Ecopath model-based food web assessment method for large water surface ecosystem

Abstract

The invention relates to the technical field of ecological system evaluation, and discloses a large water surface ecological system food web evaluation method based on an Ecopath model, which comprises the steps of model construction and parameter functionalization: constructing a food web topological structure of an ecological system, and establishing a core balance equation based on the Ecopath model; defining at least one key process parameter in the core balance equation as a dynamic parameter response function related to an environmental factor, wherein the dynamic parameter response function comprises a response function coefficient to be solved; a data acquisition step: in the ecological system, monitoring an environmental factor corresponding to the dynamic parameter response function to obtain dynamic monitoring data of the environmental factor; meanwhile, pulse type stable isotope labeling is carried out on primary producers of the ecological system. Key process parameters in an Ecopath model are innovatively defined as dynamic parameter response functions related to environmental factors, and the fundamental limitation that a traditional model can only conduct static snapshot type evaluation is overcome.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of ecosystem assessment, and particularly to a method for evaluating food web of large water surface ecosystem based on Ecopath model. BACKGROUND

[0002] Large water surface ecosystems, such as lakes, reservoirs and near-shore bays, are lifelines for maintaining regional ecological balance, ensuring drinking water safety and supporting fishery economy. In order to manage and develop these important natural resources in a sustainable manner, decision-makers must have a deep understanding of the complex food web structure and energy flow rules within the ecosystem, which is a fundamental prerequisite for formulating scientific and reasonable fishery catch limits, assessing the risk of invasive species, and predicting the impact of water environmental changes. Therefore, the present application aims to provide a technical method for accurately and dynamically evaluating the state of the food web of large water surface ecosystems, and to provide a reliable scientific basis for the protection and management of the ecosystem.

[0003] At present, one of the widely used technical means for food web quantitative assessment in the field is the Ecopath model. As an internationally recognized standardized analysis tool, the main advantage of this method is that it can integrate scattered and diverse ecological data within a region into a unified mathematical framework based on mass balance principles. By constructing a static snapshot of the food web, the Ecopath model can clearly show the overall structure of the ecosystem at a specific period, and calculate a series of macro-comprehensive indicators for comparing the health status or maturity of different ecosystems, providing an effective platform for researchers to comprehensively and structurally understand and analyze complex ecosystems.

[0004] However, due to the reliance on static snapshots in the methodology of the existing technology, it has exposed several inherent defects that cannot be overcome when applied to dynamic real ecosystems. First, its static framework based on a set of fixed parameters fundamentally cannot capture and simulate the dynamic response of the ecosystem to changes in key environmental factors such as seasonal water temperature and light, leading to a disconnection between the evaluation results and reality. Second, the large number of key parameters required by the model often rely on literature data with questionable applicability or subjective "balance" tuning, which greatly reduces the objectivity and site specificity of the evaluation results. Finally, the traditional method usually gives a single and deterministic evaluation conclusion, but cannot quantify and convey the uncertainty of the final results caused by the uncertainty of the input parameters, creating an illusion of precision for decision-makers. Therefore, the technical problem to be solved by the present application is how to break through the static and deterministic framework of the existing technology and establish a new food web evaluation method that can endogenously reflect the dynamic mechanism of the ecosystem, be constrained by objective process data, and provide probabilistic evaluation results. SUMMARY

[0005] In view of the deficiencies of the prior art, the present application provides a method for evaluating a food web of a large water surface ecosystem based on an Ecopath model, which solves the problem that the prior art food web evaluation method cannot objectively reflect the dynamic response of an ecosystem to environmental changes due to its reliance on static and highly uncertain parameters, and the evaluation results lack reliable uncertainty quantification.

[0006] To achieve the above object, the present application is implemented by the following technical solutions: a method for evaluating a food web of a large water surface ecosystem based on an Ecopath model, comprising the following steps: S1, model construction and parameter functionization step: constructing a food web topological structure of the ecosystem and establishing a core balance equation based on an Ecopath model; defining at least one key process parameter in the core balance equation as a dynamic parameter response function related to an environmental factor, the dynamic parameter response function containing a response function coefficient to be solved; S2, data acquisition step: monitoring the environmental factor corresponding to the dynamic parameter response function in the ecosystem to obtain environmental factor dynamic monitoring data; at the same time, pulse stable isotope labeling is performed on the primary producers of the ecosystem, and dynamic sampling and analysis are performed on a plurality of functional groups in the food web to obtain isotope dynamic observation data; S3, parameter inversion and calibration step: using a Bayesian inversion method, taking the isotope dynamic observation data as observation evidence and the environmental factor dynamic monitoring data as model driving, solving the response function coefficient of the dynamic parameter response function to obtain the posterior probability distribution of the response function coefficient; S4, ecosystem evaluation step: based on the posterior probability distribution of the response function coefficient, probabilistically evaluating the food web of the ecosystem.

[0007] Preferably, the key process parameters in the step S1 include unit biomass consumption rate and unit biomass production rate of the functional groups; and the dynamic parameter response function is used to represent the response mechanism of the key process parameters to the dynamic changes of the environmental factor.

[0008] Preferably, the parameter inversion and calibration step in the step S3 specifically comprises: setting a prior probability distribution for the response function coefficient to be solved; establishing a dynamic isotope flow model, which can generate a theoretical isotope dynamic curve according to a given set of response function coefficients and the environmental factor dynamic monitoring data; constructing a likelihood function for quantifying the consistency degree between the theoretical isotope dynamic curve and the isotope dynamic observation data; The posterior probability distribution of the response function coefficients is obtained by a Markov Chain Monte Carlo algorithm based on the prior probability distribution and the likelihood function.

[0009] Preferably, the dynamic isotope flow model is used to describe the transfer process of isotopes in the food web, and the unit mass growth rate and tissue metabolic turnover rate in the model are driven by the dynamic parameter response function combined with the dynamic monitoring data of the environmental factors.

[0010] Preferably, the pulsed stable isotope labeling in step S2 is a one-time addition of stable isotope tracer to the ecosystem at the start of the experiment; and the dynamic sampling is sample collection of the functional groups at multiple different time points after labeling.

[0011] Preferably, the environmental factor is at least one of water temperature, dissolved oxygen or light intensity.

[0012] Preferably, the posterior probability distribution of the response function coefficients obtained in step S3 is a probability distribution form containing the expected value and confidence interval of the response function coefficients.

[0013] Preferably, the ecosystem evaluation step in step S2 specifically includes: According to a given environmental scenario, the posterior probability distribution of the response function coefficients is used to calculate and generate the probability distribution of the macroscopic properties of the ecosystem, including total system throughput, recycling index or average trophic level.

[0014] Preferably, the core balance equation in step S1 is represented as: the yield of a functional group is equal to the sum of the total amount of being preyed on by other functional groups in the model, the total amount of being captured, the amount of other natural death, the net amount of emigration and the amount of biomass accumulation.

[0015] The large water surface ecosystem food web evaluation system based on the Ecopath model includes: A model construction module is configured to construct the food web topology of the ecosystem and establish a core balance equation containing a dynamic parameter response function, wherein the dynamic parameter response function contains response function coefficients to be solved; A data acquisition module is configured to acquire dynamic monitoring data of environmental factors and isotope dynamic observation data obtained by pulsed stable isotope labeling; A parameter inversion calibration module is configured to perform Bayesian inversion, wherein the isotope dynamic observation data is a constraint, the dynamic monitoring data of the environmental factors is a driving force, the response function coefficients of the dynamic parameter response function are solved, and a posterior probability distribution of the response function coefficients is generated; an ecosystem assessment module for probabilistically assessing a food web of the ecosystem based on the posterior probability distribution of the response function coefficients.

[0016] The present application provides a method for assessing a food web of a large water surface ecosystem based on an Ecopath model. 1. The present application overcomes the fundamental limitation of traditional models that can only perform static snapshot assessment by innovatively defining the key process parameters in the Ecopath model as dynamic parameter response functions related to environmental factors. This design enables the model to inherently and mechanistically reflect the dynamic response of the ecosystem to the external environment, thereby greatly improving the simulation fidelity and realism of the model to real ecological processes, and making the assessment results no longer static numbers that are independent of the environmental background.

[0017] 2. The present application obtains isotopic dynamic observation data reflecting real processes by using pulsed stable isotope labeling, and solves the model parameters by combining with the Bayesian inversion method, thereby getting rid of the excessive dependence on literature values, expert experience or subjective balancing operation in traditional methods. This inversion calibration process strongly constrained by process data measured on site ensures the objectivity, traceability and site specificity of the assessment results, and significantly enhances the scientific credibility of the assessment conclusions.

[0018] 3. The present application uses the Bayesian inversion framework, and the final output is not a single, determined parameter value, but a posterior probability distribution containing complete uncertainty information. In the final ecosystem assessment step, all assessment results, including key parameters and system macroscopic properties, can be presented in the form of probability distribution or Bayesian confidence interval, thereby realizing the full quantification of the uncertainty of the assessment results and providing more robust and comprehensive information support for managers to make scientific decisions based on risk. BRIEF DESCRIPTION OF DRAWINGS

[0019] Figure 1 is a flow chart of the method of the present application; Figure 2 is a system architecture diagram of the present application. DETAILED DESCRIPTION

[0020] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor fall within the scope of the present application.

[0021] Embodiment: Please refer to the accompanying Figure 1The embodiment of the present application provides a large water surface ecosystem food web evaluation method based on an Ecopath model, and comprises the following steps: S1, model construction and parameter functionization step: constructing a food web topology structure of an ecosystem, and establishing a core balance equation based on an Ecopath model; at least one key process parameter in the core balance equation is defined as a dynamic parameter response function related to an environmental factor, and the dynamic parameter response function comprises to-be-solved response function coefficients; This step aims to construct a mathematical model framework capable of endogenously reflecting the dynamic characteristics of an ecosystem for subsequent parameter inversion calibration and ecosystem evaluation. The framework is based on the Ecopath model known in the art, and the expression mode of the core parameters is innovatively improved in essence.

[0022] Firstly, the construction of the food web topology structure is performed. This process is based on prior knowledge, historical literature and preliminary field investigation of the target large water surface ecosystem, identifies key biological groups in the ecosystem, and divides them into a limited number of functional groups. The division principle of the functional groups can be based on species, feeding habits, ecological niche or their specific role in energy flow. Subsequently, according to the predator-prey relationship between the functional groups, the food web topology structure describing the material or energy flow path is preliminarily established, and the structure is usually represented in the form of a food composition matrix.

[0023] On the basis of the food web topology structure, the core balance equation describing the energy or material balance of the ecosystem is established. The equation follows the principle of energy conservation, that is, for any functional group in the food web, its total production should be equal to the sum of its total consumption in the system and all other output items. For any functional group The core balance equation thereof can be summarized as follows: ; In the formula, represents the total production of the functional group ; represents the total amount of the functional group being preyed on by all predator functional groups ; represents the total amount of the functional group being fished; represents all natural death amounts other than predation in the model; represents the net emigration amount of the functional group ; represents the biomass accumulation amount of the functional group .

[0024] To overcome the fundamental limitation of the above balance equation in the prior art, which relies on a set of static, isolated parameter values and thus fails to reflect the dynamic response of the ecosystem under real environmental fluctuations, the present application introduces a key innovation, i.e. dynamic functionalization of the key process parameters.

[0025] Instead of treating the key process parameters reflecting the rates of biological physiology and ecological processes, such as the unit biomass consumption rate or the unit biomass production rate , as constants that do not change over time, the present application innovatively defines them as one or more dynamic parameter response functions that are real-time related to external environmental factors. This design enables the intrinsic mechanism of the model to be directly coupled with observable, dynamically changing environmental conditions.

[0026] The general form of the dynamic parameter response function can be expressed as: ; wherein: is the instantaneous value of the th key process parameter of the functional group at time . is a vector composed of one or more environmental factors that are real-time monitored at time , which can include water temperature, dissolved oxygen, light intensity, etc. is a preset function form used to describe the response relationship between the parameter and the environmental factor. The selection of this function form can be based on ecological mechanism or empirical model, which is not limited to a specific form and can be an exponential function, a polynomial function, a logistic function or other suitable functions that can be foreseen by those skilled in the art. is a set of response function coefficients associated with the function , which determines the specific form of the response curve, such as the basal rate and the change sensitivity, etc.

[0027] Through the above functionalization, the technical goal of the present application has undergone a fundamental change: instead of estimating static parameter values, it is changed to solving a set of response function coefficients that can completely define the dynamic response mechanism of the parameter.

[0028] The exponential function form can be used for the functionalization of the feeding rate of poikilothermic organisms such as fish, i.e. the unit biomass consumption rate , which is significantly affected by water temperature. The Q10 temperature coefficient model can be used to express its relationship with water temperature the relationship between the dynamic parameter and the environmental factor. This is based on the general ecological understanding that, within a certain range, the metabolic rate of an organism will exponentially increase with temperature. At this point, the specific form of the dynamic parameter response function may be an exponential function, but the present application is not limited to this specific form.

[0029] In this way, the final output of step S1 is no longer a rigid model that depends on a large number of uncertain static parameters, but a flexible model framework with an internal dynamic response mechanism, ready to accept subsequent data constraints and calibration. The core unknown of this framework has changed from multiple independent parameter values to a smaller number of response function coefficients that better reflect the essential laws of the ecosystem, laying the foundation for accurate and robust solutions through Bayesian inversion in subsequent steps.

[0030] S2, data collection step: in the ecosystem, the environmental factors corresponding to the dynamic parameter response function are monitored to obtain dynamic monitoring data of the environmental factors; at the same time, the primary producers in the ecosystem are pulsed stable isotope labeled, and multiple functional groups in the food web are dynamically sampled and analyzed to obtain isotope dynamic observation data; The core purpose of this step is to obtain two sets of high-resolution time series data sets that are synchronized with each other: environmental factor dynamic monitoring data reflecting external environmental stress, and isotope dynamic observation data representing the real process of material flow within the ecosystem. These two sets of data together constitute the objective basis and data driving force for parameter inversion and calibration in subsequent steps.

[0031] This step first involves the acquisition of environmental factor dynamic monitoring data. This process aims to provide real-time and quantifiable input for the dynamic parameter response function constructed in step S1. The implementation is to deploy one or more environmental monitoring devices in the study area of the target large water surface ecosystem. These devices are used to continuously or at high temporal frequency to monitor the environmental factors determined in step S1 that are significantly related to key process parameters.

[0032] The environmental factors can include but are not limited to water temperature, dissolved oxygen, pH value, light intensity, or key nutrient salt concentration. The raw data generated by the monitoring is processed to form one or more time series data sets, i.e. environmental factor dynamic monitoring data The acquisition of this data set is the premise and foundation for ensuring that the model can capture and simulate the dynamic response of the ecosystem to real environmental fluctuations.

[0033] The key innovation of this step is to obtain the isotopic dynamic observation data through an active and controlled field experiment. This process aims to generate a signal that can be accurately tracked and record the complete kinetic process of the signal flowing in the food web, thus providing direct and process-based observation evidence for the internal process parameters of the inversion model that are difficult to measure directly.

[0034] The process further comprises: First, pulsed stable isotope labeling is performed. This operation is to add one or more stable isotope tracers to a representative area of the study water body in a pulsed or one-time manner at the initial moment of the experiment (defined as ). The representative area here can be a controlled enclosure system that can simulate the surrounding water environment, or a specific bay with relatively slow water exchange, to ensure that the tracers can be fully absorbed by primary producers in the initial stage and reduce signal loss caused by rapid dilution.

[0035] The stable isotope tracer can be an inorganic salt or an organic compound containing high-abundance carbon isotopes or nitrogen isotopes. The purpose is to create a significant isotope signal pulse higher than the natural background abundance in the energy basis of the food web, i.e., the biomass of primary producers.

[0036] Second, dynamic sampling and analysis of the food web are performed. After completing the pulsed labeling, a pre-set sampling plan covering the entire experimental period is started immediately. The plan requires the collection of biological samples from all or key functional groups defined in step S1 at multiple different time points (where is the total number of sampling times). The time interval of sampling needs to be carefully designed, with an initial intensive and later sparse, to ensure that the rapid rising stage, peak platform, and subsequent attenuation and dilution stage of the isotope signal transmission between different trophic levels can be captured.

[0037] After standard pretreatment, the collected biological samples of each functional group are measured by professional analysis instruments.

[0038] Through the above analysis, the isotopic ratio of each functional group at each sampling time point is finally obtained. All these data points together constitute the isotopic dynamic observation data set . This data set directly and intuitively depicts how the externally introduced substances are transmitted and redistributed among the various components of the ecosystem through the predator-prey relationship over time in a quantitative manner.

[0039] In summary, step S2 successfully generates two sets of time series data that are interrelated: environmental factor dynamic monitoring data describing the environment in which the system is located, and isotope dynamic observation data describing the internal response of the system. The synchronization and high resolution characteristics of the two sets of data provide an indispensable data basis for establishing a model that can reflect the real ecological process in subsequent steps, and effectively constraining and calibrating it.

[0040] S3, parameter inversion calibration step: using the Bayesian inversion method, the isotope dynamic observation data is used as observation evidence, and the environmental factor dynamic monitoring data is used as model driving to solve the response function coefficient of the dynamic parameter response function, and obtain the posterior probability distribution of the response function coefficient; This step is the core calculation and reasoning link of the method of the application, and its fundamental purpose is to establish a solid bridge between the macro ecological model and the micro process observation data, and to solve the parameter response function coefficient that best represents the dynamic mechanism of the ecological system under the constraint of the real data through a rigorous statistical inversion framework.

[0041] This step innovatively uses the statistical idea of Bayesian inversion. The essence of this idea is to regard the unknown parameters to be solved, i.e. the response function coefficients defined in step S1 and other model parameters with uncertainty, as random variables, and use the objective observation data obtained in step S2 as evidence to update the probability distribution of these parameters, thereby obtaining more accurate and uncertainty-including posterior knowledge.

[0042] According to Bayes' theorem, the posterior probability of the parameter is proportional to the product of its likelihood and prior probability, which can be generally expressed as: ; In the formula: represents the set vector of all unknown parameters to be solved in the model. The core of this set vector is the response function coefficient vector defined in step S1, which describes the dynamic response mechanism of all key process parameters In addition, in order to complete the model, other parameters with greater uncertainty in prior knowledge, such as partial food composition ratio or background mortality rate, can also be included. represents the prior probability distribution of the parameter set . represents the likelihood function of observing real data under the condition that the parameter set is given. represents the updated posterior probability distribution of the parameter set after the observation data is obtained, which is the final solution target of this step.

[0043] To achieve the above-mentioned Bayesian inversion, the step specifically comprises the following interrelated links: First, set the prior probability distribution of the parameters . The distribution is an initial, probabilistic description of the unknown parameters based on existing knowledge in the field, relevant literature reports or background investigation data obtained in step S2 before any data fitting. For the coefficients describing the temperature response, a normal distribution or a uniform distribution with a large variance centered on physiological common sense can be set. Setting a wide prior with low information content can enable subsequent parameter inference to be more driven by objective observation data.

[0044] Second, construct the likelihood function . The likelihood function is a mathematical link connecting the model theoretical prediction and the real world observation, which quantitatively evaluates the consistency between the theoretical results generated by the model under the condition of a given set of specific parameters and the isotope dynamic observation data collected in step S2.

[0045] To calculate the likelihood, a forward model is needed that can generate theoretical prediction values according to any given parameter set . For this purpose, the present method establishes a dynamic isotope flow model. The model is a mathematical system composed of ordinary differential equations, which is used to simulate the transmission, dilution and enrichment of stable isotopes between functional groups in the food web over time after pulse labeling. The internal dynamics of the isotope flow model, the growth, feeding and metabolic rate of organisms, are directly driven by the dynamic parameter response function defined in step S1. The simulation process of the model requires the response function coefficients to be solved and the environmental factor dynamic monitoring data obtained in step S2 as double inputs, so as to calculate the theoretical isotope dynamic curve .

[0046] After obtaining the theoretical curve, the likelihood can be calculated by comparing it with the real observation value . Assuming that the observation error follows a normal distribution, the likelihood function can be expressed as the connected product of the probability density functions on all data points. However, the present application is not limited to this specific error distribution assumption.

[0047] Finally, the posterior probability distribution is solved. Due to the high nonlinearity of the model and the complexity of the parameter space, the posterior probability distribution usually cannot be obtained analytically. This step uses advanced numerical calculation methods for solving. Markov chain Monte Carlo algorithm can be used, or other equivalent numerical sampling algorithms that can be foreseen by those skilled in the art.

[0048] MCMC algorithm generates a series of parameter samples that conform to the posterior probability distribution by performing an intelligent random walk in the parameter space. Each sample is a possible realization of the parameter set . When a sufficient number of samples are generated, the collection of samples constitutes an empirical approximation of the posterior probability distribution.

[0049] In summary, step S3 performs a complete Bayesian inference by integrating prior setting, dynamic model simulation, likelihood calculation and numerical solution.

[0050] S4, Ecosystem assessment step: probabilistic assessment of the food web of the ecosystem based on the posterior probability distribution of the response function coefficients.

[0051] This step is the final value embodiment of the method of the present application, and its core is to use the posterior probability distribution of the parameters obtained in the aforementioned step S3, which is calibrated by real data, to perform a completely new and probabilistic comprehensive assessment of the structure, function and dynamic response of the food web of the ecosystem, thereby providing more comprehensive and reliable decision support for ecosystem management.

[0052] The implementation of this step marks a fundamental transcendence of the traditional ecosystem assessment paradigm. Instead of generating a single, deterministic assessment result, a dynamic assessment system containing quantitative information about uncertainty is generated based on the entire posterior probability distribution output by step S3. This posterior probability distribution is essentially a collection of a large number of parameter sets sampled by Markov Chain Monte Carlo algorithms, etc., where each parameter set represents a possible true instance of the ecosystem model that is consistent with the observed data.

[0053] The implementation of this ecosystem assessment step can be further divided into the following aspects: First, direct analysis and interpretation of the posterior distribution of parameters. This process is a direct application of the output of step S3, and provides a complete probabilistic description for each response function coefficient to be solved. This includes but is not limited to obtaining the expected value, median, mode of these coefficients, and the crucial Bayesian confidence interval that characterizes the uncertainty range thereof. This analysis can clearly reveal the accuracy of the understanding of each core mechanism of the model under data constraints, providing a basis for judging the reliability of subsequent assessment.

[0054] Secondly, and most importantly, this step is the dynamic and probabilistic assessment of ecosystem properties. This step makes full use of the dynamic parameter response functions established in step S1 and the coefficient distribution calibrated in step S3. For any given environmental scenario, which can be defined by a specific set of environmental factor values .

[0055] For each parameter sample set in the step S3 posterior distribution, a complete set of key process parameter values corresponding to the given environmental scenario can be calculated through the dynamic parameter response function. The parameter sample set is thus transformed into a complete ecosystem model instance under this specific environmental scenario. By repeating this process for a large number of samples in the posterior distribution, a large ensemble of ecosystem models representing all possible states under this environmental scenario can be generated.

[0056] Based on this model ensemble, any ecosystem macroscopic property of interest can be calculated. Since the input is probabilistic, the output is naturally a probability distribution. The posterior distribution of the total system throughput can be calculated, representing the scale and intensity of the metabolic activity of the system as a whole; the posterior distribution of the Finn cycling index can be calculated, quantifying the efficiency and maturity of the internal material recycling within the system; the posterior distribution of a series of key ecological indicators such as the average trophic level of the food web, the total biomass of the system, etc. can also be calculated.

[0057] Finally, the assessment results are no longer isolated numerical values, but are presented in the form of, for example, under the 2°C warming scenario, there is a 95% probability that the total system throughput of this ecosystem is located between X and Y, greatly enhancing the scientific rigor and decision-making reference value of the assessment results.

[0058] This step also includes the diagnosis and exploration of ecosystem mechanisms. The realization of this capability is due to the unique design of the invention to functionalize parameters. By analyzing the dynamic parameter response function itself obtained after calibration, the internal operating mechanism of the ecosystem can be directly understood. By examining the posterior distribution of the response function coefficient related to water temperature in the consumption rate response function of a certain fish functional group, the sensitivity of the species to temperature changes can be directly quantified. If the posterior distribution of this coefficient is significantly greater than zero, it provides strong data support for the hypothesis that the feeding activity of this species will increase with temperature. This ability to learn ecological laws directly from data is not possessed by traditional methods.

[0059] In summary, step S4, through the deep utilization of parameter posterior probability distribution, upgrades the ecosystem assessment from a static and deterministic process to a dynamic and probabilistic analysis framework. Not only can the uncertainty of the assessment results be quantified, but the response of the ecosystem to environmental changes can also be simulated and predicted, and the underlying driving mechanisms can be revealed, thus providing unprecedented technical means for the realization of scientific and forward-looking ecosystem management.

[0060] Please refer to the attached Figure 2 The large water surface ecosystem food web assessment system based on the Ecopath model comprises: A model construction module is configured to construct a food web topology of the ecosystem and establish a core balance equation including a dynamic parameter response function, the dynamic parameter response function including response function coefficients to be solved; A data acquisition module is configured to acquire environmental factor dynamic monitoring data and isotope dynamic observation data obtained through pulse stable isotope labeling. A parameter inversion calibration module is configured to perform Bayesian inversion, the isotope dynamic observation data being a constraint and the environmental factor dynamic monitoring data being a driving force, to solve the response function coefficients of the dynamic parameter response function and generate a posterior probability distribution of the response function coefficients. An ecosystem evaluation module is configured to perform probabilistic evaluation on the food web of the ecosystem based on the posterior probability distribution of the response function coefficients.

[0061] The system of the embodiment can be used to execute the method embodiments described above, and has similar principles and technical effects, which will not be described here again.

Claims

1. A method for assessing a food web of a large water body ecosystem based on an Ecopath model, characterized by, The method comprises the following steps: S1, model construction and parameter functionization step: constructing a food web topology of the ecosystem, and establishing a core balance equation based on an Ecopath model; defining at least one key process parameter in the core balance equation as a dynamic parameter response function related to an environmental factor, the dynamic parameter response function comprising a response function coefficient to be solved; S2, data acquisition step: monitoring the environmental factor corresponding to the dynamic parameter response function in the ecosystem to obtain environmental factor dynamic monitoring data; simultaneously, pulse stable isotope labeling is performed on the primary producers of the ecosystem, and dynamic sampling and analysis are performed on a plurality of functional groups in the food web to obtain isotope dynamic observation data; S3, parameter inversion calibration step: using a Bayesian inversion method, taking the isotope dynamic observation data as observation evidence and the environmental factor dynamic monitoring data as model driving, solving the response function coefficient of the dynamic parameter response function to obtain a posterior probability distribution of the response function coefficient; S4, ecosystem evaluation step: based on the posterior probability distribution of the response function coefficient, probabilistically evaluating the food web of the ecosystem.

2. The Ecopath model-based assessment method of a pelagial food web of a large water body according to claim 1, characterized in that, The key process parameters in the step S1 include unit biomass consumption rate and unit biomass production rate of the functional groups; and the dynamic parameter response function is used to represent the response mechanism of the key process parameters to the dynamic change of the environmental factor.

3. The Ecopath-based assessment method of a large water body food web according to claim 1, wherein, The parameter inversion calibration step in the step S3 specifically comprises: setting a prior probability distribution for the response function coefficient to be solved; establishing a dynamic isotope flow model, which can generate a theoretical isotope dynamic curve according to a given set of response function coefficients and the environmental factor dynamic monitoring data; constructing a likelihood function for quantifying the consistency degree between the theoretical isotope dynamic curve and the isotope dynamic observation data; combining the prior probability distribution and the likelihood function, and solving the posterior probability distribution of the response function coefficient through a Markov Chain Monte Carlo algorithm.

4. The Ecopath-based assessment method of a pelagial food web according to claim 3, wherein, The dynamic isotope flow model is used to describe the transmission process of isotopes in the food web, and the unit mass growth rate and tissue metabolic update rate in the model are driven by the dynamic parameter response function combined with the environmental factor dynamic monitoring data.

5. The Ecopath model-based assessment method of a pelagial food web according to claim 1, wherein, The pulse stable isotope labeling in the step S2 is to add a stable isotope tracer to the ecosystem at the starting time of the experiment; and the dynamic sampling is to collect samples of the functional groups at a plurality of different time points after labeling.

6. The Ecopath-based assessment method of a large water body food web according to claim 1, wherein, The environmental factor is at least one of water temperature, dissolved oxygen or light intensity.

7. The Ecopath-based assessment method of a large water body food web according to claim 1, wherein, The posterior probability distribution of the response function coefficient obtained in the step S3 is a probability distribution form comprising an expected value and a confidence interval of the response function coefficient.

8. The Ecopath-based assessment method of a large water body food web according to claim 1, wherein, The ecosystem evaluation step in the step S2 specifically comprises: According to a given environmental scenario, a posterior probability distribution of the response function coefficients is used to calculate and generate a probability distribution of the ecosystem macro-attributes, including total system throughput, recycling index or average trophic level.

9. The Ecopath-based assessment method of a large water body food web according to claim 1, wherein, The core balance equation in the step S1 is expressed as: the yield of a functional group equals the sum of the total amount of being preyed on by other functional groups in the model, the total amount of fishing, the amount of other natural death, the net amount of emigration and the amount of biomass accumulation.

10. A system for assessing a food web of a large water ecosystem based on an Ecopath model, according to the method for assessing a food web of a large water ecosystem based on an Ecopath model according to any one of claims 1 to 9, characterized in that, The method comprises the following steps: a model construction module, configured to construct a food web topology of an ecosystem and establish a core balance equation comprising a dynamic parameter response function containing response function coefficients to be solved; a data acquisition module, configured to acquire environmental factor dynamic monitoring data and isotope dynamic observation data obtained through pulse stable isotope labeling; a parameter inversion calibration module, configured to perform Bayesian inversion, with the isotope dynamic observation data as a constraint and the environmental factor dynamic monitoring data as a driver, to solve the response function coefficients of the dynamic parameter response function and generate a posterior probability distribution of the response function coefficients; an ecosystem evaluation module, configured to probabilistically evaluate the food web of the ecosystem based on the posterior probability distribution of the response function coefficients.

Images