Method and system for analyzing ecological quality trend of crested ibis habitat

Abstract

The invention discloses a crested ibis habitat ecological quality trend analysis method and system, and relates to ecological quality monitoring. The method comprises the following steps: S1, constructing an intelligent sensing network, synchronously obtaining multi-source data of a habitat, identifying activity events of crested ibis, and generating a multi-dimensional habitat parameter table; s2, collecting environmental samples, and generating a microbial functional gene abundance matrix through metagenome sequencing and bioinformatics analysis; s3, taking the activity events of the crested ibis as behavior tags, and generating habitat function health indexes by coupling the parameter table and the matrix training machine learning prediction model; s4, performing spatial interpolation and trend analysis based on the habitat function health index to generate an ecological quality space-time evolution graph; and S5, based on the ecological quality space-time evolution graph, performing quantitative analysis by using a spatial differentiation statistical model, and generating a trend analysis report. By fusing multi-source data, real-time dynamic evaluation of habitat ecological quality and quantitative analysis of driving factors are realized, and a direct decision basis is provided for accurate protection.

Description

Technical Field

[0001] This invention relates to ecological quality monitoring, specifically a method and system for analyzing the ecological quality trends of crested ibis habitats. Background Technology

[0002] As a Class I protected wild animal in China, the survival and reproduction of the crested ibis population are highly dependent on the ecological quality of its habitat. However, continuous and precise monitoring and trend analysis of the ecological quality of the crested ibis habitat are a scientific prerequisite for effective conservation management. Currently, technical practices in this field mainly rely on the following approaches: Traditional field surveys and habitat assessment methods are fundamental, relying mainly on manual on-site investigations and recording of information such as vegetation, water bodies, and human disturbance, combined with expert experience to construct assessment models. While the data obtained in this way is specific, it is labor-intensive and slow to update, making it difficult to achieve large-scale, periodic dynamic monitoring, and the assessment results are easily influenced by subjective judgments.

[0003] Remote sensing and geographic information system (GIS) technologies can use satellite or aerial imagery to conduct macroscopic monitoring of land use and vegetation cover over large areas. However, the information obtained by this method is mostly limited to the surface landscape structure level. It is difficult to accurately capture micro-habitat characteristics that are crucial for the survival of crested ibises, such as the dynamics of shallow water areas required for foraging and the abundance of food resources. It is also impossible to directly obtain key physicochemical parameters of water bodies and soil.

[0004] Environmental DNA technology is an emerging biomonitoring method that confirms the presence of target species by detecting genetic material in environmental samples. This technology is highly sensitive and particularly suitable for detecting the distribution of rare species such as the crested ibis. However, at present, its main function remains limited to species presence detection, failing to effectively correlate and quantitatively analyze species presence signals with the functional status of the ecosystem within the habitat, such as key ecological processes like material cycling efficiency and microbial metabolic activity.

[0005] Therefore, developing a method that can automatically integrate multi-source habitat data and achieve deep coupling analysis of ecological structure and function, thereby accurately quantifying the evolution trend of ecological quality and its intrinsic driving forces, has become an urgent technical need to improve the effectiveness of habitat protection and management for endangered species. Summary of the Invention

[0006] Based on the shortcomings of the prior art described above, the purpose of this invention is to provide a method and system for analyzing the ecological quality trend of crested ibis habitats, so as to solve the above-mentioned technical problems.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a method for analyzing the ecological quality trend of crested ibis habitat, comprising: S1: Construct an intelligent sensing network to synchronously acquire multi-source habitat data and identify crested ibis activity events, and generate a multi-dimensional habitat parameter table through edge computing fusion; S2: Collect environmental samples based on preset environmental locations, and generate a microbial functional gene abundance matrix through metagenomic sequencing and bioinformatics analysis; S3: Using crested ibis activity events as behavioral labels, coupled with a multidimensional habitat parameter table and a functional gene abundance matrix, a machine learning prediction model is trained to generate a habitat functional health index; S4: Based on the spatiotemporal sequence of habitat functional health index, perform spatial interpolation and trend analysis to generate a spatiotemporal evolution map of ecological quality; S5: Based on the spatiotemporal evolution map of ecological quality, quantitative analysis is performed using a spatial differentiation statistical model to generate a trend analysis report that clearly identifies the dominant driving factors of ecological quality changes.

[0008] The present invention is further configured such that S1 includes: A smart sensing network covering the target area is constructed, the smart sensing network including: an acoustic monitoring unit, a remote sensing acquisition unit, and an in-situ physicochemical monitoring unit; Raw environmental audio data is collected using an acoustic monitoring unit; Utilize remote sensing to acquire unit multispectral image data and thermal infrared radiation data; Raw test data of water and soil were collected using in-situ physicochemical monitoring units; The raw audio data, multispectral image data, thermal infrared radiation data, and raw monitoring data are processed and fused to construct a multidimensional habitat parameter table.

[0009] The present invention is further configured such that the processing and fusion include: The raw audio data is processed by time-domain and frequency-domain analysis. Soundscape index is extracted based on acoustic statistical analysis method. At the same time, crested ibis activity events are identified based on pattern recognition method. Acoustic parameters are constructed by combining soundscape index and specific biological sound source events. Radiometric correction and feature extraction were performed on multispectral image data. Vegetation-related spectral indices were generated based on spectral index calculation methods, and habitat type identification was completed based on image classification methods. The thermal infrared radiation data is processed and the land surface temperature is obtained based on the thermal infrared inversion model. Remote sensing parameters are constructed by combining spectral vegetation index, habitat classification and land surface temperature. Multidimensional habitat parameters are generated by performing time synchronization, spatial correlation, and edge-side fusion processing on acoustic parameters, remote sensing parameters, and physicochemical parameters.

[0010] The present invention further specifies that S2 includes: a sequencing data generation step and a matrix construction step.

[0011] The present invention is further configured such that the sequencing data generation step includes: Collect environmental samples according to the preset environmental sampling points; Total microbial DNA was extracted from environmental samples; The total DNA was sequenced using a shotgun sequencing strategy to obtain gene sequence data.

[0012] The present invention is further configured such that the matrix construction step includes: Quality control, splicing and assembly, and gene prediction are performed on gene sequence data to obtain a non-redundant set of gene sequences; The gene sequence set is compared with a pre-defined database of known functional genes to obtain functional gene annotation information; Based on the sequence alignment results, the frequency or reading of various functional genes in each sample was statistically analyzed, normalized, and a microbial functional gene abundance matrix was formed.

[0013] The present invention is further configured such that S3 includes: A fused feature set was constructed by coupling a multidimensional habitat parameter table and a functional gene abundance matrix. Acquire identified crested ibis activity events and use these events as behavioral labels; By using behavioral labels as supervision signals, a pre-defined machine learning prediction model is trained using a fused feature set. For the target area, input features are formed based on a multidimensional habitat parameter table and functional gene abundance, which are then input into a trained machine learning prediction model to output the corresponding habitat functional health index.

[0014] The present invention is further configured such that S4 includes: Spatiotemporal sequence data were constructed based on habitat functional health indices at multiple time points. Spatial interpolation is performed on the spatiotemporal sequence data to generate continuous spatial distribution surfaces at each time point; Based on the spatial distribution surface, the trend of each spatial location is calculated in the time series to obtain the mass change trend of each location; Based on the trend of quality change, a spatiotemporal evolution map of ecological quality is generated using map-making methods.

[0015] The present invention is further configured such that S5 includes: Spatial overlay and correlation of the spatiotemporal evolution map of ecological quality and the multidimensional habitat parameter table; Using a geographic detector model, we calculated the independent explanatory power of each habitat parameter for the spatial differentiation of ecological quality; Analyze the explanatory power of the interaction between any two habitat parameters on the spatial differentiation of ecological quality; Based on the ranking and numerical values ​​of independent and interactive explanatory power, the dominant driving factors and their corresponding influence strengths are determined; A trend analysis report is generated based on the dominant driving factors and their strength.

[0016] This invention also provides a system for analyzing the ecological quality trend of crested ibis habitat, the system comprising: Intelligent sensing and data fusion module: Constructs an intelligent sensing network, synchronously acquires multi-source habitat data and identifies crested ibis activity events, and generates a multi-dimensional habitat parameter table through edge computing fusion; Environmental genomics analysis module: Based on environmental samples collected from preset environmental locations, a microbial functional gene abundance matrix is ​​generated through metagenomic sequencing and bioinformatics analysis; Intelligent Health Index Assessment Module: Using crested ibis activity events as behavioral labels, coupling a multidimensional habitat parameter table with a functional gene abundance matrix, training a machine learning prediction model, and generating a habitat functional health index; Spatiotemporal dynamic map generation module: Based on the spatiotemporal sequence of habitat functional health index, spatial interpolation and trend analysis are performed to generate a spatiotemporal evolution map of ecological quality; Driving Factor Analysis and Report Generation Module: Based on the spatiotemporal evolution map of ecological quality, quantitative analysis is performed using a spatial differentiation statistical model to generate a trend analysis report that clearly identifies the dominant driving factors of ecological quality changes.

[0017] This invention provides a method and system for analyzing the ecological quality trend of crested ibis habitat. The method utilizes an intelligent sensing and data fusion module: S1: Constructing an intelligent sensing network to synchronously acquire multi-source habitat data and identify crested ibis activity events, and generating a multi-dimensional habitat parameter table through edge computing fusion; S2: Collecting environmental samples based on preset environmental locations, and generating a microbial functional gene abundance matrix through metagenomic sequencing and bioinformatics analysis; S3: Using crested ibis activity events as behavioral tags, coupling the multi-dimensional habitat parameter table and the functional gene abundance matrix, training a machine learning prediction model, and generating a habitat functional health index; S4: Based on the spatiotemporal sequence of the habitat functional health index, performing spatial interpolation and trend analysis to generate a spatiotemporal evolution map of ecological quality; S5: Based on the spatiotemporal evolution map of ecological quality, using a spatial differentiation statistical model for quantitative analysis, generating a trend analysis report that clearly identifies the dominant driving factors of ecological quality changes. The beneficial effects include: Achieving in-depth correlation and comprehensive evaluation of habitat information across multiple dimensions: By integrating macro-environmental remote sensing data, real-time acoustic biological monitoring data, in-situ physicochemical parameters, and micro-microbial functional gene data within a unified spatiotemporal framework, this approach overcomes the limitations of traditional methods, which suffer from single-dimensional assessment and fragmented data. It enables collaborative analysis and comprehensive evaluation of habitat structure, function, and species activity.

[0018] Achieving a leap from static assessment to dynamic intelligent diagnosis: By using machine learning models coupled with multi-source dynamic data, the generated habitat functional health index can reflect the dynamic changes in ecological quality in real time and quantitatively, overcoming the shortcomings of traditional methods that are lagging and static, and realizing intelligent perception and forward-looking early warning of ecological quality trends.

[0019] Provides quantitative and precise decision-making basis for protection and management: By using spatial differentiation models to quantitatively analyze the dominant driving factors and their contribution to changes in ecological quality, management decisions are elevated from "experience-based judgment" to "evidence-based policy implementation," providing direct and clear scientific basis for implementing specific protection measures such as precise habitat restoration, water level regulation, and disturbance elimination.

[0020] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings: Figure 1 A flowchart illustrating a method for analyzing the ecological quality trend of crested ibis habitat, as shown in an exemplary embodiment of the present invention; Figure 2 This is a schematic diagram illustrating the structure of a crested ibis habitat ecological quality trend analysis system, which is an exemplary embodiment of the present invention. Detailed Implementation

[0022] The embodiments of the present invention will be described below with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be understood that the preferred embodiments are only for illustrating the present invention and not for limiting the scope of protection of the present invention.

[0023] It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. Therefore, the drawings only show the components related to the present invention and are not drawn according to the actual number, shape and size of the components in the actual implementation. In the actual implementation, the form, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0024] In the following description, numerous details are explored to provide a more thorough explanation of embodiments of the invention. However, it will be apparent to those skilled in the art that embodiments of the invention may be practiced without these specific details. In other embodiments, well-known structures and devices are shown in block diagram form rather than in detail to avoid obscuring embodiments of the invention.

[0025] Example 1: A method for analyzing the ecological quality trend of crested ibis habitat, such as Figure 1 As shown, it includes: S1: Construct an intelligent sensing network to synchronously acquire multi-source habitat data and identify crested ibis activity events, and generate a multi-dimensional habitat parameter table through edge computing fusion; S2: Collect environmental samples based on preset environmental locations, and generate a microbial functional gene abundance matrix through metagenomic sequencing and bioinformatics analysis; S3: Using crested ibis activity events as behavioral labels, coupled with a multidimensional habitat parameter table and a functional gene abundance matrix, a machine learning prediction model is trained to generate a habitat functional health index; S4: Based on the spatiotemporal sequence of habitat functional health index, perform spatial interpolation and trend analysis to generate a spatiotemporal evolution map of ecological quality; S5: Based on the spatiotemporal evolution map of ecological quality, quantitative analysis is performed using a spatial differentiation statistical model to generate a trend analysis report that clearly identifies the dominant driving factors of ecological quality changes.

[0026] The present invention is further configured such that S1 includes: A smart sensing network covering the target area is constructed, the smart sensing network including: an acoustic monitoring unit, a remote sensing acquisition unit, and an in-situ physicochemical monitoring unit; Raw environmental audio data is collected using an acoustic monitoring unit; Utilize remote sensing to acquire unit multispectral image data and thermal infrared radiation data; Raw test data of water and soil were collected using in-situ physicochemical monitoring units; The raw audio data, multispectral image data, thermal infrared radiation data, and raw monitoring data are processed and fused to construct a multidimensional habitat parameter table.

[0027] The processing and fusion include: The raw audio data is processed by time-domain and frequency-domain analysis. Soundscape index is extracted based on acoustic statistical analysis method. At the same time, crested ibis activity events are identified based on pattern recognition method. Acoustic parameters are constructed by combining soundscape index and specific biological sound source events. Radiometric correction and feature extraction were performed on multispectral image data. Vegetation-related spectral indices were generated based on spectral index calculation methods, and habitat type identification was completed based on image classification methods. The thermal infrared radiation data is processed and the land surface temperature is obtained based on the thermal infrared inversion model. Remote sensing parameters are constructed by combining spectral vegetation index, habitat classification and land surface temperature. Multidimensional habitat parameters are generated by performing time synchronization, spatial correlation, and edge-side fusion processing on acoustic, remote sensing, and physicochemical parameters. Specifically, a multidimensional habitat parameter table is generated by constructing an intelligent sensing network covering the target habitat. This intelligent sensing network consists of three units: an acoustic monitoring unit, a remote sensing acquisition unit, and an in-situ physicochemical monitoring unit. First, the acoustic monitoring unit consists of waterproof digital recording nodes arranged in a regular grid, which periodically record raw environmental audio at a specific sampling rate, such as 48kHz, along with coordinate and time information. The remote sensing acquisition unit uses drones equipped with multispectral and thermal imaging cameras to perform flight missions according to pre-planned preset routes, simultaneously acquiring multi-band reflectivity images and thermal infrared radiation data. The in-situ physicochemical monitoring unit automatically and continuously collects raw detection data, including but not limited to temperature, pH, dissolved oxygen, and humidity, through water quality and soil sensors deployed in typical habitats. Subsequently, in the data processing and fusion stage, the raw acoustic audio is bandpass filtered at the edge computing nodes and then segmented into short-time frames. An acoustic complexity index is obtained by calculating the sum of amplitude variations in each frame's frequency sub-band using a standard algorithm. A bioacoustic index is obtained by calculating the signal intensity within a specific biological activity frequency band using a standard algorithm. A standardized differential soundscape index is obtained by calculating the energy ratio of the biological frequency band to the anthropogenic noise frequency band using a standard algorithm. Simultaneously, by converting the audio into a Mel spectrogram and inputting it into an existing, pre-trained bird sound recognition convolutional neural network model such as BirdNet, specific biological sound source events such as the crested ibis call are automatically identified, and the duration and corresponding timestamp of the crested ibis call are obtained. The raw remote sensing images acquired by the UAV are radiometrically calibrated, atmospherically corrected, and geometrically corrected in the cloud to generate orthophotos. Based on this, the normalized differential vegetation index and normalized differential water index are calculated, surface temperature is retrieved, and an object-oriented multi-scale segmentation and random forest classification method is used to automatically identify and classify habitat types. Data uploaded from in-situ sensors undergoes automatic cleaning and moving average filtering based on rule-based thresholds. For example, thresholds are set based on physically feasible ranges (e.g., water temperature 0-40°C, dissolved oxygen ≥0 mg / L) to obtain standardized physicochemical parameter sequences. Finally, in the data fusion stage, all records are aligned to a standard time slice using Coordinated Universal Time (UTC) as the unified time base. Through spatial connectivity or point-polygon overlay analysis in the geographic information system, the coordinates of acoustic nodes and sensors are assigned to polygonal habitat patches of their locations, and the unique ID of each patch is recorded. The system then aggregates and calculates the mean acoustic index, cumulative duration of crested ibis calls, and mean physicochemical parameters within the spatiotemporal unit using time and patch ID as a joint index. This data is then combined with the inherent attributes of the patches to ultimately generate a comprehensive data table for each record, containing habitat type, vegetation index, temperature, acoustic characteristics, species activity, and physicochemical conditions—a multidimensional habitat parameter table.

[0028] The present invention is further configured such that S2 includes a sequencing data generation step and a matrix construction step. Specifically, the sequencing data generation step extracts total microbial DNA from environmental samples and performs metagenomic sequencing, converting physical samples such as soil and water into digitized raw gene sequence data, thereby achieving comprehensive acquisition of the genetic information of the microbial community in the habitat and providing basic data raw materials for subsequent functional analysis; the matrix construction step performs quality control, functional annotation, and standardized quantification of the gene sequence data to construct a microbial functional gene abundance matrix from the raw sequences. The microbial functional gene abundance matrix reveals key micro-ecological functions of the habitat and generates quantitative functional indicators that can be coupled with macro-habitat parameters, providing an intrinsic mechanistic basis for comprehensive evaluation.

[0029] The present invention is further configured such that the sequencing data generation step includes: Collect environmental samples according to the preset environmental sampling points; Total microbial DNA was extracted from environmental samples; The total DNA was sequenced using a shotgun sequencing strategy to obtain gene sequence data. Specifically, the entire sequencing data generation process begins at a pre-defined habitat sampling point with geographical coordinates. Collection methods include manual collection by staff, collection using drones, or collection using robots. This embodiment exemplifies the manual collection method, where staff use pre-sterilized specialized tools and strictly adhere to aseptic operating procedures to simultaneously collect soil core and surface water samples. Soil samples are directly placed into sterile tubes, while water samples are immediately filtered on-site through a sterile filter membrane to enrich microorganisms. All samples are immediately placed in an ultra-low temperature environment for preservation after collection and rapidly transported to the laboratory to maintain the stability of biomolecules. Upon arrival at the laboratory, the samples are removed from the ultra-low temperature environment and weighed or processed on a low-temperature operating table. For soil or filter membranes, a commercially available environmental DNA extraction kit is used. The core process involves placing the sample and a test tube containing lysis beads and lysis buffer together on a high-speed vortex mixer for vigorous shaking, using mechanical force to break the microbial cell walls; simultaneously, the chemical components in the lysis buffer dissolve the cell membrane and degrade proteins, releasing DNA into the solution. Subsequently, solid impurities were removed by centrifugation, and the DNA-containing supernatant was transferred to a centrifuge column lined with a silica gel membrane. Under specific salt solution conditions, DNA molecules specifically adsorbed onto the silica gel membrane, while impurities were washed away by the washing buffer. Finally, the purified DNA was dissolved from the membrane using a low-ionic-strength elution buffer to obtain the total DNA extract of environmental microorganisms, the concentration and purity of which were quantified by ultraviolet absorption spectroscopy. After obtaining qualified total DNA, the library preparation stage began. First, long-chain DNA was randomly fragmented into short fragments of specific lengths using an ultrasonic device or enzymatic method. The ends of these fragments were repaired to blunt ends, and then universal adapter sequences required for sequencing were ligated to both ends. These adapters contained the information necessary for subsequent amplification, sequencing, and differentiation of different samples. The successfully ligated library fragments were amplified and enriched through a limited number of rounds of PCR to form the final sequencing library. The quality of the sequencing library, including fragment size distribution and molar concentration, was accurately determined using a bioanalyzer based on capillary electrophoresis to obtain the fragment size distribution and molar concentration. Multiple quality-tested sample libraries are mixed together in a calculated ratio and loaded into the flow cell of a high-throughput sequencer (such as the Illumina platform). Inside the flow cell, each library fragment binds to complementary oligonucleotides on the chip via adapters at both ends and undergoes in situ amplification via bridge PCR, forming a "cluster" of thousands of identical DNA templates. Once sequencing begins, four reversible terminator dNTPs with different fluorescent labels and DNA polymerase are added cyclically. In each round of the reaction, only one dNTP complementary to the template strand is incorporated into the extended strand, releasing its unique fluorescent signal.A high-resolution camera captures fluorescence signals, which are then identified by software to determine the type of base incorporated at each location. Subsequently, the terminator and fluorescent groups are chemically removed, preparing for the next round of incorporation. This "synthesis-imaging-removal" cycle is repeated to achieve de novo reading of each DNA cluster sequence. Finally, the sequencing control software converts the fluorescence signal images of all clusters into base sequences and outputs them in the internationally recognized FASTQ file format. This file contains not only the ATCG base sequence of each read but also a sequencing quality score for each base. Thus, the environmental sample is successfully converted into raw, digitized genetic sequence data.

[0030] The present invention is further configured such that the matrix construction step includes: Quality control, splicing and assembly, and gene prediction are performed on gene sequence data to obtain a non-redundant set of gene sequences; The gene sequence set is compared with a pre-defined database of known functional genes to obtain functional gene annotation information; Based on sequence alignment results, the frequency or readings of various functional genes in each sample are statistically analyzed and normalized to form a microbial functional gene abundance matrix. Specifically, the entire matrix construction process is based on the original gene sequence data files and can be divided into two main stages: data preprocessing and feature generation and aggregation. The first data preprocessing stage involves parallel quality control of all files using Fastp software. The software reads the sequencer's data and converts it into an integer value using a standard formula (Q=-10*log10(P)) based on the Phred scoring system, providing a quality score for each base. Here, Q represents the quality score, and P is the sequencing error probability calculated by the sequencer based on the fluorescence signal intensity model during base recognition. Then, using a default threshold (e.g., Q set to 20) combined with sliding window analysis along the read segment, low-quality portions at both ends are adaptively removed, along with residual sequencing adapter sequences. Simultaneously, the average quality score within the window is calculated through sliding window scanning, and reads with excessively low average quality scores or excessively high proportions of N bases are filtered out, ultimately generating a "clean" high-quality read file for each sample. Next, the assembly stage begins. High-quality read files from all the samples are merged and submitted as input to the MEGAHIT software. MEGAHIT first breaks down all reads into shorter, fixed-length subsequences and counts their frequencies, constructing a complex graph network. Then, it searches this network for optimal paths to connect overlapping subsequences, thus splicing hundreds of millions of short reads into hundreds of thousands of longer continuous sequences. These continuous sequences represent potential genomic fragments within the habitat's microbial community. Next, gene prediction is performed on all the assembled continuous sequences: MetaGeneMark software scans each sequence. Its built-in Hidden Markov Model identifies segments within the sequence that conform to gene coding rules, i.e., it finds the start codon and continues prediction until a stop codon is encountered, thus defining a complete gene coding region. Finally, the software outputs the DNA sequences of all predicted genes. To remove repetitive or highly similar gene sequences generated from different samples or during assembly, CD-HIT software was used to cluster all predicted genes. CD-HIT rapidly compares the sequence similarity between genes pairwise, grouping genes with a similarity exceeding 95% into the same cluster, and selecting the longest sequence from each cluster as a representative. Ultimately, a non-redundant reference gene set composed of unique gene sequences was obtained. The second feature generation and aggregation stage involves functional annotation: using a standard genetic code table, the DNA sequences in the non-redundant reference gene set were translated into corresponding amino acid sequences. Then, DIAMOND software was used to perform high-speed alignment of these amino acid sequences with selected functional gene databases such as KEGG and eggNOG.DIAMOND rapidly finds the most similar known functional genes in the database for each gene by establishing a dual index of query sequences and target databases. By setting strict statistical significance thresholds, reliable alignment results are selected, thus assigning functional descriptions and metabolic pathway classification information to most genes in the gene set. Finally, gene abundance is quantified and standardized. Using Bowtie2 software, each sample's individual high-quality read files are accurately aligned back to the aforementioned non-redundant gene set. The number of reads successfully aligned to each gene is counted to obtain the raw count of that gene in the sample. Since the total sequencing volume varies across samples, and gene lengths differ, directly comparing raw counts is meaningless. Therefore, TPM standardization is performed for each sample: first, the raw read count of each gene is divided by its length; then, the length-corrected read counts of all genes in a sample are summed to calculate a scaling factor per million; finally, the length-corrected read count of each gene is divided by this scaling factor to obtain the standardized abundance value of that gene in the sample. Finally, the standardized abundance values ​​of all functional genes from all samples were compiled into a two-dimensional table, thus generating the final microbial functional gene abundance matrix. In this matrix, each row corresponds to a habitat sampling point, and each column corresponds to a specific functional gene or pathway. The values ​​in the microbial functional gene abundance matrix quantitatively reflect the activity level of specific functions of microbial communities at different locations, providing core, mechanistic microscopic functional indicators for subsequent comprehensive ecological quality assessment.

[0031] The present invention is further configured such that S3 includes: A fused feature set was constructed by coupling a multidimensional habitat parameter table and a functional gene abundance matrix. Acquire identified crested ibis activity events and use these events as behavioral labels; By using behavioral labels as supervision signals, a pre-defined machine learning prediction model is trained using a fused feature set. For the target area, input features are formed based on a multidimensional habitat parameter table and functional gene abundance. These features are then input into a trained machine learning prediction model, outputting the corresponding habitat functional health index. Specifically, firstly, data fusion and feature engineering are performed. The system merges two data tables into a single master table based on the shared spatiotemporal identifiers of each data point, ensuring that each sample point possesses both environmental and gene functional characteristics. Next, this master table is cleaned and transformed: for occasionally missing values, the median value of all data in that column is used to fill in the gaps; for all numerical columns, such as vegetation indices, temperature, and the abundance of various genes, standardization is performed—first subtracting the average of all data in that column, then dividing by its fluctuation range (standard deviation) to ensure that these numbers are within a similar range. Furthermore, based on ecological knowledge, potentially related columns can be multiplied or compared to create new feature columns. Finally, using the preliminary results of the model to be trained, the importance of each feature column to the final prediction is analyzed, and those feature columns that are almost ineffective are removed, forming a clean and standardized "fusion feature set." Next, the "answer" for model learning, i.e., behavioral labels, is generated. The system reads acoustic monitoring records and checks whether crested ibis calls are detected for each sample point (i.e., data from a specific patch within a specific time period) in the fused feature set. If the call duration is greater than zero, the sample point is labeled "1," indicating that crested ibises are using it; otherwise, it is labeled "0." This label column is added to the last column of the fused feature set. Since in reality, "useful" samples are usually far fewer than "unuseful" samples, to prevent the model from overemphasizing the majority class, a portion of the "0" labeled samples is randomly selected to roughly balance the number of "0" and "1" samples. Then, the core model training phase begins. First, the complete labeled feature set is randomly shuffled and divided into two parts in a roughly 7:3 ratio, with the larger part serving as the training set and the smaller part as the test set. The efficient gradient boosting decision tree algorithm LightGBM is selected as the model. During training, the model uses the features of the training set as input and the corresponding labels as the "standard answers" to begin the learning process. Its learning mechanism involves iteratively building many simple decision trees, each attempting to correct the residuals predicted by the previous tree. To find the optimal model parameters (such as tree complexity and learning speed), a cross-validation strategy is employed: the training set is divided into five parts, and four parts are used for training and one for validation in turn, repeated five times to ensure stable evaluation, while automatically searching for a parameter combination that yields the best validation results. After training, the model is finally evaluated using a test set that was never used in training: the model calculates a predicted value between 0 and 1 for each sample in the test set, and the system calculates metrics such as AUC by analyzing the degree of agreement between these predicted values ​​and the true labels.Only when these indicators perform well and the training process is stable are the model parameters and structure fixed and saved as the final prediction model. Finally, for any new target area to be evaluated (e.g., all patches in a new habitat map, or monitoring data from a new period), environmental and genetic data for that area are first acquired following the S1 and S2 procedures. Then, these new data must undergo identical feature cleaning, standardization, and transformation using the same steps and parameters as during training (i.e., using the mean, standard deviation, and feature list saved during training) to generate a new feature vector with a uniform format. This feature vector is then input into the saved LightGBM model. Inside the model, this data flows sequentially through each previously trained decision tree, branching into different branches based on the feature values, eventually reaching a leaf node that provides a base score. The base scores from all trees are summed, and this sum is mapped to a range of 0 to 1 using a sigmoid function. The final output probability value is the "Habitat Functional Health Index" for the target area.

[0032] The present invention is further configured such that S4 includes: Spatiotemporal sequence data were constructed based on habitat functional health indices at multiple time points. Spatial interpolation is performed on the spatiotemporal sequence data to generate continuous spatial distribution surfaces at each time point; Based on the spatial distribution surface, the trend of each spatial location is calculated in the time series to obtain the mass change trend of each location; Based on the trend of quality changes, a spatiotemporal evolution map of ecological quality is generated using map-making methods. Specifically, the system first reads the calculated habitat functional health index data, where each data point contains the coordinates of the center point of its corresponding spatial unit, the time of calculation, and the habitat functional health index value itself. The program organizes this information into a standard table, where each row clearly records a triplet of "where-when-value", thus constructing a discrete spatiotemporal point dataset for spatial analysis. Next, spatial interpolation is performed to generate a continuous surface. The system splits the above dataset into multiple independent subsets according to the time field, with each subset representing all sampling point data for a specific month or season. For each time subset, the system calls the Kriging interpolation algorithm. This algorithm first analyzes the spatial clustering patterns and variability of these discrete points, automatically fitting a mathematical model describing the rule that "the farther the distance, the less similar the attributes". Then, guided by this model, the algorithm delineates a regular grid for the entire study area and estimates the value of each intersection point in the grid (i.e., the center point of each cell in the future raster map). During the estimation process, the algorithm searches for all known sampling points within a certain distance of the unknown point. Based on their distances to the unknown point and their spatial relationships, it calculates an optimal set of weights and calculates a weighted average of the values ​​from these known points to obtain the predicted value for the unknown point. After this process iterates through all grid points, a smooth digital image (raster layer) representing the ecological quality distribution at that time point is generated, fully covering the study area. This process is repeated for all time slices, resulting in a chronologically ordered sequence of raster images. Then, trend mining along the time dimension begins. The system reads the chronologically ordered raster sequence. For each fixed pixel location within the geographical area of ​​the coverage region, the program extracts the value of that location sequentially from all raster images along the time axis, forming a location-specific ecological quality time series. For this time series, the system employs the Sennheiser slope estimation algorithm for analysis. The algorithm considers all possible pairwise time point combinations within the series, calculates the change in ecological quality corresponding to each pair by dividing the time interval, obtaining numerous instantaneous rates of change. The median of these instantaneous rates of change is then taken as the long-term, robust trend slope for that location. Simultaneously, the system performs the Mann-Kendall test: it examines the entire time series, checking whether the values ​​exhibit an increasing, decreasing, or random rank relationship over time, and calculates a p-value to determine the significance of the trend using a set of standard statistics. The system independently performs the above calculations for each pixel location within the study area. Finally, all calculated "trend slopes" and "significance p-values" are filled into two new, blank geographic grid maps, generating two new raster layers—one for the trend slope distribution and the other for the statistical significance distribution. Finally, comprehensive mapping and result generation are performed.The mapping system merges and visualizes the two raster layers. First, based on the "significance p-value" layer, it categorizes pixels into two main classes: "statistically significant" and "insignificant." Then, within the significant pixels, it further subdivides them into categories such as "significant improvement" and "significant degradation" based on the sign and magnitude of the "trend slope." Insignificant pixels are categorized as "insignificant change." The system assigns specific colors to each pixel category according to preset mapping standards; for example, dark green represents significant improvement, dark red represents significant degradation, and light gray represents insignificant change. Subsequently, the system overlays reference background information such as administrative divisions and river systems onto this thematic layer, adds a legend, scale bar, north arrow, and descriptive title, completing the overall map layout. Finally, this thematic map, comprehensively reflecting the spatiotemporal evolution trend, pattern, and reliability of ecological quality, is output as the final visualization result: the spatiotemporal evolution map of ecological quality.

[0033] The present invention is further configured such that S5 includes: Spatial overlay and correlation of the spatiotemporal evolution map of ecological quality and the multidimensional habitat parameter table; Using a geographic detector model, we calculated the independent explanatory power of each habitat parameter for the spatial differentiation of ecological quality; Analyze the explanatory power of the interaction between any two habitat parameters on the spatial differentiation of ecological quality; Based on the ranking and numerical values ​​of independent and interactive explanatory power, the dominant driving factors and their corresponding influence strengths are determined; Based on the dominant driving factors and their intensity, a trend analysis report is generated. Specifically, the ecological quality trend map generated by S4 is deeply correlated with the abundant habitat parameters from S1 and S2. First, a pixel-level long-term change slope raster map is extracted from the trend map. Simultaneously, all candidate driving factors in the multidimensional habitat parameter table, such as vegetation indices for different seasons, average soundscape indices, key water quality parameters, and abundance of specific microbial functional genes, are all converted into raster layers completely consistent with the spatial baseline of the trend slope map using spatial interpolation methods. Then, the system evenly distributes tens of thousands of sampling points within the analysis area, batch extracting the values ​​of each point on all raster maps, forming a large, wide table where each row contains a location point, representing "trend value - values ​​of various environmental factors," preparing for subsequent statistical analysis. Next, the system calls a geographic detector model to mine driving factors. The model first automatically classifies each continuous habitat factor according to the natural discontinuities of its numerical distribution, converting it into a categorical variable. Then, the core operation is performed for each factor: the model divides the entire study area into several sub-regions according to the factor's category. Using the principle of variance decomposition, it calculates how much the ecological quality trend differs between different sub-regions, and how similar they are within the same sub-region, if only this factor is considered. Finally, an independent explanatory power (q-value) between 0 and 1 is calculated. This q-value clearly tells us, for example, that after classifying "shallow water area," it can explain 45% of the spatial differences in ecological quality trends on its own. The system performs this calculation for all factors and generates a ranking of independent explanatory power (q-values). Afterward, the system delves deeper into the complex relationships between factors. It automatically performs pairwise combination analysis on the top-ranked factors. For example, it creates a new layer to identify areas that are "both high vegetation cover and low anthropogenic disturbance," and then calculates the explanatory power (q-value) of the interaction effect of this new combination of factors on the trend. By comparing the interaction q-value (q(X1∩X2)) with the sum of the independent q-values ​​of the two factors (q(X1)+q(X2)) and the maximum value (Max(q(X1),q(X2))), the system can determine whether the two factors have a synergistic reinforcing relationship ("one plus one is greater than two") or are independent. This step may reveal hidden mechanisms such as "high vegetation cover only has a strong promoting effect on ecological quality under low disturbance background." Based on the above quantitative results, the system intelligently screens out the dominant driving factors by combining preset thresholds (e.g., q>0.2) and the strength of the interaction. For example, the conclusion may show that "the dominant driving factor is 'spring shallow water area', which has the strongest independent explanatory power; followed by 'nighttime noise level', and there is a strong nonlinear reinforcing interaction effect between the two, indicating that noise will drastically amplify the negative impact of habitat loss." Finally, all analysis results are automatically imported into the report generation template. A report framework including abstract, methods, results and discussion, and recommendations is created.The results section includes trend change maps, bar charts ranking the explanatory power values ​​of driving factors, and diagrams illustrating key interactions. In the discussion section, based on the aforementioned charts and data, it automatically generates mechanistic interpretations of the causes of ecological degradation or improvement. In the management recommendations section, the system outputs highly targeted suggestions based on the spatial distribution and interaction types of the dominant driving factors, such as "prioritizing water system restoration projects in the northeastern region, where shallow water areas are currently lacking but historically present, and simultaneously implementing noise control measures on major traffic arteries within this region." This comprehensive trend analysis report integrates spatial patterns, statistical evidence, and ecological mechanisms.

[0034] Example 2: Please see Figure 2 This exemplary system for analyzing the ecological quality trends of crested ibis habitat includes: Intelligent sensing and data fusion module: Constructs an intelligent sensing network, synchronously acquires multi-source habitat data and identifies crested ibis activity events, and generates a multi-dimensional habitat parameter table through edge computing fusion; Environmental genomics analysis module: Based on environmental samples collected from preset environmental locations, a microbial functional gene abundance matrix is ​​generated through metagenomic sequencing and bioinformatics analysis; Intelligent Health Index Assessment Module: Using crested ibis activity events as behavioral labels, coupling a multidimensional habitat parameter table with a functional gene abundance matrix, training a machine learning prediction model, and generating a habitat functional health index; Spatiotemporal dynamic map generation module: Based on the spatiotemporal sequence of habitat functional health index, spatial interpolation and trend analysis are performed to generate a spatiotemporal evolution map of ecological quality; Driving Factor Analysis and Report Generation Module: Based on the spatiotemporal evolution map of ecological quality, quantitative analysis is performed using a spatial differentiation statistical model to generate a trend analysis report that clearly identifies the dominant driving factors of ecological quality changes.

[0035] It should be noted that the crested ibis habitat ecological quality trend analysis system and the crested ibis habitat ecological quality trend analysis method provided in the above embodiments belong to the same concept. The specific operation methods of each module and unit have been described in detail in the method embodiments and will not be repeated here. In practical applications, the crested ibis habitat ecological quality trend analysis system provided in the above embodiments can be assigned to different functional modules as needed, that is, the internal structure of the system can be divided into different functional modules to complete all or part of the functions described above. This is not a limitation here.

[0036] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for analyzing the ecological quality trend of crested ibis habitat, characterized in that, include: S1: Construct an intelligent sensing network to synchronously acquire multi-source habitat data and identify crested ibis activity events, and generate a multi-dimensional habitat parameter table through edge computing fusion; S2: Collect environmental samples based on preset environmental locations, and generate a microbial functional gene abundance matrix through metagenomic sequencing and bioinformatics analysis; S3: Using crested ibis activity events as behavioral labels, coupled with a multidimensional habitat parameter table and a functional gene abundance matrix, a machine learning prediction model is trained to generate a habitat functional health index; S4: Based on the spatiotemporal sequence of habitat functional health index, perform spatial interpolation and trend analysis to generate a spatiotemporal evolution map of ecological quality; S5: Based on the spatiotemporal evolution map of ecological quality, quantitative analysis is performed using a spatial differentiation statistical model to generate a trend analysis report that clearly identifies the dominant driving factors of ecological quality changes.

2. The method for analyzing the ecological quality trend of crested ibis habitat according to claim 1, characterized in that, S1 includes: A smart sensing network covering the target area is constructed, the smart sensing network including: an acoustic monitoring unit, a remote sensing acquisition unit, and an in-situ physicochemical monitoring unit; Raw environmental audio data is collected using an acoustic monitoring unit; Utilize remote sensing to acquire unit multispectral image data and thermal infrared radiation data; Raw test data of water and soil were collected using in-situ physicochemical monitoring units; The raw audio data, multispectral image data, thermal infrared radiation data, and raw monitoring data are processed and fused to construct a multidimensional habitat parameter table.

3. The method for analyzing the ecological quality trend of crested ibis habitat according to claim 2, characterized in that, The processing and fusion include: The raw audio data is processed by time-domain and frequency-domain analysis. Soundscape index is extracted based on acoustic statistical analysis method. At the same time, crested ibis activity events are identified based on pattern recognition method. Acoustic parameters are constructed by combining soundscape index and specific biological sound source events. Radiometric correction and feature extraction were performed on multispectral image data. Vegetation-related spectral indices were generated based on spectral index calculation methods, and habitat type identification was completed based on image classification methods. The thermal infrared radiation data is processed and the land surface temperature is obtained based on the thermal infrared inversion model. Remote sensing parameters are constructed by combining spectral vegetation index, habitat classification and land surface temperature. Multidimensional habitat parameters are generated by performing time synchronization, spatial correlation, and edge-side fusion processing on acoustic parameters, remote sensing parameters, and physicochemical parameters.

4. The method for analyzing the ecological quality trend of crested ibis habitat according to claim 1, characterized in that, S2 includes: a sequencing data generation step and a matrix construction step.

5. The method for analyzing the ecological quality trend of crested ibis habitat according to claim 4, characterized in that, The sequencing data generation steps include: Collect environmental samples according to the preset environmental sampling points; Total microbial DNA was extracted from environmental samples; The total DNA was sequenced using a shotgun sequencing strategy to obtain gene sequence data.

6. The method for analyzing the ecological quality trend of crested ibis habitat according to claim 5, characterized in that, The matrix construction steps include: Quality control, splicing and assembly, and gene prediction are performed on gene sequence data to obtain a non-redundant set of gene sequences; The gene sequence set is compared with a pre-defined database of known functional genes to obtain functional gene annotation information; Based on the sequence alignment results, the frequency or reading of various functional genes in each sample was statistically analyzed, normalized, and a microbial functional gene abundance matrix was formed.

7. The method for analyzing the ecological quality trend of crested ibis habitat according to claim 1, characterized in that, S3 includes: A fused feature set was constructed by coupling a multidimensional habitat parameter table and a functional gene abundance matrix. Acquire identified crested ibis activity events and use these events as behavioral labels; By using behavioral labels as supervision signals, a pre-defined machine learning prediction model is trained using a fused feature set. For the target area, input features are formed based on a multidimensional habitat parameter table and functional gene abundance, which are then input into a trained machine learning prediction model to output the corresponding habitat functional health index.

8. The method for analyzing the ecological quality trend of crested ibis habitat according to claim 1, characterized in that, S4 includes: Spatiotemporal sequence data were constructed based on habitat functional health indices at multiple time points. Spatial interpolation is performed on the spatiotemporal sequence data to generate continuous spatial distribution surfaces at each time point; Based on the spatial distribution surface, the trend of each spatial location is calculated in the time series to obtain the mass change trend of each location; Based on the trend of quality change, a spatiotemporal evolution map of ecological quality is generated using map-making methods.

9. The method for analyzing the ecological quality trend of crested ibis habitat according to claim 1, characterized in that, S5 includes: Spatial overlay and correlation of the spatiotemporal evolution map of ecological quality and the multidimensional habitat parameter table; Using a geographic detector model, we calculated the independent explanatory power of each habitat parameter for the spatial differentiation of ecological quality; Analyze the explanatory power of the interaction between any two habitat parameters on the spatial differentiation of ecological quality; Based on the ranking and numerical values ​​of independent and interactive explanatory power, the dominant driving factors and their corresponding influence strengths are determined; A trend analysis report is generated based on the dominant driving factors and their strength.

10. A system for analyzing the ecological quality trend of crested ibis habitat, used to implement the method for analyzing the ecological quality trend of crested ibis habitat as described in any one of claims 1-9, characterized in that, include: Intelligent sensing and data fusion module: Constructs an intelligent sensing network, synchronously acquires multi-source habitat data and identifies crested ibis activity events, and generates a multi-dimensional habitat parameter table through edge computing fusion; Environmental genomics analysis module: Based on environmental samples collected from preset environmental locations, a microbial functional gene abundance matrix is ​​generated through metagenomic sequencing and bioinformatics analysis; Intelligent Health Index Assessment Module: Using crested ibis activity events as behavioral labels, coupling a multidimensional habitat parameter table with a functional gene abundance matrix, training a machine learning prediction model, and generating a habitat functional health index; Spatiotemporal dynamic map generation module: Based on the spatiotemporal sequence of habitat functional health index, spatial interpolation and trend analysis are performed to generate a spatiotemporal evolution map of ecological quality; Driving Factor Analysis and Report Generation Module: Based on the spatiotemporal evolution map of ecological quality, quantitative analysis is performed using a spatial differentiation statistical model to generate a trend analysis report that clearly identifies the dominant driving factors of ecological quality changes.

Images