A method for predicting the potential distribution and risk assessment of Chinese white dolphins

Abstract

The present invention discloses a method for predicting the potential distribution and risk assessment of Chinese white dolphins, including data collection, data preprocessing, and arranging the data into a format that can be recognized by the HMSC package in R Studio, wherein a specific data frame needs to be set, and the data frame specifically includes: a species data frame, an environmental data frame, and a research scope data frame, and an HMSC model object is created and a calculation method is set, and model fitting and diagnosis are performed, and the species distribution is predicted through the trained HMSC model, and finally the results are visualized and analyzed. The present invention analyzes the potential distribution of Chinese white dolphins and their bait fish under current and future environmental changes, as well as the energy flow relationship between biological food chains by combining a joint species distribution model and an Ecopath model.

Description

Technical Field

[0001] The present application relates to the field of marine biostatistics and big data processing, and mainly to a method for predicting the potential distribution and risk assessment of Chinese white dolphins. Background Art

[0002] At present, domestic and foreign research mainly focuses on analyzing the distribution and abundance of Chinese white dolphins, and less on the impact of global changes and forage fish on Chinese white dolphins. Global change affects most marine areas, leading to the decline or even loss of marine species, reduced fish production, and reduced habitats, which in turn affects the structure, process and function of ecosystems, thereby changing marine ecosystem services, including food supply and coastal protection. Global change has had a significant impact on many economic fish and wildlife by changing habitats or food webs. As the top predator in the marine ecosystem, the Chinese white dolphin plays an important role in maintaining ecological balance. Their predation activities can help control the number of economic fish populations. In some cases, the presence of predators can promote the diversity of prey species because they usually prey more on those species with a larger number, thereby providing other species with fewer numbers with opportunities to grow. Therefore, the significant impact of the reduction of forage fish for Chinese white dolphins cannot be ignored when choosing research methods.

[0003] With the development of technology, the application of species distribution models is becoming more and more extensive. It predicts the distribution range of animals and plants and explores the influencing factors by establishing a quantitative relationship between species and their important environmental variables. At present, the most widely used model is the maximum entropy (MaxEnt) model. The MaxEnt model can obtain relatively accurate prediction results with a small amount of sample data, and can handle complex situations such as nonlinear relationships, but it mainly analyzes a single species, and the explanatory power of species interactions needs to be strengthened. Because global changes not only affect the living environment of top species, but more importantly, the reduction in the number of medium and low-level fish leads to changes in interspecies relationships and ecosystem functions, which is very likely to lead to the loss and extinction of Chinese white dolphins. The joint species distribution model is an ecosystem model based on genetic algorithms. It predicts the distribution of species in a new environment by simulating the ecological adaptability of species and the sustainability of the environment. It can take into account the interactions between species and the complexity of the environment and explain the interactions between species.

[0004] The species distribution model has outstanding advantages in explaining environmental variables and species distribution, and can better predict the potential distribution of species. However, the explanation of energy conversion between species needs to be strengthened. Therefore, the energy flow model can be combined to analyze the changes in food chain structure, simulate the spatial changes of Chinese white dolphins, improve accuracy and credibility, and provide a reference for weighing the maximum foraging probability and minimum mortality risk rate of Chinese white dolphins. The energy flow model can take into account a variety of factors such as fish, plankton, benthic habitats, human activities, and climate change, providing a reference for the survival and development of Chinese white dolphins. At present, the models for analyzing food chains and food webs mainly include the top predator model, Ecopath model, MICE model (Model of Intermediate Complexity for Ecosystem assessments) and Ecospace model. By comparison, it is found that the Ecopath model requires less data and is easier to build and use. In addition, by defining management strategies and climate change scenarios, the output results of the model under different scenarios can be analyzed, the impact of management strategies and climate change on the interactions between ecosystem components, energy flow and trophic level structure can be evaluated, and which management measures can effectively enhance the resistance and resilience of the ecosystem can be identified. The Ecopath model is based on the optimal foraging theory, which expects animals to maximize foraging benefits while minimizing risks. At the same time, it can take into account a variety of factors such as fish, plankton, benthic habitats, human activities, climate change, etc., to improve the accuracy and credibility of predictions. Summary of the invention

[0005] This paper takes global change as a perspective, takes the Chinese white dolphin and its forage fish as an example, and makes a potential prediction of the potential distribution of the Chinese white dolphin and its forage fish through a joint species distribution model. Through the Ecopath model analysis, the feeding intensity of the Chinese white dolphin and its forage fish driven by global change is analyzed, and the management system under different scenarios and the risk effects on the Chinese white dolphin ecosystem under the global change scenario are analyzed, providing theoretical support and scientific methods for the protection and management of the Chinese white dolphin.

[0006] According to one aspect of the present invention, a method for predicting the potential distribution and risk assessment of Chinese white dolphins is proposed, comprising:

[0007] S1. Taking the Chinese white dolphins and bait fish in the study area as sample points, collecting the occurrence record data and spatial location data of the sample points, and collecting future environmental data as environmental variables;

[0008] S2. Standardize the environmental variables, perform visual analysis on the spatial location data, and remove data whose coordinates are not in the study area;

[0009] S3, organizing the occurrence record data into a format that can be recognized by the HMSC package in R Studio, wherein a specific data frame needs to be set, and the data frame specifically includes: a species data frame, an environmental data frame, and a spatial data frame, and creating an HMSC model object and setting a calculation method to generate a trained HMSC model;

[0010] S4. Fitting the model using the fitting method provided by the HMSC package, evaluating the model fitting based on the explanatory power, and evaluating the model fitting degree based on the predictive ability of the two-fold cross validation;

[0011] S5, using the fitting results of the model in step S4 to analyze the relationship between the Chinese white dolphin and bait fish species, including symbiotic relationships and competitive relationships, and to analyze and explain the spatial pattern of species distribution;

[0012] S6, inputting the future environmental data into the trained HMSC model in step S3 to predict the potential distribution, further processing according to the prediction results, and analyzing the position changes between the species hotspot distribution area and the current distribution area;

[0013] S7. Visualize the location changes using GIS analysis to display model results, including potential distribution maps of species and importance scores of environmental variables;

[0014] S8. Use the relationship between species obtained in step S5 to perform functional division on bait fish, collect biological information of fish as parameters, input the parameters into the Ecopath model, analyze the energy conversion between trophic levels, and obtain the transmission efficiency.

[0015] Furthermore, the step S1 also includes: determining the main risk factors affecting the Chinese white dolphin and bait fish through the Citespace model, including climate factors, marine environmental factors and human activity factors.

[0016] The Citespace model refers to a scientific knowledge graph construction and analysis method used by the Citespace software, a tool used in the fields of scientometrics and information visualization.

[0017] In the Citespace model, by calculating statistical indicators (such as the strength of emergent words, growth rate, etc.), we can identify new concepts, new technologies or new research directions that suddenly become significant in a specific period of time. These emergent items often indicate the development trend of the scientific frontier. Citespace displays the above analysis results in a graphical way, including timeline diagrams, co-word network diagrams, cluster diagrams, etc., making the complex scientific knowledge structure and dynamic evolution process intuitive and easy to understand.

[0018] Furthermore, the future environmental data includes: environmental variable prediction data for future scenarios, including environmental prediction data and marine environmental data under different greenhouse gas emission scenarios.

[0019] Specifically, environmental variable prediction data for future scenarios usually involve long-term trend analysis of multiple factors such as climate, ecosystems, environmental pollution, and resource consumption.

[0020] Furthermore, the standardization process in step S2 specifically includes: using ArcGIS 10.8 platform to standardize the environmental variables, that is, resampling the study area, the spatial location and the future environmental data.

[0021] ArcGIS 10.8 is a professional geographic information system (GIS) software platform that aims to provide users with a comprehensive set of tools and services to achieve efficient processing, analysis, visualization and management of geographic data, and support cross-platform data interaction, analysis and deployment.

[0022] Furthermore, the creation of the HMSC model object in step S3 specifically includes: using the functions in the HMSC package to define the model, including selecting the model type and other model parameters.

[0023] When using the functions in the HMSC package to define a model, you can use function parameters to set and adjust various hyperparameters, structural parameters, etc. of the model. This parameterization method makes the model configuration flexible and easy to manage, and facilitates model tuning and comparative experiments.

[0024] Furthermore, the HMSC model object specifically includes: a presence-absence data model, a Poisson model for count data, and a log-normal Poisson model.

[0025] The presence-missing data model helps maintain the integrity of the data set and avoids the reduction of sample size and loss of information due to discarding observations with missing values. By properly filling or processing missing values, all available data can be fully utilized for analysis, improving the accuracy of the analysis.

[0026] The Poisson model for counting data is based on the Poisson distribution. It can not only describe the distribution characteristics of observed data, but also be used to predict the possibility of future events and provide a basis for decision-making.

[0027] The lognormal Poisson model handles overdispersion or underdispersion of the Poisson distribution, provides finer control and explanatory power, and can capture more complex variability in count data.

[0028] Furthermore, setting the calculation method in step S3 specifically means: setting the calculation method used for model estimation and the number of iterations and step size of the algorithm.

[0029] The number of iterations and step size are controlled to ensure the convergence, stability, generalization ability and computational efficiency of the model training process, to avoid overfitting and underfitting, and the ultimate goal is to find a model that can perform well on training data and have strong predictive ability for unknown data. The setting of parameters usually needs to be combined with factors such as specific problems, model types, data characteristics and computing resources, and the optimal or approximately optimal values ​​are determined through experimental adjustment or automatic parameter adjustment technology. In this invention, the parameters are set according to the HMSC model.

[0030] Furthermore, the step S5 also includes constructing a beta graph to display the estimated values ​​of the species niche parameters, analyzing the impact of different influencing factors on the Chinese white dolphin and bait fish according to the beta graph, and using random effects to reveal the correlation between species.

[0031] The different influencing factors come from different environmental factors in the future environmental data.

[0032] Beta plots are used to visualize a graphical representation of species niche breadth and niche overlap. Niche breadth reflects the range of resource utilization of a species in its living environment, while niche overlap indicates how similar different species are in their resource utilization.

[0033] The beta plot graphically displays the quantitative estimation results of niche parameters, which helps to understand the ecological relationships between species and the structure of the ecosystem.

[0034] The beta plot showing the estimated values ​​of species niche parameters is a statistical chart that intuitively displays the species niche breadth and niche overlap, which helps to understand and explain the role of sample points in the ecosystem, competitive relationships, and the maintenance mechanism of ecosystem functions.

[0035] Furthermore, the parameters of step S8 specifically include: biomass, production, bioconversion efficiency, consumption, trophic level, and other parameters that can be automatically estimated by the model without input.

[0036] The selection of parameters is based on the principle of energy input and output balance that needs to be followed in the Ecopath model, that is, the energy conservation formula: Input = Output.

[0037] Further, according to the relationship between the Chinese white dolphin and the bait fish species studied by the present invention, the specific formula for energy conservation is: the biomass of prey (Production volume / Prey biomass ) Bioconversion efficiency - net migration = predator biomass (consumption / Predator biomass ) Trophic level.

[0038] The energy conservation formula ensures that the total amount of energy flowing in the ecosystem is constant during the construction of the Ecopath model, that is, energy will not be created or disappear out of thin air, and all energy inputs must be balanced by corresponding output pathways. This principle ensures the biophysical consistency of the model, so that the results obtained through the Ecopath model analysis can reflect the true energy flow state of the ecosystem, providing a scientific basis for further ecological assessment and management decisions. Complying with the principle of energy conservation is the basis for building a reasonable and effective Ecopath model.

[0039] According to a second aspect of the present invention, a computer-readable storage medium is provided, on which one or more computer programs are stored. When the one or more computer programs are executed by a computer processor, any of the above methods is implemented.

[0040] The above one or more technical solutions in the embodiments of the present application have at least one of the following technical effects:

[0041] The present invention provides an analytical method for the interaction between marine organisms in the food chain. This comprehensive method helps to better understand the response and evolution of ecosystems. By combining the joint species distribution model and the Ecopath model to analyze the potential distribution of Chinese white dolphins and their forage fish under current and future environmental factors, the present invention can reveal the specific impacts of climate change and human activities on their distribution.

[0042] Secondly, by considering the interactions and energy flows between food chains, the present invention can predict the overall response and adaptability of the ecosystem when facing future scenarios. This cross-scale research method can not only provide a scientific basis for the demarcation of Chinese white dolphin protection areas, but also help formulate more effective management measures to mitigate the negative impacts of climate change and human activities on marine ecosystems. This method helps to better protect and restore marine biodiversity, ensure the sustainability of ecosystem services, and provide important support for the sustainable management of marine ecosystems.

[0043] The present invention can also provide a scientific reference for the protection and survival of other top predators, reduce the risk of food chain disruption for the survival of top predators, and provide protection for the safety of the marine ecosystem. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] The accompanying drawings are included to provide a further understanding of the embodiments and are incorporated into and constitute a part of this specification. The accompanying drawings illustrate the embodiments and are used together with the description to explain the principles of the present invention. It will be easy to recognize other embodiments and many expected advantages of the embodiments because they become better understood by reference to the following detailed description. The elements of the drawings are not necessarily to scale with each other. The same reference numerals refer to corresponding similar parts.

[0045] Figure 1 A schematic diagram of the potential distribution prediction and risk assessment process of Chinese white dolphins according to an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0046] The present application will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the relevant invention, rather than to limit the invention. It should also be noted that, for ease of description, only the parts related to the relevant invention are shown in the accompanying drawings.

[0047] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0048] Figure 1 FIG. 4 shows a schematic diagram of a potential distribution prediction and risk assessment process of a Chinese white dolphin according to an embodiment of the present invention. Figure 1 As shown,

[0049] Step S1, data collection: taking the Chinese white dolphins and bait fish in the study area as sample points, collecting the occurrence record data and spatial location data of the sample points, and collecting future environmental data as environmental variables.

[0050] The occurrence records include but are not limited to longitude and latitude, family name, species name, and peak season.

[0051] The Citespace model was used to identify the main risk factors affecting Chinese white dolphins and forage fish, including climate factors, marine environmental factors and human activities.

[0052] In this embodiment, the risk factors are determined based on "web of science" or "CNKI" as the basic data source, and the subject selection is: TS=((Chinese white dolphin) AND (bait fish OR feeding fish OR diet) AND (influencing factor OR environmental factor OR threat) AND (distribution or distribution prediction); TS=((chinese white dolphin orsousa chinese or Indo-pacific humpback dolphin) AND (fishes fed or bait fish)AND (influencing OR environment factors OR threats) AND (DistributionPrediction or Distribution) as the search statement, sorted by subject relevance, read the search records one by one, delete irrelevant items such as conference papers, conference reviews, conference notices, forewords, news reports, etc., each document contains information such as author, institution, keywords, abstract, publication date, etc., mainly journal papers, and the Refworks format file is output by the document management center and named download_XX.txt and imported into Citespace software for processing.

[0053] The future environmental data include: environmental variable prediction data for future scenarios, including climate data and marine environmental data under different greenhouse gas emission scenarios.

[0054] Future environmental data can be obtained from long-term climate trend analysis data stored in the databases of government meteorological departments (such as the National Meteorological Administration) or scientific research institutions, including climate data, marine environmental data, and human activity factors.

[0055] Climate factors include annual average temperature and annual precipitation; marine environmental data include coastline Euclidean distance, seabed topography elevation, slope, slope aspect, annual average sea surface temperature, annual sea surface temperature range, annual average salinity, annual average current speed, dissolved oxygen concentration, pH, and chlorophyll concentration. Human activity factors include the catch of bait fish, ship routes, and Euclidean distance to the waterway. In this embodiment, sea temperature, salinity, sea surface current speed, chlorophyll a, dissolved oxygen, and pH in 2050 and 2100 are used as research factors for future environmental data.

[0056] Step S2, data preprocessing: standardize the environmental variables, visualize the spatial location data, and remove data whose coordinates are not in the study area.

[0057] The standardization process specifically includes: using ArcGIS 10.8 platform to standardize the environmental variables, that is, resampling the study area, the spatial location and the future environmental data to ensure that the accuracy remains consistent.

[0058] The spatial location data of the species collected in step S1 are visualized and analyzed to eliminate data whose coordinates are not in the study area or whose species data are too small to represent the distribution data of the species.

[0059] Step S3, model training: the occurrence record data is organized into a format recognizable by the HMSC package in R Studio, wherein a specific data frame needs to be set, the data frame specifically includes: species data frame, environment data frame and space data frame, and an HMSC model object is created and a calculation method is set to generate a trained HMSC model.

[0060] The creating of the HMSC model object specifically includes: using the functions in the HMSC package to define the model, including selecting the model type and other model parameters.

[0061] In this embodiment, the environment data frame is taken as an example, and the data is converted into a specific data frame in R Studio. The relevant statements are as follows:

[0062] XData = data.frame( Route = da$study area, clim_1 = da$annual average temperature, clim_2 = da$annual precipitation, Marine_1 = da$Euclidean distance from coastline, Marine_2 = da$seafloor topography elevation, Marine_3 = da$slope, Marine_4 = da$slope aspect, Marine_5 = da$annual average salinity, Marine_6= da$annual average sea surface temperature, Marine_7 = da$annual sea surface temperature range, Marine_8 = da$annual average current speed, Marine_9 = da$dissolved oxygen concentration, Marine_10 = da$pH, Marine_11 = da$chlorophyll concentration, POP_1 = da$catch, POP_2 = da$route, POP_3 = da$Euclidean distance to waterway)

[0063] The HMSC model objects specifically include: presence-absence data model, Poisson model for count data, and lognormal Poisson model.

[0064] In this example, the presence-absence data model is selected to study the Chinese white dolphin and its bait fish, and the reference statement is as follows:

[0065] Y = as.matrix(da[,-c(1:n)])>0;

[0066] Use da[,-c(1:n)] to select all columns in the data frame except columns 1 to n. as.matrix converts this data frame to a matrix. This will generate a matrix consisting of logical values ​​(TRUE or FALSE), where TRUE means that the value at the corresponding position in the original data is greater than zero, and FALSE means that the value at the corresponding position in the original data is equal to zero.

[0067] If the weights of various environmental factors are determined in step S1, in this embodiment, the weights are all taken as 1, and the environmental factor formula setting can refer to the following statement:

[0068] XFormula =~ clim_1 + clim_2 + Marine_1 + Marine_2 + Marine_3 +Marine_4 + Marine_5 + Marine_6 + Marine_7 + Marine_8 + Marine_9 + Marine_10 +Marine_11 + POP_1 + POP_2 + POP_3;

[0069] To build a model, you can refer to the following statements:

[0070] m.FULL = Hmsc(Y = species matrix, XData = environmental data, XFormula = XFormula, distr = "probit", studyDesign = study area range, ranLevels = list(route=rL));

[0071] This statement includes only the model of environmental variables, without considering species trait data and phylogenetic relationships. Only the effects of environmental variables on the response variable are included. The ranLevels parameter specifies the hierarchy of random effects.

[0072] The setting of the calculation method specifically means: setting the calculation method used for model estimation and the number of iterations and step size of the algorithm.

[0073] The number of iterations can be 10, 100, or 1000. The more iterations there are, the more accurate the data will be. The experimental statements in R Studio in this embodiment are as follows:

[0074] nChains = 2 ##Set the number of MCMC chains to 2. Multiple chains are helpful to check the convergence of the MCMC algorithm.

[0075] nParallel = 2 ## (optional setting) The number of chains to run in parallel. This can improve computational efficiency but may be subject to hardware and software limitations.

[0076] samples = 100 ##Set the number of MCMC samples collected for each chain to 100.

[0077] for (thin in c(1, 10)) {

[0078] transient = 50 * thin

[0079] m = sampleMcmc(m, thin = thin, samples = samples, transient =transient, nChains = nChains, initPar = "fixed effects", nParallel =nParallel)

[0080] filename = file.path(model.directory, paste0("model_chains_",as.character(nChains), "_samples_", as.character(samples), "_thin_",as.character(thin)))

[0081] save(m, file = filename)

[0082] }

[0083] Code comments:

[0084] ## nParallel: Set the number of parallel computing chains to improve computing efficiency. In practical applications, this setting is limited by hardware (number of CPU cores) and software (support for parallel computing).

[0085] ## samples: Set the number of samples collected in the MCMC simulation for each chain to 100.

[0086] ## for loop: By looping to set different thin parameter values ​​(1 and 10), each thin value will affect the interval of samples. This allows the model performance and convergence to be tested at different thinning levels.

[0087] ## transient(burn-in): Calculate the number of burn-in iterations according to the value of thin to ensure that each run of the MCMC chain starts collecting data from a stable state.

[0088] ## sampleMcmc(): Call this function to apply the above parameter settings and run the MCMC simulation.

[0089] ## Save model: Save the MCMC model object of each run to a file. The file name contains the number of chains, number of samples and thin value, which is helpful for subsequent inspection and analysis.

[0090] Step S4, model fitting: the model is fitted using the fitting method provided by the HMSC package, the model fitting is evaluated based on the explanatory power, and the model fitting degree is evaluated based on the predictive ability of the two-fold cross validation.

[0091] In this example, the fit of the model is evaluated based on the explanatory power, and the reference statement is as follows:

[0092] preds = computePredictedValues(m) ##Calculate the predicted values ​​of the model and store these predicted values ​​in the preds variable.

[0093] MF = evaluateModelFit(hM=m, predY=preds) ##Evaluate the model's fit, that is, see how big the gap is between the model's predicted value and the actual observed value. The result of this function will be stored in the MF variable.

[0094] In this embodiment, the model fit is evaluated based on the prediction ability of two-fold cross validation, and the reference statement is as follows:

[0095] nParallel = 2

[0096] run.cross.validation = TRUE

[0097] filename=file.path(model.directory, paste0("CV_chains_",as.character(nChains),"_samples_",as.character(samples),"_thin_",as.character(m$thin)))

[0098] if (run.cross.validation) {

[0099] partition=createPartition(m,nfolds=2,column="route") preds=computePredictedValues(m,partition=partition,nParallel=nParallel)

[0100] MFCV=evaluateModelFit(hM=m,predY=preds)

[0101] save(partition,MFCV,file=filename)

[0102] }

[0103] else{

[0104] load(filename)

[0105] }

[0106] tmp=c(mean(MF$AUC),mean(MFCV$AUC),mean(MF$TjurR2),mean(MFCV$TjurR2))

[0107] names(tmp)=c("AUC","AUC(CV)","TjurR2","TjurR2(CV)")

[0108] Code comments:

[0109] ## Parallel computing settings: nParallel = 2 specifies the number of threads running in parallel to 2, which helps improve computing efficiency.

[0110] ## Cross validation condition: run.cross.validation = TRUE Set a condition to control whether to perform cross validation.

[0111] ## File path and naming: filename is used to define the path and name of the file that saves the cross-validation results, including the number of chains, number of samples, and thin value.

[0112] ## Execution of cross validation: If run.cross.validation is true, execute the statements in if.

[0113] ## Create partitions: createPartition(m, nfolds=2, column="route") Create partitions for cross-validation based on the route column, divided into two folds.

[0114] ## Calculate predicted values: computePredictedValues(m, partition=partition,nParallel=nParallel) Calculates model predicted values ​​using the given partition and parallel settings.

[0115] ## Model fitting evaluation: evaluateModelFit(hM=m, predY=preds) Evaluate the model fitting effect under cross validation and save the results.

[0116] ## Save results: Use the save() function to save the partition and model fitting evaluation results to a file.

[0117] ## Load results: If not running cross validation, load results from a previously saved file.

[0118] ## Performance metric calculation: Calculate the average of model performance metrics, such as AUC and Tjur's R², and name the resulting vector.

[0119] The AUC value refers to the area under the receiver operating characteristic curve (ROC) curve (area under curve, AUC), which ranges from 0 to 1 and reflects the classification effectiveness of the model prediction. Generally speaking, the closer the AUC is to 1, the better the performance of the model. Tjur's R² is an indicator that measures the effectiveness of the model in predicting positive and negative examples. The value range is between -1 and 1, where 1 indicates the ability to perfectly predict positive and negative examples, 0 indicates that the model's prediction effect is equivalent to random guessing, and -1 indicates that the model is completely opposite.

[0120] Step S5, influencing factors: using the fitting results of the model in step S4, analyze the relationship between the Chinese white dolphin and bait fish species, including symbiotic relationships and competitive relationships, and analyze and explain the spatial pattern of species distribution.

[0121] The step S5 also includes constructing a beta graph to display the estimated values ​​of the species niche parameters. In this embodiment, the reference statement is as follows:

[0122] postBeta=getPostEstimate(m, parName="Beta") ##Get the posterior estimate of the model parameter Beta and store the result in the postBeta variable.

[0123] plotBeta(m, post = postBeta, plotTree = TRUE, spNamesNumbers = c (FALSE, FALSE)) ##Draw a graph of the Beta parameter.

[0124] According to the obtained Beta graph, the impact of different influencing factors on Chinese white dolphins and bait fish is analyzed. It is mainly divided into positive and negative sides. For most species, the linear response to spring temperature is mainly positive, which means that the probability of occurrence of most species reaches its peak under warm climate conditions. If the extreme value is outside the "climate" range of the study, it indicates that there is a nonlinear increase or decrease response within the range of the study. If you need to get an estimate of the peak, you can use the posterior mean of postBeta to get a rough estimate.

[0125] Understanding how environmental factors affect the distribution of fish is essential for developing sustainable fishery management plans. This information can help optimize fishing seasons and areas to reduce impacts on fish populations and ecosystems. After determining the impact of each factor, the top few factors with the largest impact can be screened out. By collecting data on their future changes, it can be used to analyze the potential distribution of Chinese white dolphins and forage fish, and predict the risks and opportunities that Chinese white dolphins may face.

[0126] In this example, random effects are used to reveal the association between species, and the reference statement is as follows:

[0127] OmegaCor = computeAssociations(m) ##Calculate the associations between species.

[0128] supportLevel=0.95 ##The confidence interval is set to 95%.

[0129] toPlot=((OmegaCor[[1]]$support>supportLevel)+(OmegaCor[[1]]$support<(1-supportLevel))>0)*OmegaCor[[1]]$mean ##The values ​​to be plotted are calculated. The associated mean is plotted only when the associated support is greater than supportLevel or less than 1-supportLevel.

[0130] toPlot=sign(toPlot)

[0131] plotOrder=corrMatOrder(OmegaCor[[1]]$mean,order="AOE") ##Sort the species so that clusters of species with co-occurrence relationships are easiest to observe.

[0132] corrplot(toPlot[plotOrder,plotOrder],method="color",tl.cex=0.5,col=colorRampPalette(c("blue", "white", "red"))(255)) ##Draw a correlation graph. The colors of the correlation graph indicate the association between species: red indicates positive association and blue indicates negative association.

[0133] Due to the reduction of fishery resources, the diet of Chinese white dolphins has also changed greatly. Understanding the interactions between species can help managers develop more effective ecological restoration plans. For example, clarifying the food sources of Chinese white dolphins is of great guiding significance for selecting fish for value-added release.

[0134] Step S6, model prediction: input the future environmental data into the trained HMSC model in step S3 to predict the potential distribution, perform further processing based on the prediction results, and analyze the position changes between the species hotspot distribution area and the current distribution area.

[0135] The prediction results are further processed, such as converting the probability into a prediction of occurrence / non-occurrence, or converting the prediction value into a more meaningful indicator (such as species abundance). The reference statement in this embodiment is as follows:

[0136] Gradient = prepareGradient(m, XDataNew = XData.grid , sDataNew = list(Route = xy.grid)) ##The function is used to prepare the gradient information required for prediction on new environment and spatial data. Among them, XDataNew and sDataNew specify the new environment variable data and spatial data respectively.

[0137] nParallel=2

[0138] predY=predict(m,Gradient=Gradient,expected=TRUE,nParallel=nParallel)##Used for model prediction, expected=TRUE means that the predicted value is the expected value of the probability of species occurrence.

[0139] EpredY=Reduce("+",predY) / length(predY) ##Calculate the average of the prediction results. The Reduce function merges all elements in the predY list by addition.

[0140] dim(EpredY) ##Returns the dimension of EpredY, possibly for debugging or validating data structures.

[0141] Cm=EpredY[,50] ##Extract the 50th column from the EpredY matrix, representing the predicted occurrence probability of species numbered 50.

[0142] S=rowSums(EpredY) ##Calculate the species richness at each predicted location.

[0143] CWM=(EpredY%*%m$Tr) / matrix(rep(S,m$nt),ncol=m$nt) ##Calculate the community weighted average features for each predicted position.

[0144] xy=grid[,1:2] ##Extract the first two columns from the grid data frame as coordinate information.

[0145] M=XData.grid$Marine environmental factors## means only taking marine factors.

[0146] C=grid$climate factors## means only taking climate factors.

[0147] mapData=data.frame(xy,C,S,Cm,CWM,M,stringsAsFactors=TRUE) ##Combine all variables into a new data frame for subsequent analysis or display.

[0148] Based on the prediction results of potential distribution, the location changes of species hotspot distribution areas and current distribution areas are analyzed.

[0149] In this example, the current distribution range of Chinese white dolphins in Xiamen Bay is ; Under the influence of global change, the potential distribution range changes to ; Using GIS for overlay analysis, we can know and The overlapping area is The formula is:

[0150] (1) When >0, Δa=( - ) / .

[0151] If Δa>=30%, it means that the potential distribution range has expanded significantly and the protected area needs to be expanded.

[0152] If Δa<30% and Δa>=0, the protected area can remain basically unchanged.

[0153] If Δa<0, that is, the potential distribution range is smaller than the current distribution range, the protected area can be reduced.

[0154] In addition, the specific locations of potential distribution inside and outside protected areas need to be considered. Even if the overall potential distribution range does not increase, some new areas may become hotspots.

[0155] (2) When =0, indicating that there is no intersection between the current distribution and the potential distribution. It is necessary to consider establishing ecological corridors to connect these scattered habitats, as well as replanning human activities such as fishing activities and offshore construction to ensure that no additional threats are posed to white dolphins.

[0156] Step S7, result display: using GIS analysis to visualize the location changes and display model results, the model results including potential distribution maps of species and importance scores of the environmental variables;

[0157] Visualization presents data relationships, patterns, outliers, etc. in a graphical way, which helps to reveal the deep connections, trends and laws hidden behind a large amount of data. It can help users discover the structure, distribution characteristics, anomalies, etc. of data, provide deep insights into data, and promote decision-making and problem solving.

[0158] Step S8, overall analysis: Use the relationship between species obtained in step S5 to perform functional division on bait fish, collect biological information of fish as parameters, input the parameters into the Ecopath model, analyze the energy conversion between trophic levels, and obtain the transmission efficiency.

[0159] In this embodiment, the functions are divided as follows: a Chinese white dolphin; b pelagic fish; c demersal fish; d large demersal fish; e large benthic fish; f small benthic fish; g zooplankton fish; h omnivorous fish; i cephalopods; j stomatopods; k crabs; l shrimps; m bivalves; n carnivorous zooplankton; o herbivorous zooplankton; p phytoplankton and q organic debris.

[0160] The Ecopath model needs to comply with the principle of energy input and output balance, the formula is: prey biomass (Production volume / Prey biomass ) Bioconversion efficiency - net migration = predator biomass (consumption / Predator Biomass ) Trophic level.

[0161] The parameters include: biomass, production, bioconversion efficiency, consumption, trophic level, and other parameters that are not input can be automatically estimated by the model. The trophic level is a required input parameter.

[0162] In step S8, it is necessary to define management strategies and climate change scenarios, and determine management measures that may affect the interaction between ecosystem components, such as fishery quotas, protected area establishment, pollution control, etc. According to the analysis of influencing factors in S5, select the main influencing factors. Then adjust the relevant parameters to simulate the biomass, predation rate or mortality rate of Chinese white dolphins under different scenarios. For each scenario, run the Ecopath model to simulate its potential impact on ecosystem structure and function. Then analyze and compare the results to evaluate the ecosystem response. Analyze the output results of the model under different scenarios, evaluate the impact of management strategies and climate change on the interaction between ecosystem components, energy flow and trophic level structure, and identify which management measures can effectively enhance the resistance and resilience of the ecosystem. Finally, formulate adaptive management measures. Based on the results of model analysis, formulate management measures aimed at enhancing ecosystem adaptability and mitigating the negative impact of climate change, and continuously monitor ecosystem responses, adjust management strategies according to monitoring results and new information, and achieve adaptive management.

[0163] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0164] The modules involved in the embodiments of the present application may be implemented by software or by hardware.

[0165] The above description is only a preferred embodiment of the present application and an explanation of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the present application is not limited to the technical solution formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above invention concept. For example, the above features are replaced with the technical features with similar functions disclosed in this application (but not limited to) by each other to form a technical solution.

Claims

1. A method for predicting the potential distribution and risk assessment of Chinese white dolphins, characterized in that: The following steps are involved: S1. Take the Chinese white dolphins and bait fish in the study area as sample points, collect the occurrence record data and spatial location data of the sample points, including the latitude and longitude, family name, species name and peak season of the Chinese white dolphins and bait fish, and collect future environmental data as environmental variables; S2. Using ArcGIS 10.8 platform to perform resampling and standardization processing on the environmental variables, visually analyzing the spatial location data, and eliminating data whose coordinates are not in the study area; S3, organizing the occurrence record data into a format that can be recognized by the HMSC package in R Studio, wherein a specific data frame needs to be set, and the data frame specifically includes: a species data frame, an environmental data frame, and a spatial data frame, and creating an HMSC model object and setting a calculation method, using the presence-absence data model to generate a trained HMSC model, wherein the setting of the calculation method specifically means: setting a calculation method for model estimation and the number of iterations and step size of the algorithm, and the HMSC model adopts an MCMC model, wherein the MCMC chain is equal to 2, and the set chains are calculated in parallel, and the MCMC model object of each run is saved to a file, and the file name contains the number of chains, the number of samples, and the thin value; S4. Use the fitting method provided by the HMSC package to fit the model, evaluate the model fit according to the explanatory power, specifically use the evaluateModelFit() function to evaluate the model fit, and evaluate the model fit according to the prediction ability of the two-fold cross validation, wherein AUC and Tjur's R2 are used to calculate the performance indicators, and use the file containing the number of chains, the number of samples and the thin value as the model fitting file; S5. Analyze the relationship between the Chinese white dolphin and bait fish species using the fitting results of the model in step S4, including symbiotic relationships and competitive relationships, specifically including: constructing a beta graph to display the estimated values ​​of the species niche parameters, analyzing the impact of different influencing factors on the Chinese white dolphin and bait fish according to the beta graph, and using random effects to reveal the association between species, wherein the specific implementation of the random effect also includes: setting the confidence interval to 95%, and the beta graph will only draw the mean of the association when the support of the association is greater than the confidence interval or less than 1-confidence interval; S6, inputting the future environmental data into the trained HMSC model in step S3 to predict the potential distribution, and further processing according to the prediction results, including converting the prediction results into results of occurrence or non-occurrence, and analyzing the position changes between the species hotspot distribution area and the current distribution area, wherein the specific steps of prediction include: setting the gradient information required for prediction on the new environmental and spatial data, and the expected value of the occurrence probability to be predicted, calculating the average value of the prediction results and returning the dimension of the average value, extracting the selected numbered columns from the average value matrix as the predicted occurrence probability of the species, calculating the species richness of each predicted position, and calculating the community weighted average characteristics of each predicted position, extracting the first two columns from the selected data frame as coordinate information, then extracting marine environmental factors and climate factors, and combining the coordinate information, the climate factors, the species richness, the predicted occurrence probability of the species, the community weighted average characteristics and the marine environmental factors to generate a new data frame for further processing; S7. Visualize the position change using GIS overlay analysis to display model results, where the model results include potential distribution maps of species and importance scores of environmental variables, wherein the GIS overlay analysis specifically includes: Δa=(A2-A1) / A1,A3>0 Among them, A1 represents the current distribution range of Chinese white dolphins in the designated area, A2 represents the potential distribution range of Chinese white dolphins driven by global changes, and A3 represents the overlapping range of A1 and A2; If Δa>=30%, it means that the potential distribution range has significantly expanded, and the protection scope of the designated area needs to be expanded; If Δa<30% and Δa>=0, it means that the specified area can remain basically unchanged; If Δa<0, it means that the potential distribution range is smaller than the current designated area, and the protection range of the designated area needs to be reduced; If A3=0, it means that the current distribution and the potential distribution have no intersection; S8. Use the relationship between species obtained in step S5 to functionally divide the bait fish, collect the biological information of the fish as parameters, input the parameters into the Ecopath model, analyze the energy conversion between each trophic level, obtain the transmission efficiency, and use the Ecopath model to simulate the implementation effects of different management strategies for the existing management strategies for Chinese white dolphins and bait fish, and define adaptive management strategies. The Ecopath model needs to comply with the principle of energy input and output balance, and the formula is: prey biomass × (production / prey biomass) × biological conversion efficiency - net migration = predator biomass × (consumption / predator biomass) × trophic level.

2. The method according to claim 1, characterized in that: The step S1 also includes: determining the main risk factors affecting the Chinese white dolphin and bait fish through the Citespace model, including climate factors, marine environmental factors and human activity factors.

3. The method according to claim 1, characterized in that: The future environmental data include: environmental variable prediction data of future scenarios, and the environmental variable prediction data include climate data and marine environmental data under different greenhouse gas emission scenarios.

4. The method according to claim 1, characterized in that: The step S3 of creating the HMSC model object specifically includes: using the functions in the HMSC package to define the model, including selecting the model type, and the HMSC model object specifically includes: presence-absence data model, Poisson model for count data, and lognormal Poisson model.

5. The method according to claim 1, characterized in that: The different influencing factors come from different environmental factors in the future environmental data.

6. The method according to claim 1, characterized in that: The parameters of step S8 specifically include: biomass, production, bioconversion efficiency, consumption and trophic level.

Images