in the initial environmental stress indicator species list to obtain the regression coefficients and significance test values ​​corresponding to each indicator species; A22, obtaining the response significance level of each indicator species based on the regression coefficients and significance test values.

[0032] This embodiment employs multiple linear regression analysis as an external validation method independent of the multi-species occupation model. By calculating the regression coefficients (representing the direction and intensity of the response) and significance test values ​​(representing whether the response is significant) of each indicator species, statistical evidence for cross-validation is provided for indicator species selection. The indicator species retained after successful validation demonstrate statistically significant environmental stress response relationships, ensuring the reliability of the final indicator species list.

[0033] In one embodiment, the multi-species occupancy model includes an occupancy sub-model and a detection sub-model. The formula for the occupancy sub-model is as follows: , , , in, This represents the actual state of existence of species i at location j; This represents the actual occupancy probability of species i at location j; This indicates that it follows a Bernoulli distribution; This represents a vector of environmental covariates related to the occupancy of species i. express transpose; This represents the response coefficient of species i to environmental stress; (·)express Join function; This represents the mean of the response coefficients for all species. express The covariance matrix, This indicates that the distribution follows a normal distribution; The formula corresponding to the detection sub-model is as follows: , ,

[0034] in, This indicates whether species i was detected in the k-th replicate at the j-th location; This represents the probability of species i being detected in the kth repeated instance at the j-th location; This represents a vector of covariates that influence the probability of detecting species i at the j-th location and the k-th repetition during the detection process. express transpose; The probability coefficient for detecting species i is represented by . This represents the mean of the detection probability coefficients for all species; express The covariance matrix.

[0035] This embodiment describes the binary state of species presence and detection results using a Bernoulli distribution, maps probabilities to linear space using a Logit join function, and regularizes response coefficients using a normal prior distribution. These three elements work together to achieve robust estimations of the species' true occupancy probability, detection probability, and environmental response coefficients. The hierarchical Bayesian framework enables information sharing among species, improves the estimation accuracy for small sample species, and provides a reliable quantitative basis for indicator species screening.

[0036] In one embodiment, obtaining initial species abundance information of multiple aquatic groups within the study area in step S1 includes: collecting environmental DNA samples at various sampling points within the study area; performing PCR amplification on the environmental DNA samples using multiple primer combinations to obtain amplification products; performing high-throughput sequencing on the amplification products to obtain sequencing data; performing bioinformatics analysis on the sequencing data, comparing it with a reference database to complete species annotation, and obtaining annotation results; and calculating the abundance of each species at each sampling point based on the annotation results to obtain initial species abundance information.

[0037] As a preferred embodiment, the multi-primer combination includes at least two of Mt12S, Mt16S, MtCOI, 18S, and trnL. Mt12S and Mt16S are mitochondrial genes, which evolve rapidly and are suitable for distinguishing closely related species; they are commonly used markers for fish eDNA monitoring. MtCOI is the mitochondrial cytochrome c oxidase subunit I gene, a standard gene for animal DNA barcoding. 18S is a nuclear gene encoding the small ribosomal subunit RNA; its sequence is conserved, making it suitable for broad-spectrum analysis at higher taxonomic levels. trnL is a chloroplast gene, one of the core markers for terrestrial plant DNA barcoding.

[0038] This embodiment employs environmental DNA technology for aquatic organism monitoring, offering the advantage of non-invasive sampling and no disturbance to the ecosystem. Multiple primer combinations can simultaneously amplify multiple groups, including waterbirds, fish, plankton, benthic animals, and aquatic plants, enabling joint monitoring of multiple groups and avoiding group bias or omissions caused by single primers. High-throughput sequencing can obtain massive amounts of data, improving species detection rates. This embodiment covers different trophic levels of the food chain and adapts to different habitat types, including water bodies, water columns, substrates, and riparian zones, allowing different groups to complement each other's response mechanisms to environmental stress, thus improving the comprehensiveness and accuracy of stress tracing.

[0039] In one embodiment, step S4 specifically includes the following steps B1 and B2: B1, standardizing the target species abundance information into a binary matrix, where each element in the binary matrix is ​​a first value, indicating that the species corresponding to that element was detected at the sampling point, and each element in the binary matrix is ​​a second value, indicating that the species corresponding to that element was not detected at the sampling point; B2, iteratively solving the response coefficients of the pre-constructed multi-species occupancy model based on the binary matrix and the target environmental stress factor to obtain the environmental stress effect quantity corresponding to each species in the target group.

[0040] In this embodiment, the first value is preferably 1, and the second value is preferably 0. Converting species abundance information into a detected / undetected binary form is because multi-species occupancy models use whether a species is detected as an observation variable to infer the true occupancy status, and the binary matrix can be directly used as the model's response variable input. After data cleaning, species with excessively low detection rates (e.g., below 2%) and duplicate samples are removed, effectively correcting sampling bias. The standardized binary matrix is ​​matched with an environmental stress database to form a unified dataset required for model operation. This embodiment, by converting species abundance information into a binary response variable acceptable to the model and integrating it with the explanatory variables, provides standardized input data for subsequent iterative solutions to model parameters. As an example, the MCMC (Markov Chain Monte Carlo) algorithm can be used for iterative solutions.

[0041] In one embodiment, the method of this application further includes: dividing the study area into multiple ecological zones based on the geographical floristics and ecosystem type of the study area.

[0042] This embodiment adapts to the spatial heterogeneity of different geographical regions through ecological zoning, overcoming the shortcomings of traditional methods that are only applicable locally and difficult to promote across the entire region. By using zoning modeling, the selection of indicator species for each ecological zone is tailored to the actual species composition and environmental characteristics of that zone, realizing a standardized selection framework applicable to different habitat ecosystems in large-scale watersheds, and meeting the differentiated needs of ecological monitoring across the entire region.

[0043] In one embodiment, step S3 specifically includes the following steps C1 and C2: C1, extracting candidate environmental stress factors from environmental data; C2, performing multicollinearity diagnosis on the candidate environmental stress factors, and identifying environmental stress factors with independent influence as target environmental stress factors.

[0044] This embodiment eliminates the multicollinearity effect among candidate environmental stress factors by performing multicollinearity diagnosis, thus avoiding distortion in model parameter estimation. The target environmental stress factors retained after screening each have independent explanatory power, improving the reliability of causal inference between indicator species and stresses, while reducing the dimensionality of model input variables and lowering the risk of overfitting.

[0045] In one embodiment, step C2 specifically includes the following steps C21 and C22: C21, calculating the variance inflation factor (VIF) value of each candidate environmental stress factor; C22, determining the candidate environmental stress factors whose variance inflation factor value is less than or equal to a preset factor threshold (preferably 5) as the target environmental stress factors.

[0046] This embodiment uses the variance inflation factor (VIF) test as a specific method for diagnosing multicollinearity. Variables with a VIF value less than or equal to 5 are retained, and variables with a VIF value greater than 5 are removed. This threshold is a recognized standard in the field of statistics and has objectivity and repeatability.

[0047] According to one embodiment of this application, an electronic device is provided, such as... Figure 9 As shown, the electronic device includes a processor 110, a communication interface 120, a memory 130, and a communication bus 140. The processor 110 is communicatively connected to the communication interface 120 and the memory 130 via the communication bus 140.

[0048] Processor 110 is the core computing and control unit of the electronic device, and can adopt a heterogeneous computing architecture of CPU and GPU. The CPU is responsible for serial logic operations such as data loading, flow control, sampling point management, environmental stress factor screening, and indicator species list sorting; the GPU is responsible for large-scale matrix operations, probability density calculation of Bernoulli distribution and normal distribution, batch mapping of Logit connection function, and parallel updating of high-dimensional response coefficient vector in the MCMC iterative solution process of the multi-species occupation model. When executing the computer program, processor 110 implements the environmental stress indicator species screening method based on the multi-species occupation model described above.

[0049] The communication interface 120 is used to realize data interaction between the electronic device and the external environment, supporting wired communication (such as Ethernet interface, USB interface) and wireless communication (such as Wi-Fi module, 4G / 5G module), and transmitting the acquired data to the processor 110 or memory 130 through the communication bus 140. The communication interface 120 is also used to output the list of indications generated by the processor 110 to an external monitoring platform or terminal device.

[0050] The memory 130 is used to store computer programs, species abundance data, environmental stress data, model parameters, and intermediate data during the calculation process. Among them, the model parameters include the occupancy probability, detection probability, environmental stress response coefficient vector, hyperparameters, and the detection / non-detection binary matrix corresponding to each species during the MCMC iteration process.

[0051] The communication bus 140 is used to connect the processor 110, the communication interface 120 and the memory 130 to realize data transfer and control signal transmission among the three.

[0052] When the electronic device executes the computer program, the processor 110 acquires the initial species abundance information and environmental data of multiple groups of aquatic organisms in the study area through the communication interface 120, loads the model parameters and preprocessed target species abundance matrix and target environmental stress factor data from the memory 130, and sequentially performs data preprocessing, ecological zoning matching, VIF multicollinearity diagnosis, multi-species occupation model MCMC iterative solution, environmental stress effect output and ranking, indicator species preliminary list generation, and multiple linear regression significance verification. Finally, it outputs the target environmental stress indicator species list classified by stress type and ecological zoning through the communication interface 120 or the display device.

[0053] According to this embodiment, the automated execution of the environmental stress indicator species screening method described in this application is realized through the coordinated operation of the hardware architecture of the electronic device and the computer program. At the same time, it supports the persistent storage and fast loading of multi-species occupation model parameters in the device, and accelerates the batch update and probability density calculation of multi-species response coefficient vectors in MCMC iteration through GPU parallel computing, which significantly improves the computational efficiency and real-time performance of environmental stress indicator species screening in multi-group and multi-ecological-region scenarios.

[0054] The following describes a specific embodiment of this application, which takes the Hanjiang River Basin (including the main stream, major tributaries, and Danjiangkou Reservoir area) as the target research area.

[0055] (1) Overview and ecological zoning of the study area The Han River basin spans multiple geographical regions, with its main stream flowing from the upper mountainous areas through the middle hilly areas to the lower plains. Its ecosystem types encompass various categories, including mountain streams, hilly rivers, plain channels, and reservoir areas. First, based on geographical regions and ecosystem types, the Han River basin was divided into several ecological zones. Specifically, according to the basin's topography, water system structure, and biogeographical characteristics, the study area was divided into an upper mountain ecological zone, a middle hilly ecological zone, a lower plain ecological zone, and a reservoir area ecological zone. Significant differences exist in the biological community composition and environmental stress characteristics of each zone: the upper mountainous areas have better river connectivity and less human disturbance; the middle hilly areas have intensive agricultural activities; the lower plains face prominent urbanization and eutrophication problems; and the reservoir area faces multiple stresses, including altered hydrological conditions and wetland shrinkage. Subsequent indicator species screening was conducted for each zone.

[0056] (2) Aquatic organism monitoring and initial species abundance information acquisition Sampling sites were established within each ecological zone, totaling 63 sites in the main stream, major tributaries, and Danjiangkou Reservoir area, covering different river sections and habitat types. Three replicate samplings were conducted at each site, resulting in a total of 189 eDNA samples. One liter of surface water was collected from each sample and filtered through a 0.45 μm filter to enrich eDNA. The samples were then stored at low temperatures and transported to the laboratory. DNA extraction was performed on the collected samples, and PCR amplification was conducted using a multi-primer combination of Mt12S, Mt16S, MtCOI, 18S, and trnL to simultaneously obtain DNA fragments from five major groups: waterbirds, fish, benthic animals, plankton, and aquatic plants. Libraries were constructed from the amplified products and sequenced. After quality control, redundancy removal, and clustering, the raw data were compared with the National Center for Biotechnology Information (NCBI) and the Hanjiang River Basin Local Species Reference Database to complete species annotation. The annotation and quantitative results were organized and an initial species abundance matrix was constructed, with sampling sites as rows, species as columns, and species abundance values ​​as cells. This matrix contains all detected species and their abundance information for all 63 sampling sites and five major taxa.

[0057] (3) Construction of environmental stress database Water quality data (water temperature, pH, dissolved oxygen, conductivity, total nitrogen, total phosphorus, ammonia nitrogen, permanganate index, etc.), hydrological data (flow rate, velocity, water level, etc.), land use data (cultivated land, construction land, forest land, water area proportion, etc.), and remote sensing data were collected from 63 sampling points to establish an environmental and spatial basic database for the study area. Candidate environmental stress factors were extracted from the environmental data, including cultivated land proportion, pesticide pollution risk index, construction land proportion, population density, urban light index, total nitrogen, total phosphorus, river connectivity index, maximum monthly average temperature, floodplain area change rate, and forest area reduction rate. The VIF value of each candidate environmental stress factor was calculated, and highly collinear variables with VIF values ​​greater than 5 were removed. Finally, seven target environmental stress factors with independent impacts on aquatic biological communities were selected, including agricultural activities, urbanization, eutrophication, river fragmentation, global warming, wetland shrinkage, and forest loss. All environmental and stress data were spatially matched with sampling points, unified in a projected coordinate system, and normalized to construct an environmental stress database for the Hanjiang River Basin. This step ensured low collinearity among the environmental stress factors entering the model, avoided distortion in model parameter estimation, and provided standardized explanatory variable data for subsequent screening of indicator species in each ecological zone.

[0058] (4) Extraction of target species abundance matrix and data preprocessing For the upstream mountainous ecological zone, fish species abundance data at each sampling point within the zone are extracted from the initial species abundance matrix to serve as the target species abundance matrix for the corresponding target taxa. Based on this target species abundance matrix, species with a detection rate below 2% and duplicate samples are removed to correct sampling bias. The abundance data in the cleaned species abundance matrix are standardized and converted into a species detection / non-detection binary matrix (1 for presence, 0 for absence), which serves as the response variable for subsequent model analysis. The species detection / non-detection matrix is ​​matched with an environmental stress database to form a unified dataset required for model operation. The data preprocessing method for other ecological zones (midstream hills, downstream plains, and reservoir area) is the same, except that the selection of target taxa can be adjusted according to the ecological characteristics and monitoring needs of each zone—for example, fish and plankton can be monitored simultaneously in the reservoir area, while phytoplankton and benthic animals can be focused on in the downstream plains. The following detailed explanation of the model operation is based on the fish target taxa in the midstream hilly ecological zone as an example.

[0059] (5) Construction and iterative solution of multi-species occupation model A multi-species occupation model was constructed using the `spOccupancy` package in R. A binary response variable was created using a standardized detection / non-detection matrix of fish target species abundance information, and explanatory variables were environmental stress factors related to the target zoning (agricultural activities, urbanization, river fragmentation, etc.). The model prior used an uninformative prior distribution. The MCMC iteration parameters were: 3000 total iterations, 1000 burn periods, a bookkeeping interval of 50, and parallel computation on four independent chains to ensure parameter convergence and stability. The model includes an occupation sub-model and a detection sub-model. The formula corresponding to the occupancy sub-model is as follows:

[0060] in, This represents the actual existence status of species i at location j (1 = existence, 0 = non-existence); This represents the actual occupancy probability of species i at location j; This represents the vector of environmental covariates (i.e., target environmental stress factors) associated with the occupation of species i. This represents the response coefficient of species i to environmental stress; This represents the mean of the response coefficients for all species. express The covariance matrix.

[0061] The formula corresponding to the detection sub-model is as follows:

[0062] in, Indicates whether species i was detected in the kth repetition at the jth location (1 = detected, 0 = not detected); This represents the probability of species i being detected in the kth repeated instance at the j-th location; This represents a vector of covariates (such as survey duration, weather, survey time, etc.) that affect the probability of species i being detected for the kth time at the jth location during the detection process. The probability coefficient for detecting species i is represented by . This represents the mean of the detection probability coefficients for all species; express The covariance matrix.

[0063] Run the model and evaluate its goodness of fit. Figure 2This figure shows the goodness-of-fit test results of the multi-species occupation model in this embodiment. The horizontal axis represents the true values ​​of species detected by field sampling, and the vertical axis represents the fitted predicted values ​​output by the hierarchical Bayesian multi-species occupation model. The black diagonal line in the figure is the baseline for perfect fit. The orange scatter points are the predicted samples after correcting the detection error using the multi-species occupation model of this application, and the light blue scatter points are the predicted samples of the traditional statistical model that does not distinguish between the detection and occupation processes. Both types of scatter points are concentrated along the baseline diagonal, proving that both models have basic fitting capabilities. However, the orange corrected samples are closer to the diagonal and have a smaller deviation. The light blue traditional model samples are mostly distributed below the diagonal, indicating a significant underestimation bias. This intuitively verifies that the multi-species occupation model of this application can effectively eliminate the numerical underestimation problem caused by sampling omissions by hierarchically separating the true occupation process from the field detection process. The prediction accuracy is significantly better than that of traditional methods, providing goodness-of-fit data support for this application to accurately output the unbiased stress effect and ensure the reliability of the indicator species screening results. The model's Fit p-value is 0.3324, indicating that the model fits well. After iterative solution of the model, the true occupancy probability and environmental stress response coefficient of each fish species in the midstream hilly ecological zone are output.

[0064] Figure 3 The box plot shows the effect sizes of environmental stress factors at the fish community level, illustrating the direction and dispersion of responses of all fish species included in the analysis to each stress factor. The horizontal axis represents the stress effect values ​​output by the multi-species occupancy model, with the vertical dashed line 0 as the boundary. An effect value less than 0 indicates a decrease in species abundance and a negative response as stress intensifies, while an effect value greater than 0 indicates an increase in species occupancy probability and a positive response as stress intensifies. The vertical axis lists six core environmental stress factors: pesticide risk index, urbanization, flow, connectivity, farmland area ratio, and maximum monthly average temperature. The boxes in the figure represent the quartile distribution and median of the effect values ​​of all species, and the discrete blue dots represent the independent effect values ​​of individual species. The distribution pattern shows that the species effect value corresponding to the maximum monthly average temperature is generally distributed in the negative range, indicating that most species have a negative response to climate warming. The species effect values ​​of pesticide risk, connectivity, urbanization, and farmland area ratio are generally located to the right of 0, indicating that most species have a positive tolerance to these stresses. The effect value corresponding to river flow spans both positive and negative ranges, with significant differences in hydrological responses among different species. This figure visualizes the direction and dispersion of the responses of all species to multiple stresses, and intuitively distinguishes between sensitive negative response species and tolerant positive response species.

[0065] (6) Determination of the preliminary list of indicator species for each ecological zone Based on the preset species detection probability threshold (e.g., 20%) and the results after sorting the environmental stress effect by absolute value, fish species with a detection probability greater than 20% in the midstream hilly ecological zone are retained; species with an absolute value of the target stress effect ranking in the top 10% of their respective groups are selected to obtain a preliminary list of indicator species for the fish groups in the midstream hilly ecological zone for seven types of environmental stress. Similarly, for the upstream mountainous ecological zone, the downstream plain ecological zone, and the reservoir ecological zone, steps (4) to (6) above are repeated for their respective ecological characteristics and target groups (e.g., upstream mountainous areas focus on river connectivity indicator species, downstream plains focus on eutrophication indicator species, and reservoir areas focus on water level change and wetland shrinkage indicator species) to obtain a preliminary list of indicator species for each zone. Figure 4 A scatter plot ranking the effect sizes of fish species on river connectivity stress is presented, showing the ranking results of fish species effect sizes on river connectivity stress in the midstream hilly ecological zone. The vertical axis lists all candidate fish indicator species names selected by the method in this application, and the horizontal axis represents the species connectivity effect value output by the multi-species occupancy model. The larger the value, the stronger the positive response of the species to the increase in river connectivity and the more sensitive it is to river fragmentation stress. The effect values ​​of all candidate fish species are distributed in the positive range above 0.6, indicating that the fish species in this zone have a significantly increased occupancy probability as river connectivity increases. Among them, *Brachys simonii* and *Brachys sichuanensis* have the highest effect values ​​and the strongest response to connectivity changes, and were given priority in the initial selection list.

[0066] (7) Merging, verifying and outputting the final list of indicators The initial lists of environmental stress indicator species from various ecological zones and populations were merged, and duplicate species were removed to obtain a set of candidate indicator species for the Hanjiang River Basin. Multiple linear regression was used to verify each candidate indicator species, retaining those with significant environmental stress responses (P < 0.05, where P is the significance test value of the regression coefficient).

[0067] Figure 5 This is a fitting validation plot for the occupancy probability of *Clerodendrum chinense* as a function of river fragmentation. The horizontal axis represents the river fragmentation index; a higher value indicates poorer river connectivity and a higher degree of fragmentation. The vertical axis represents the occupancy probability of this species, which is the true probability of species presence output by the multi-species occupancy model in this application. The scatter plots represent the measured data at each sampling point, the solid line is the multiple linear regression fitting line, and the shaded area represents the 95% confidence interval of the fitted value. The scatter plot distribution shows that in the low fragmentation range (40-60), the true occupancy probability of this species is generally below 0.3; as fragmentation increases, the occupancy probability continuously increases, reaching approximately 0.95 when the fragmentation approaches 100. The multiple linear regression fitting results show that the response coefficient estimate is Estimate = 0.16, indicating that this species has a significant response to river fragmentation stress; the coefficient of determination... =0.44, indicating that river fragmentation can explain 44% of the species' distribution variation; P<0.001 (P<0.05), indicating that *Hemiberlesia lataniae* exhibits a significant avoidance response to river fragmentation stress and can be considered a sensitive indicator species for river fragmentation, thus it should be retained. The 95% confidence interval narrows synchronously with increasing fragmentation, indicating high sample aggregation and strong result stability under high fragmentation gradients. This verifies that the multi-species occupancy model in this application can accurately separate and detect noise, quantitatively characterize the positive response pattern of species to river fragmentation stress, and provide a reliable quantitative basis for indicator species screening.

[0068] Figure 6 This is a scatter plot showing the response of sensitive indicator species numbers under river fragmentation stress gradients. The horizontal axis represents the river fragmentation index, and the vertical axis represents the number of sensitive indicator species (i.e., the number of fish indicator species sensitive to river fragmentation). The scatter plots represent the measured data at each sampling point, the solid line is the multiple linear regression fitting line, and the shaded area represents the 95% confidence interval of the fitted values. The scatter plot distribution and the fitted trend line show that the number of sensitive indicator species decreases as the river fragmentation index increases. In the low river fragmentation range (horizontal axis values ​​approach 0), the average number of sensitive indicator species at each sampling point is close to 2. As river fragmentation continues to increase and river connectivity continues to degrade, the number of sensitive indicator species at each sampling point decreases synchronously. When the fragmentation reaches 0.8, the average number of sensitive indicator species drops to below 0.5. The multiple linear regression results show that... =0.18, P=0.003 (P<0.05), indicating that although river fragmentation explains 18% of the variation in the number of sensitive indicator species, the response relationship is statistically significant, demonstrating that the selected assemblage of sensitive fish indicator species can effectively respond to river fragmentation stress, thus passing the applicability verification. Furthermore, the 95% confidence interval widens slowly with increasing fragmentation, and the sample dispersion is even higher in the high fragmentation interval, with the number of sensitive indicator species dropping directly to 0 at some locations. This proves that river fragmentation significantly reduces the distribution of sensitive indicator species, and these species can accurately characterize the intensity of river connectivity impairment stress.

[0069] Figure 7 and Figure 8 These are validation plots of environmental stress indicator species for phytoplankton sensitive and tolerant species, respectively, indicating eutrophication. The horizontal axis of both plots represents the eutrophication stress gradient, indicating the degree of eutrophication in the water body from low to high within the study area; the vertical axis represents the relative abundance of sensitive and tolerant phytoplankton species, reflecting the proportion of these groups in the community; the discrete points in the plots are the measured data at each sampling point, the solid line is the linear regression fitting curve, and the shaded area is the 95% confidence interval of the regression.

[0070] Figure 7 The linear regression result is =0.26, P=0.009 (P<0.05), indicating a highly significant negative correlation between the relative abundance of sensitive species and the eutrophication stress gradient, with the fitted model explaining 26%. The fitted curve shows a continuous downward trend, indicating that as the eutrophication stress level on the horizontal axis increases, the relative abundance of sensitive groups decreases significantly. In the low eutrophication range (around 0.3 on the horizontal axis), the relative abundance of sensitive species remains at a relatively high level of 0.3~0.75. When the eutrophication gradient increases to 0.6, the average relative abundance of sensitive species drops below 0. This shows that eutrophication pollution significantly inhibits the survival and reproduction of sensitive indicator species. The stronger the stress, the lower the proportion of sensitive species in the community, and this group can serve as a negative indicator of eutrophication. The 95% confidence interval continuously widens with the increase of the eutrophication gradient, indicating that the data dispersion of sampling points in the high-pollution range is higher, and the community structure of sensitive species fluctuates more.

[0071] Figure 8 The linear regression results were R²=0.13, P=0.04, and the trend line showed a significant positive correlation. As the degree of eutrophication stress increased, the relative abundance of tolerant species generally showed an upward trend. However, R²=0.13 indicates that the model can only explain 13% of the data variation, and the sample dispersion is high. This suggests that eutrophication is a significant factor affecting the relative abundance of tolerant species, but other environmental factors also jointly regulate the changes in the abundance of tolerant species. This result confirms that the tolerant indicator species screened by this invention have a positive response to eutrophication stress and can be used to characterize the level of eutrophication stress in water bodies.

[0072] All four sets of validation results showed that the selected indicator species had statistically significant response relationships to the corresponding environmental stresses, thus passing the applicability validation. The final output was a list of environmental stress indicator species for each ecological zone and species group (waterbirds, fish, plankton, benthic animals, and aquatic plants) in the Hanjiang River Basin. The list was categorized by stress type (agricultural activities, urbanization, eutrophication, river fragmentation, climate warming, wetland shrinkage, and forest loss) and ecological zone, directly serving the diagnosis of the basin's ecological health and the tracing of environmental stress sources.

[0073] This embodiment implements the method of this application in four ecological zones of the Hanjiang River Basin (upstream mountainous area, midstream hilly area, downstream plain, and reservoir area) to systematically verify the technical effects of this application: First, the VIF test is used to screen out seven target environmental stress factors with independent influence from environmental data, eliminating collinearity among factors and avoiding distortion in model parameter estimation; Second, the multi-species occupation model iteratively solves the eDNA data from 63 sampling points, five major taxa, and multiple repeated samplings. The model fit test p-value is 0.3324, indicating that the model fits well and the output environmental stress effect size can accurately reflect the true response direction and intensity of each species to different stresses; Third, a dual screening mechanism of species detection probability threshold (>20%) and effect size absolute value ranking (top 10%) ensures that the initially selected indicator species are both regionally representative and have strong environmental response; Finally, multiple linear regression is used to verify each candidate indicator species. Figures 5 to 8 The verification results (all verification P-values ​​are less than 0.05) indicate that all selected indicator species passed the applicability verification, and the final list of indicator species has statistically reliable stress response relationships. This embodiment fully demonstrates that the method of this application can achieve standardized screening of indicator species with full coverage, regional adaptation, multi-group compatibility, and simultaneous identification of multiple stresses in complex watershed scenarios.

[0074] In summary, the indicator species screening method provided in this application, through the organic combination of aquatic multi-group monitoring and species dominance models, realizes the transformation of indicator species screening from empirical qualitative to model quantitative methods. It can efficiently, accurately, and objectively screen indicator species applicable to environmental stresses at the global scale, while also considering multi-group compatibility and spatial applicability. This application fully utilizes the response relationship between aquatic organisms and environmental stresses, combined with the quantitative capabilities of species dominance models in distribution and response, to construct a standardized, repeatable, and scalable indicator species screening system. This system can provide stable and reliable technical support for aquatic ecological health diagnosis, environmental stress tracing, biological integrity assessment, and ecological environment management, meeting the needs of routine, high-precision, and large-scale ecological monitoring and evaluation applications in watersheds and reservoirs.

[0075] The above description is merely a preferred embodiment of this application and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in the embodiments of this application is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described inventive concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features with similar functions disclosed in the embodiments of this disclosure.

Claims

1. A method for screening environmental stress indicator species based on a multi-species occupancy model, characterized in that, include: To obtain initial species abundance information of multiple groups of aquatic organisms in the study area and environmental data of the study area; Based on the target taxa corresponding to the target ecological partition, the target species abundance information corresponding to the target ecological partition is determined from the initial species abundance information. The target ecological partition is any one of multiple ecological partitions within the study area. Based on the environmental data, target environmental stress factors related to the target group are determined; Based on the target species abundance information and the target environmental stress factor, the response coefficients of the pre-constructed multi-species occupation model are iteratively solved to obtain the environmental stress effect quantity corresponding to each species in the target group. The multi-species occupation model adopts a hierarchical Bayesian framework. The target species abundance information is used as the response variable of the multi-species occupation model, and the target environmental stress factor is used as the explanatory variable of the multi-species occupation model. Based on the environmental stress effect of each species in the target taxa, a list of target environmental stress indicator species corresponding to the target ecological zone is determined.

2. The method according to claim 1, characterized in that, The determination of the target environmental stress indicator species list corresponding to the target ecological zone based on the environmental stress effect magnitude of each species in the target taxa includes: Based on the preset species detection probability threshold and the results of sorting environmental stress effect magnitudes by absolute value, an initial list of environmental stress indicator species is obtained from the target group. The environmental stress response significance of the indicator species in the initial list of environmental stress indicator species is verified to obtain the response significance level of each indicator species; Select the indicator species whose response significance level is less than a preset threshold from among the indicator species to obtain the list of indicator species for the target environmental stress.

3. The method for screening environmental stress indicator species according to claim 2, characterized in that, The step of verifying the significance of environmental stress responses of the indicator species in the initial list of environmental stress indicator species to obtain the response significance level of each indicator species includes: Multiple linear regression analysis was performed on the indicator species in the initial list of environmental stress indicator species to obtain the regression coefficients and significance test values ​​for each indicator species; Based on the regression coefficients and significance test values, the significance level of the response of each indicator species is obtained.

4. The method for screening environmental stress indicator species according to claim 1, characterized in that, The multi-species occupancy model includes an occupancy sub-model and a detection sub-model. The formula corresponding to the occupancy sub-model is as follows: , , ; in, This represents the actual state of existence of species i at location j; This represents the actual occupancy probability of species i at location j; This indicates that it follows a Bernoulli distribution; This represents a vector of environmental covariates related to the occupancy of species i. express transpose; This represents the response coefficient of species i to environmental stress; (·)express Join function; This represents the mean of the response coefficients for all species. express The covariance matrix, This indicates that the distribution follows a normal distribution; The formula corresponding to the detection sub-model is as follows: , , ; in, This indicates whether species i was detected in the k-th replicate at the j-th location; This represents the probability of species i being detected in the kth repeated instance at the j-th location; This represents a vector of covariates that influence the probability of detecting species i at the j-th location and the k-th repetition during the detection process. express transpose; The probability coefficient for detecting species i is represented by . This represents the mean of the detection probability coefficients for all species; express The covariance matrix.

5. The method according to claim 1, characterized in that, The acquisition of initial species abundance information of multiple groups of aquatic organisms within the study area includes: Environmental DNA samples were collected at various sampling points within the study area. The environmental DNA sample was amplified by PCR using a multi-primer combination to obtain the amplification product. The amplified products were subjected to high-throughput sequencing to obtain sequencing data; Bioinformatics analysis was performed on the sequencing data, and species annotation was completed by comparing it with a reference database to obtain the annotation results; The abundance of each species at each sampling point is calculated based on the annotation results to obtain the initial species abundance information.

6. The method according to claim 5, characterized in that, The step of iteratively solving the response coefficients of a pre-constructed multi-species occupancy model based on the target species abundance information and the target environmental stress factor to obtain the environmental stress effect quantity corresponding to each species in the target group includes: The abundance information of the target species is standardized into a binary matrix. When each element in the binary matrix is ​​a first value, it means that the species corresponding to that element is detected at that sampling point. When each element in the binary matrix is ​​a second value, it means that the species corresponding to that element is not detected at that sampling point. The response coefficients of the pre-constructed multi-species occupation model are iteratively solved based on the binary matrix and the target environmental stress factor to obtain the environmental stress effect quantity corresponding to each species in the target group.

7. The method according to claim 1, characterized in that, The method further includes: Based on the geographical distribution and ecosystem type of the study area, the study area is divided into multiple ecological zones.

8. The method according to claim 1, characterized in that, The step of determining the target environmental stress factors related to the target group based on the environmental data includes: Extract candidate environmental stress factors from the environmental data; Multicollinearity diagnosis is performed on the candidate environmental stress factors, and environmental stress factors with independent influence are identified as the target environmental stress factors.

9. The method according to claim 8, characterized in that, The step of performing multicollinearity diagnosis on the candidate environmental stress factors, and identifying environmental stress factors with independent influence as the target environmental stress factors, includes: Calculate the variance inflation factor value for each of the candidate environmental stress factors; Candidate environmental stress factors whose variance inflation factor value is less than or equal to a preset factor threshold are determined as the target environmental stress factors.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable by the processor, characterized in that, The processor, when executing the computer program, implements the environmental stress indicator screening method as described in any one of claims 1 to 9.