River fish habitat connectivity assessment method and system

Abstract

The invention belongs to but is not limited to the technical field of ecological environmental protection, and discloses a river fish habitat connectivity assessment method, which comprises the following steps: collecting historical site data of target river fish and bioclimatic data influencing river fish distribution to establish a variable set; the method comprises the following steps: screening variables without collinearity by utilizing VIF, constructing a species distribution model of fishes in a target river by utilizing a machine learning algorithm, and screening environmental factors with relatively high contribution degree of the model; according to a machine learning sub-model selected by AUC and TSS, establishing a set model and simulating HS of target river fish habitat suitability distribution; dam information of the target evaluation drainage basin is collected, and a tree connectivity index (DCI) is calculated; stacking the HS and the DCI, calculating a habitat connectivity index (FRCI) of the fish in the target river, and evaluating the habitat connectivity. According to the method, the habitat suitability, the structure connectivity and the function connectivity are considered, the credibility and the preciseness of evaluation of the habitat connectivity are improved, and a scientific basis and a practical tool are provided for protection and management of a river ecosystem.

Description

Technical Field

[0001] The present invention belongs to, but is not limited to, the field of ecological environmental protection technologies, and particularly relates to a method for evaluating the connectivity of river fish habitats. Background Art

[0002] Under the background of climate change and human activities, the loss of river biodiversity is one of the most concerned issues in ecological protection at home and abroad. Among them, river fish have irreplaceable ecological and economic values, and the protection and restoration of their habitats have become the focus and key point of domestic ecological protection work. Therefore, effective and comprehensive evaluation of the connectivity of river fish habitats is of great significance for implementing ecological protection policies. The existing methods for evaluating the connectivity of river fish habitats mainly focus on habitat suitability or habitat structural connectivity, and cannot consider both at the same time, and often ignore the important biological factor of functional connectivity, which leads to poor effects of river fish habitat protection and restoration measures. Based on the species distribution model, the habitat suitability distribution of the target river fish is simulated and superimposed with the tree-shaped connectivity index that simultaneously considers structural connectivity and functional connectivity to achieve the evaluation of the connectivity of the target river fish habitat, which can provide an effective reference for the formulation of habitat protection and restoration measures for the target river fish. The existing methods for evaluating the connectivity of river fish habitats often cannot consider habitat suitability, structural connectivity and functional connectivity at the same time, and it is difficult to effectively evaluate the connectivity of river fish habitats. Therefore, this method for evaluating the connectivity of river fish habitats is of great significance.

[0003] Existing technical problems solved by the technical solution in industrial applications:

[0004] 1. Existing problem 1: Lack of a systematic habitat connectivity evaluation method

[0005] In the prior art, for the evaluation of the connectivity of river fish habitats, most use simple statistics or empirical judgments, lacking scientific model-based and quantitative evaluation methods.

[0006] The connectivity evaluation fails to comprehensively consider the distribution suitability of river fish and the water system connectivity, and it is difficult to reflect the real impact of dam construction on the habitat.

[0007] 2. Existing problem 2: Unable to accurately integrate multi-dimensional environmental data

[0008] Traditional methods have limited processing capabilities for multi-dimensional environmental factors (such as bioclimatic data, human activity data, and geographic information data) that affect fish habitats, and it is difficult to eliminate the collinearity between factors, resulting in inaccurate model predictions.

[0009] It is impossible to effectively screen out the key factors that have the greatest impact on the distribution of target fish, lacking a data-driven scientific basis.

[0010] 3. Existing Problem 3: Insufficient model performance and lack of dynamic evaluation

[0011] Most methods fail to use machine learning models for high-precision dynamic evaluation and ignore the differences in the periods before and after dam construction.

[0012] Existing evaluation methods cannot combine habitat suitability distribution and river connectivity, making it difficult to comprehensively reflect the dynamic changes of habitats and the suitability connectivity status.

[0013] 4. Existing Problem 4: Difficulty in directly applying evaluation results

[0014] Lack of unified evaluation indicators makes it difficult for evaluation results to provide scientific guidance for actual river ecological restoration, dam optimization design, and fish resource protection.

[0015] It cannot provide specific quantitative data support for the decision-making of related industries, affecting the application value. Summary of the Invention

[0016] Aiming at the problems existing in the prior art, the present invention provides a method for evaluating the connectivity of river fish habitats.

[0017] The present invention is implemented as follows. A method for evaluating the connectivity of river fish habitats includes the following steps:

[0018] Step 1: Collect the historical distribution sites of the target river fish and collect bioclimatic data affecting fish distribution;

[0019] Step 2: Use the variance inflation factor (VIF) to screen non-collinear variables, preliminarily select a machine learning algorithm, construct a species distribution model of the target river fish, and screen environmental factors with relatively high model contribution degrees;

[0020] Step 3: According to two evaluation indicators, the area under the receiver operating characteristic curve (AUC) and the true skill statistic (TSS), select a machine learning sub-model with excellent performance to realize the establishment of an ensemble model and the simulation of habitat suitability distribution (HS);

[0021] Step 4: Collect information on the completion time, dam height, installed capacity, reservoir capacity, and geographical location of the dams in the evaluation target area, and calculate the dendritic connectivity index (DCI) according to the river network water system segmentation;

[0022] Step 5: Overlay the habitat suitability distribution (HS) of the target river fish and the dendritic connectivity index (DCI) to evaluate the connectivity of the river fish habitat.

[0023] Furthermore, the specific steps of Step 2 are as follows:

[0024] Step 2.1: Initially select no less than 4 sub-models from the set M of available machine learning models as the candidate model set M pre ;

[0025] Step 2.2: Calculate the variance inflation factor value (VIF) of each factor in the candidate climate factor set Env, and eliminate the environmental factors with a VIF order of magnitude higher than other factors; the calculation formula of the variance inflation factor value (VIF) is as follows:

[0026]

[0027] In the formula, j---the jth environmental factor;

[0028] ---the sum of squares of the relative change amounts of this environmental factor and all other environmental factors;

[0029] Step 2.3: Use the remaining environmental factors for model simulation, and according to the contribution result of the set model set M pre Eliminate the variables with relatively small contributions, and only retain 2 climate factors related to temperature and 2 temperature factors related to precipitation as the modeling factor data set Env S .

[0030] Furthermore, the specific steps of Step 3 are as follows:

[0031] Step 3.1: Each model in the candidate model set M pre Runs ten times;

[0032] Step 3.2: Exclude the grids where the distribution sites exist, and randomly generate missing points 10 times the number of distribution sites;

[0033] Step 3.3: Randomly select 70% of the total distribution site and missing point data as the training set to fit the algorithm, and the remaining 30% is reserved for evaluating the algorithm performance;

[0034] Step 3.4: Use the area under the receiver operating characteristic curve (AUC) and the true skill statistic (TSS) to evaluate on the evaluation data set;

[0035] Step 3.5: When the sub-model set M s Is determined, simulate the habitat of the target river fish to obtain its habitat suitability distribution (HS);

[0036] Furthermore, the evaluation method in Step 3.4 includes: averaging the AUC and TSS of all sub-models to obtain the AUC pre And TSS pre , if AUC pre > 0.9, TSS preIf it is > 0.75, it is considered that the performance of the set model is excellent, and the selected sub-model set is the final sub-model set M s ; Otherwise, reselect the sub-model set M according to the AUC and TSS of the individual models pre .

[0037] The calculation formula of TSS is as follows:

[0038] TSS = Specificity + Sensitivity - 1 (2)

[0039] In the formula, Specificity is the model specificity, and the calculation formula is:

[0040] Specificity = True negative cases / (True negative cases + False positive cases) (3)

[0041] Sensitivity is the model sensitivity, and the calculation formula is:

[0042] Sensitivity = True positive cases / (True positive cases + False negative cases) (4)

[0043] Furthermore, the specific steps of step 4 are as follows:

[0044] Step 4.1: Collect the basic data for making the dendritic connectivity index (DCI), including the completion time, dam height, installed capacity, reservoir capacity, and geographical location information of the dam;

[0045] Step 4.2: Recheck the geographical location of the dam to ensure that the dam is located on the river line of the evaluation target basin;

[0046] Step 4.3: Segment the river network system of the evaluation target basin according to the geographical location of the dam;

[0047] Step 4.4: On the basis that the passing rates between the dams under construction are independent of each other, the calculation formula of the dendritic connectivity index (DCI) of a river section with a total length of L is as follows:

[0048]

[0049] l' = l × w 1 × w 2 (7)

[0050] Among them, i, j are the river section numbers; when there are M dams between the sub-river sections i and j, and are the upstream and downstream passing rates of the mth dam respectively; L is the total length of the river section, l i , l j is the length of the sub-river sections i, j; l' is the weighted length of the river section, w 1is the weight assigned according to river classification, w 2 is the weight assigned according to the distance between the river section and the dam.

[0051] Furthermore, the specific steps of Step 5 are as follows:

[0052] Step 5: Overlay the habitat suitability distribution (HS) of the target river fish and the dendritic connectivity index (DCI) to obtain the habitat connectivity evaluation index (FRCI) of the target river fish. The calculation formula is as follows:

[0053] FRCI = DCI × HS (8)

[0054] Another object of the present invention is to provide a river fish habitat connectivity evaluation system for a river fish habitat connectivity evaluation method, including:

[0055] The variable set module is established for collecting historical distribution site data of the target river fish and collecting bioclimatic (Bioclimate) data affecting fish distribution;

[0056] The environmental factor screening module is used to screen non-collinear variables using the variance inflation factor (VIF), initially select a machine learning algorithm, construct a species distribution model of the target river fish, and screen environmental factors with relatively high model contribution degrees;

[0057] The algorithm set and simulation module is used to select a machine learning sub-model with superior performance according to two evaluation indexes, the area under the receiver operating characteristic curve (AUC) and the true skill statistic (TSS), to realize the establishment of the ensemble model and the simulation of the habitat suitability distribution (HS);

[0058] The connectivity index calculation module is used to calculate the dendritic connectivity index (DCI) by river section according to the dam information of the evaluation target basin collected;

[0059] The habitat connectivity evaluation module is used to overlay the habitat suitability distribution (HS) of the target river fish and the dendritic connectivity index (DCI) to evaluate the habitat connectivity of the river fish.

[0060] The dam impact evaluation module is used to calculate and evaluate the impact of dam construction on the habitat of the target river fish based on the output results of the habitat model of the target river fish after dam construction and the habitat model before dam construction.

[0061] Another object of the present invention is to provide a computer device. The computer device includes a memory and a processor. When the computer program stored in the memory is executed by the processor, the processor executes the steps of the river fish habitat connectivity evaluation.

[0062] Another object of the present invention is to provide a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to execute the steps of the river fish habitat connectivity assessment described above.

[0063] Another object of the present invention is to provide an information data processing terminal, which includes the river fish habitat connectivity assessment system described above.

[0064] Combined with the above technical solutions and solved technical problems, the advantages and positive effects of the technical solutions to be protected by the present invention are as follows:

[0065] First, the river fish habitat connectivity assessment method of the present invention takes into account habitat suitability, structural connectivity, and functional connectivity, comprehensively and effectively evaluates the habitat connectivity of river fish, and is superior to traditional methods that only consider one or two of these aspects.

[0066] The present invention provides a river fish habitat connectivity assessment method, which solves the problem that the prior art is not comprehensive enough in evaluating the habitat connectivity of river fish. Through a systematic process, from data collection, variable screening, model construction, connectivity calculation to habitat connectivity assessment, a comprehensive assessment and quantification of the river fish habitat connectivity are achieved. This method improves the rigor and credibility of the river fish habitat connectivity assessment and provides a scientific basis for formulating protection and management strategies.

[0067] First of all, the present invention collects data on the distribution sites of target river fish and bioclimate data that affect the distribution of target river fish. The diversity and integrity of this data provide a solid foundation for the accuracy of subsequent models. By using the variance inflation factor (VIF) to screen non-collinear variables and the relative contribution of the model to screen environmental factors with higher importance, the reliability and scientific nature of the model are further improved.

[0068] Secondly, the present invention uses two evaluation indicators, the area under the receiver operating characteristic curve (AUC) and the true skill statistic (TSS), to select machine learning submodels with superior performance, and realizes the establishment of an ensemble model and the simulation of the distribution of fish habitat suitability. This process not only ensures the efficiency and accuracy of the model, but also ensures the stability and reliability of the model results through multiple runs and evaluations.

[0069] Finally, the present invention calculates the dendritic connectivity index (DCI) for each river section by collecting dam information in the target evaluation basin, and superimposes it with the habitat suitability (HS) of the fish in the target river, comprehensively evaluating the habitat connectivity of the fish in the target river. This evaluation method quantifies the habitat connectivity by calculating indicators, thus providing key data support for the formulation of protection and management measures. Compared with traditional methods, the present invention has achieved significant technological progress in data processing, model construction, and habitat connectivity evaluation.

[0070] The present invention has successfully solved the problem that the prior art is not comprehensive enough in evaluating the habitat connectivity of river fish through a systematic data processing, model construction, and habitat connectivity evaluation method. Its significant technological progress is reflected in considering habitat suitability, structural connectivity, and functional connectivity, improving the credibility and rigor of habitat connectivity evaluation, and providing a scientific basis and practical tool for the protection and management of river ecosystems. The popularization and application of this method will have a profound impact on the cause of river ecological protection.

[0071] Second, the present invention has successfully solved multiple technical problems in the evaluation of river fish habitat connectivity in the prior art through a series of main parameters, algorithms, and mathematical models, achieving significant technological progress. First, the present invention uses the distribution site data of the fish in the target river and bioclimate data to establish a comprehensive variable set. The collinearity of the variables is tested by the variance inflation factor (VIF) to ensure the stability and scientificity of the model. This process overcomes the problem of inaccurate models caused by variable collinearity in traditional methods, significantly improving the accuracy and efficiency of variable selection.

[0072] Secondly, the present invention introduces the species distribution model (SDMs) and combines machine learning algorithms to construct a river fish habitat distribution model. Two evaluation indicators, the area under the receiver operating characteristic curve (AUC) and the true skill statistic (TSS), are used to screen sub-models with superior performance, realizing the establishment of an ensemble model and the simulation of the habitat suitability distribution (HS) of the fish in the target river. This method overcomes the problem of insufficient prediction ability of a single model in the prior art, significantly improving the prediction accuracy and stability of the model.

[0073] Finally, the present invention proposes a method of index superposition. By superimposing the dendritic connectivity index (DCI) and the habitat suitability distribution (HS), the habitat connectivity of the fish in the target river is quantitatively evaluated. This method effectively solves the problem that the prior art does not consider enough factors in evaluating the habitat connectivity of river fish, significantly improving the rigor and credibility of the evaluation results.

[0074] Third, the significant technological progress of the technical solution of the present invention:

[0075] 1. Provide a comprehensive quantitative assessment method for habitat connectivity

[0076] The present invention evaluates the connectivity of river fish habitats by superimposing the habitat suitability distribution (HS) and the dendritic connectivity index (DCI), and for the first time realizes the comprehensive evaluation of the suitable distribution of target fish and the connectivity of river networks.

[0077] The assessment method covers the spatio-temporal dynamic impacts of dam construction on habitats and overcomes the limitations of traditional methods.

[0078] 2. Introduce an efficient machine learning model and factor screening mechanism

[0079] The performance of the model is optimized by using the maximum entropy algorithm (MaxEnt) and the ensemble of multiple sub-models, which significantly improves the prediction accuracy of the habitat suitability distribution.

[0080] The variance inflation factor (VIF) method is introduced to eliminate collinear variables, and key modeling factors are comprehensively screened through AUC and TSS to ensure the scientific nature of the model and the credibility of the results.

[0081] 3. Integrate multi-dimensional data to construct a dynamic assessment system

[0082] In the method, bioclimatic data (such as temperature, precipitation), human activity-related data (such as land use), and geographical information data (such as river network structure) are integrated to realize the dynamic modeling of multi-dimensional data.

[0083] Through the dynamic comparison before and after dam construction, the suitability change of fish habitats is quantitatively evaluated, providing an accurate assessment of the impact of dams on the ecosystem.

[0084] 4. Establish a standardized connectivity evaluation index

[0085] The present invention innovatively constructs a habitat connectivity evaluation index (FRCI), which quantitatively combines habitat suitability and river connectivity, providing a unified standard for ecological assessment.

[0086] Through the quantitative results of the evaluation index, the dynamic changes of fish habitat connectivity can be intuitively displayed, supporting scientific decision-making for relevant protection and management measures.

[0087] 5. Significantly improve the practical industrial application value

[0088] The evaluation results can directly guide the design of water conservancy projects (such as dam optimization, fish passage setting) and ecological restoration measures (such as river channel connection plan optimization).

[0089] The method is applicable to multiple industrial fields such as dam construction, river ecological protection, and fish resource management, providing balanced technical support for river development and protection.

[0090] Technical advantages and significant application value:

[0091] 1. High precision and wide applicability

[0092] By introducing a machine learning model and establishing a dynamic evaluation system, the evaluation accuracy has been significantly improved, and it can be applied to various river systems and fish habitat connectivity evaluation scenarios.

[0093] The method has strong universality and is applicable to multiple fields such as dam construction, ecological protection, and river planning.

[0094] 2. Promote scientific decision-making

[0095] The evaluation method provides a quantitative scientific basis for decision-makers, supporting dam design and river management decisions based on ecological protection needs.

[0096] It provides clear ecological protection indicators and technical paths, supporting cross-sectoral collaboration and industrial optimization.

[0097] 3. Balance development and protection

[0098] The present invention provides technical support for the coordination of ecological protection and economic development. While ensuring the economic benefits of the dam, it minimizes the damage to fish habitats and contributes to the sustainable development of the basin.

[0099] The method for evaluating the connectivity of river fish habitats proposed by the present invention significantly improves the accuracy, scientificity, and practicality of the evaluation in industrial applications, overcoming the deficiencies of traditional evaluation methods. Its innovation and technological progress will have a wide impact in the fields of water conservancy projects, ecological protection, and basin management. Brief description of the drawings

[0100] Figure 1 is the flowchart of the method for evaluating the connectivity of river fish habitats provided by the embodiments of the present invention.

[0101] Figure 2 is the distribution map of the river network and the sites of Schizothorax dolichonema in the Jinsha River Basin provided by the embodiments of the present invention.

[0102] Figure 3 is the habitat suitability distribution map of Schizothorax dolichonema in the periods of 1970 - 2000 and 2001 - 2020 provided by the embodiments of the present invention.

[0103] Figure 4 is the distribution map of the dendritic connectivity index of the Jinsha River Basin in the periods of 1970 - 2000 and 2001 - 2020 provided by the embodiments of the present invention.

[0104] Figure 5It is the distribution map of the habitat connectivity index of Schizothorax dolichonema in the periods of 1970 - 2000 and 2001 - 2020 provided by the embodiments of the present invention.

[0105] Figure 6 It is the schematic diagram of the river fish habitat connectivity assessment system provided by the embodiments of the present invention. Specific embodiments

[0106] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0107] As Figure 1 shown, a river fish habitat connectivity assessment method provided by the embodiments of the present invention includes the following steps:

[0108] Step 1: Collect the historical distribution sites of target river fish and collect bioclimate data affecting fish distribution;

[0109] Step 2: Use the variance inflation factor (VIF) to screen variables without collinearity, initially select a machine learning algorithm, construct a species distribution model of target river fish, and screen environmental factors with relatively high model contribution degrees;

[0110] Step 3: According to two evaluation indicators, the area under the receiver operating characteristic curve (AUC) and the true skill statistic (TSS), select a machine learning sub-model with excellent performance to realize the establishment of an ensemble model and the simulation of habitat suitability distribution (HS);

[0111] Step 4: Collect information on the completion time, dam height, installed capacity, reservoir capacity and geographical location of the dams in the target evaluation area, and calculate the dendritic connectivity index (DCI) according to the river network water system segmentation;

[0112] Step 5: Overlay the habitat suitability distribution (HS) of the target river fish and the dendritic connectivity index (DCI) to evaluate the habitat connectivity of the river fish.

[0113] Furthermore, the specific steps of Step 2 are as follows:

[0114] Step 2.1: Initially select no less than 4 sub-models from the available machine learning model set M as the candidate model set M pre ;

[0115] Step 2.2: Calculate the variance inflation factor value (VIF) of each factor in the candidate climate factor set Env, and eliminate environmental factors with a VIF order of magnitude higher than other factors; the calculation formula of the variance inflation factor value (VIF) is as follows:

[0116]

[0117] Wherein, j---the jth environmental factor;

[0118] ---the sum of squares of the relative change amounts of this environmental factor and all other environmental factors;

[0119] Step 2.3: Use the remaining environmental factors for model simulation. According to the contribution degree results of the ensemble model set M pre Eliminate the variables with relatively small relative contribution degrees, and only retain 2 climate factors related to temperature and 2 temperature factors related to precipitation as the modeling factor dataset Env S .

[0120] Furthermore, the specific steps of Step 3 are as follows:

[0121] Step 3.1: For the candidate model set M pre Each model runs ten times;

[0122] Step 3.2: Exclude the grids where the distribution sites exist, and randomly generate missing points that are 10 times the number of distribution sites;

[0123] Step 3.3: Randomly select 70% of the total distribution site and missing point data as the training set to fit the algorithm, and the remaining 30% is reserved for evaluating the algorithm performance;

[0124] Step 3.4: Use the area under the receiver operating characteristic curve (AUC) and the true skill statistic (TSS) to evaluate on the evaluation dataset;

[0125] Step 3.5: When the sub-model set M s is determined, simulate the habitat of the target river fish to obtain its habitat suitability distribution (HS);

[0126] Furthermore, the evaluation method in Step 3.4 includes: averaging the AUC and TSS of all sub-models to obtain the AUC pre and TSS pre of the ensemble model. If AUC pre > 0.9 and TSS pre > 0.75, it is considered that the performance of the ensemble model is excellent, and the selected sub-model set is the final sub-model set M s ; otherwise, reselect the sub-model set M according to the AUC and TSS of individual models pre .

[0127] The calculation formula of TSS is as follows:

[0128] TSS = Specificity + Sensitivity - 1 (2)

[0129] Wherein, Specificity is the model specificity, and the calculation formula is:

[0130] Specificity = True negative cases / (True negative cases + False positive cases) (3)

[0131] Sensitivity is the model sensitivity, and the calculation formula is:

[0132] Sensitivity = True positive cases / (True positive cases + False negative cases) (4)

[0133] Furthermore, the specific steps of Step 4 are as follows:

[0134] Step 4.1: Collect the basic data for making the dendritic connectivity index (DCI), including the completion time, dam height, installed capacity, reservoir capacity, and geographical location information of the dam;

[0135] Step 4.2: Recheck the geographical location of the dam to ensure that the dam is located on the river line of the target evaluation basin;

[0136] Step 4.3: Segment the river network system of the target evaluation basin according to the geographical location of the dam;

[0137] Step 4.4: On the basis that the passing rates between the dams under construction are independent of each other, the calculation formula for the dendritic connectivity index (DCI) of a river section with a total length of L is as follows:

[0138]

[0139] l' = l × w 1 ×w 2 (7)

[0140] Wherein, i and j are the river section numbers; when there are M dams between the sub-river sections i and j, and are the upstream and downstream passing rates of the mth dam respectively; L is the total length of the river section, l i ,l j is the length of the sub-river sections i and j; l' is the weighted length of the river section, w 1 is the weight assigned according to the river classification, w 2 is the weight assigned according to the distance between the river section and the dam body; and refer to Table 1 for the values, and w 1 and w 2 refer to Table 2 for the values.

[0141] Further, the specific steps of Step 5 are as follows:

[0142] Step 5: Superimpose the habitat suitability distribution (HS) of the target river fish and the dendritic connectivity index (DCI) to obtain the habitat connectivity evaluation index (FRCI) of the target river fish. The calculation formula is as follows:

[0143] FRCI = DCI × HS (8)

[0144] Table 1 Dam height and corresponding passing rate (P)

[0145]

[0146] Table 2 Weight coefficients of different river reaches, including the weight coefficient (w 1 ) based on river classification and the weight coefficient (w 2 ) of the distance between the river reach and the dam

[0147]

[0148] As Figure 6 shown, a river fish habitat connectivity evaluation system for a river fish habitat connectivity evaluation method provided by an embodiment of the present invention includes:

[0149] A variable set module is established for collecting historical distribution site data of target river fish and collecting bioclimate data affecting fish distribution;

[0150] An environmental factor screening module is used to screen non - collinear variables using the variance inflation factor (VIF), preliminarily select a machine learning algorithm, construct a species distribution model of target river fish, and screen environmental factors with relatively high model contribution degrees;

[0151] An algorithm set and simulation module is used to select a machine learning sub - model with excellent performance according to two evaluation indexes, the area under the receiver operating characteristic curve (AUC) and the true skill statistic (TSS), and realize the establishment of an ensemble model and the simulation of habitat suitability distribution (HS);

[0152] A connectivity index calculation module is used to calculate the dendritic connectivity index (DCI) by river reach according to the dam information of the evaluated target basin;

[0153] A habitat connectivity evaluation module is used to superimpose the habitat suitability distribution (HS) of the target river fish and the dendritic connectivity index (DCI) to evaluate the habitat connectivity of the river fish.

[0154] The Jinsha River Basin (24°36′ - 35°44′N, 90°30′ - 105°15′E;Figure 2 ) Located in the upper reaches of the Yangtze River Basin, with a basin area of 473,200 km² 2 , accounting for about 26% of the entire Yangtze River Basin. The climate types in this basin are complex and diverse, ranging from the alpine subfrigid zone to the subtropical humid monsoon climate. The water energy resources in the Jinsha River Basin account for more than 40% of the Yangtze River Basin. The Jinsha River and its tributary, the Yalong River, are two important hydropower bases planned in China, and their hydropower development has extremely important strategic significance. The Jinsha River Basin is located on the edge of the Qinghai-Tibet Plateau region in the geographical division, with a complex fish composition, extremely rich fish resources, and is famous for a large number of fish endemic species. The large-scale hydropower construction in the Jinsha River Basin has significantly affected the aquatic ecological environment. The dam obstruction has changed the river environment with rapids and dangerous shoals into a reservoir environment with still water and slow flow, changing the connectivity of the fish habitats in the river. The decline in connectivity caused by dam construction is the most serious human activity pressure faced by fish habitats, and climate change will exacerbate the pressure faced by fish habitats. Therefore, it is necessary to consider the habitat suitability driven by climate change, the structural connectivity affected by dam construction, and the functional connectivity determined by the river fish itself to evaluate the habitat connectivity of endemic river fish in the Jinsha River Basin, so as to implement habitat protection and restoration. Schizothorax dolichonema is an endemic fish of the Cyprinidae family in the Jinsha River Basin. Therefore, a method for evaluating the habitat connectivity of river fish in the present invention takes Schizothorax dolichonema as an example, establishes a species distribution model (SDMs), simulates the distribution of habitat suitability (HS) of Schizothorax dolichonema, calculates the dendritic connectivity index (DCI) of the Jinsha River Basin, and superimposes HS and DCI to evaluate the habitat connectivity of Schizothorax dolichonema in multiple periods, and formulates habitat restoration and protection strategies. The specific process is as follows:

[0155] Collect the distribution site data of Schizothorax dolichonema in the period from 1970 to 2000, check the correctness of the spatial positions of these data to ensure that these points fall on the river line ( Figure 2 ). Exclude the grids where the distribution site data are located, and use a randomization method to generate 10 times the number of absence point data of the presence points in the remaining grids, and merge them into a total data set. Input the merged data set, the modeling factor set Env and the candidate model set M preDebug the species distribution models (SDMs), select isothermality, mean temperature of the driest quarter, precipitation of the driest month, and precipitation of the coldest quarter as the model building factors according to the contribution results, and select the generalized linear model (GLM), surface range envelope model (SRE), flexible discriminant analysis (FDA), and maximum entropy model (MAXENT) as sub-models to form an ensemble model according to the values of the area under the receiver operating characteristic curve (ROC) and true skill statistic (TSS). Simulate the habitat suitability distribution of Schizothorax dolichonema during the period of 1970 - 2000, and predict the habitat suitability distribution during the period of 2001 - 2020 ( Figure 3 ). According to the dam information collected in the Jinsha River Basin, draw the dendritic connectivity index distribution during the periods of 1970 - 2000 and 2001 - 2020 ( Figure 4 ). Overlay the suitability distribution of Schizothorax dolichonema and the dendritic connectivity index distribution to evaluate the habitat connectivity during the periods of 1970 - 2000 and 2001 - 2020 ( Figure 5 ).

[0156] The building factors of the species distribution model of Schizothorax dolichonema all passed the collinearity test, and the AUC and TSS values of the model were 0.97 and 0.85 respectively, indicating excellent model performance. Figure 3 It shows that new habitats of Schizothorax dolichonema have emerged at the source of the Jinsha River, and some habitats have been lost in the lower reaches of the Jinsha River. Figure 4 It shows that the connectivity of the Jinsha River during the period of 2001 - 2020 has decreased throughout the basin compared with that during the period of 1970 - 2000, especially in the middle and lower reaches and the tributaries of the Yalong River. Figure 5 It shows that the habitat connectivity of Schizothorax dolichonema in the source area of the Jinsha River, the tributaries of the Yalong River, and the lower reaches of the Jinsha River has deteriorated significantly during the period of 2001 - 2020.

[0157] Example 1: Ecological impact assessment of water conservancy projects

[0158] When building a new dam in a river basin, it is usually necessary to evaluate its impact on the ecosystem, especially on fish habitats. Through the evaluation method of the present invention, the changes in the suitability and connectivity of the target fish habitats can be quantitatively analyzed, providing a scientific basis for the design and optimization of water conservancy projects.

[0159] 1. Project requirements: A certain river basin plans to build a hydropower dam, and it is necessary to evaluate its potential impact on the habitat connectivity and distribution of specific fish (such as Chinese sturgeon and Chinese paddlefish) in the basin.

[0160] 2. Implementation steps:

[0161] Data collection: Collect the distribution data of target fish before dam construction, bioclimatic data (temperature, precipitation), as well as the geographical location, dam height, and passing rate of the dam.

[0162] Habitat simulation: Use the machine learning model (such as MaxEnt) in the method of the present invention to construct a habitat suitability model (HS) for fish before and after dam construction.

[0163] Connectivity calculation: Combine the passing rate data of the dam and the river network segmentation to calculate the dendritic connectivity index (DCI) within the basin.

[0164] Comprehensive evaluation: Overlay HS and DCI to generate a fish habitat connectivity evaluation index (FRCI).

[0165] 3. Results and applications: The evaluation results show that the connectivity of some key river sections has decreased significantly. Optimized suggestions for designing fishways or regulating the flow rate on the dam are proposed to reduce the damage to the habitats of target fish.

[0166] Industrial value:

[0167] This method provides a scientific ecological assessment tool for water conservancy project construction, reduces the damage of the project to fish habitats, and promotes the balance between ecological protection and project development.

[0168] Example 2: Basin ecological restoration and protection planning

[0169] For a basin with multiple dams already built, the connectivity of fish habitats is usually severely affected. The present invention can be used to evaluate the impact of existing dams on river fish habitats and guide the planning of basin ecological restoration projects.

[0170] 1. Project requirements: There are multiple dams distributed in a certain river basin, and the population of target fish (such as the four major Chinese carps) has decreased significantly. An ecological restoration plan needs to be formulated through scientific evaluation.

[0171] 2. Implementation steps:

[0172] Data collection: Collect the historical and current distribution data of fish in the basin, and record the construction time, dam height, passing rate, and geographical location of each dam.

[0173] Model evaluation: Use the method of the present invention to simulate the distribution of fish habitat suitability and calculate the habitat suitability model (HS).

[0174] Connectivity analysis: Combine the river network segmentation and the dam passing rate to calculate the dendritic connectivity index (DCI) of the basin, and overlay HS and DCI to generate the habitat connectivity evaluation index (FRCI).

[0175] Optimization plan: Identify the river reaches with the worst connectivity based on the evaluation results, and formulate restoration measures, such as demolishing abandoned dams, adding fishways or improving the passing rate of existing fishways.

[0176] 3. Results and applications: After adding fishways to key river reaches, the habitat connectivity index (FRCI) has increased significantly, providing effective support for the restoration of the target fish population in the basin.

[0177] This method provides a quantitative basis for ecological restoration in the basin, ensures the scientificity and accuracy of restoration measures, improves the connectivity and suitability of river fish habitats, and promotes the restoration and protection of biodiversity.

[0178] The Jinsha River Basin has three characteristics: complex climate characteristics, concentrated hydropower development, and rich endemic fish resources. Among them, the habitats of endemic fish are threatened by multiple pressures. Evaluating the habitats of these fish generally involves establishing species distribution models and changing climate factors to explain habitat changes, ignoring other dynamic factors, such as changes in structural connectivity caused by hydropower development and differences in functional connectivity determined by biological factors. Therefore, it may lead to a decline in the effectiveness of assessing the habitat connectivity of target river fish.

[0179] The present invention innovatively proposes a method for evaluating the habitat connectivity of river fish. In the embodiment, it reveals a significant decline in the habitat connectivity of Schizothorax dolichonema ( Figure 5 ) rather than just migrating upstream ( Figure 3 ), which can improve the efficiency of protecting Schizothorax dolichonema and avoid wasting manpower and material resources.

[0180] The above is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, any modifications, equivalent replacements, and improvements made within the spirit and principle of the present invention shall be covered by the protection scope of the present invention.

Claims

1. A method for assessing connectivity of river fish habitats, characterized in that: The following steps are involved: Step 1: Collect historical distribution sites of target river fish and bioclimatic data that affect fish distribution; Step 2: Use the variance inflation factor to screen variables that do not have collinearity, preliminarily select a machine learning algorithm, build a species distribution model for target river fish, and screen environmental factors with relatively high contribution to the model; Step 3: Based on the two evaluation indicators of the area under the receiver operating characteristic curve and the true skill statistic, select the machine learning sub-model with superior performance to achieve the establishment of the ensemble model and the simulation of habitat suitability distribution; Step 4: Collect information on the construction time, dam height, installed capacity, reservoir capacity and geographical location of the dams in the target area, and calculate the tree connectivity index based on the river network segmentation; Step 5: Superimpose the habitat suitability distribution and tree connectivity index of the target river fish to assess the connectivity of the river fish habitat.

2. The method for assessing river fish habitat connectivity according to claim 1, characterized in that: The specific steps of step 2 are as follows: Step 2.1: Preliminarily select no less than 4 sub-models from the available machine learning model set M as the candidate model set M pre ; Step 2.2: Calculate the variance inflation factor (VIF) of each factor in the candidate climate factor set Env, and remove environmental factors whose VIF magnitude is higher than that of other factors; the calculation formula of the variance inflation factor (VIF) is as follows: In the formula, j---the jth environmental factor; ---The sum of the squares of the relative changes of this environmental factor and all other environmental factors; Step 2.3: Use the remaining environmental factors to simulate the model according to the ensemble model set M pre Contribution results: Variables with relatively small contributions are eliminated, and only two temperature-related climate factors and two precipitation-related temperature factors are retained as the modeling factor dataset Env S .

3. The method for assessing river fish habitat connectivity according to claim 1, characterized in that: The specific steps of step 3 are as follows: Step 3.1: Selected model set M pre Each model was run ten times; Step 3.2: Exclude the grids where the distribution sites exist and randomly generate missing points 10 times the number of distribution sites; Step 3.3: Randomly select 70% of the total distribution sites and missing point data as the training set to fit the algorithm, and the remaining 30% is reserved for evaluating the algorithm performance; Step 3.4: Evaluate on the evaluation dataset using the area under the receiver operating characteristic curve (AUC) and the true skill statistic (TSS); Step 3.5: When the sub-model set M s After determination, the habitat of the target river fish is simulated to obtain its habitat suitability distribution (HS); Furthermore, the evaluation method in step 3.4 includes: averaging the AUC and TSS of all sub-models to obtain the AUC of the ensemble model pre and TSS pre , if AUC pre >0.9,TSS pre >0.75, the ensemble model is considered to have superior performance, and the selected sub-model set is the final sub-model set M s Otherwise, reselect the submodel set M based on the AUC and TSS of the single model pre .

4. The method for assessing river fish habitat connectivity according to claim 1, characterized in that: The specific steps of step 4 are as follows: Step 4.1: Collect basic data for making the dendritic connectivity index (DCI), including the construction time, dam height, installed capacity, reservoir capacity and geographical location of the dam; Step 4.2: Review the geographical location of the dam to ensure that it is located on the river line of the target watershed for assessment; Step 4.3: Segment the river network of the target basin according to the geographical location of the dam; Step 4.4: On the basis of the independent passing rates between dams, the dendritic connectivity index (DCI) of a river section with a total length of L is calculated as follows: l'=l×w1×w2 (7) Where i and j are the river section numbers. When there are M dams between sub-sections i and j, and are the upstream and downstream passing rates of the mth dam respectively; L is the total length of the river section, l i ,l j is the length of sub-section i, j; l' is the weighted length of the section, w1 is the weight assigned according to the river classification, and w2 is the weight assigned according to the distance between the section and the dam.

5. The method for assessing river fish habitat connectivity according to claim 1, characterized in that: The specific steps of step 5 are as follows: Step 5: Superimpose the habitat suitability distribution (HS) and the dendritic connectivity index (DCI) of the target river fish to obtain the habitat connectivity evaluation index (FRCI) of the target river fish. The calculation formula is as follows: FRCI=DCI×HS (8).

6. A river fish habitat connectivity assessment system according to the river fish habitat connectivity assessment method according to any one of claims 1 to 5, characterized in that: include: (1) Data collection module: used to collect historical distribution sites of target river fish and bioclimatic data that affect fish distribution, including factors such as temperature and precipitation; (2) Data processing module: used to screen variables that do not have collinearity based on the variance inflation factor (VIF), select sub-models from the machine learning model set, build a species distribution model of target river fish, and screen environmental factors with relatively high contribution to the model; (3) Model evaluation module: used to evaluate model performance based on the area under the receiver operating characteristic curve (AUC) and true skill statistic (TSS), determine the ensemble model with superior performance, and simulate the habitat suitability distribution (HS) of target river fish; (4) River network connectivity calculation module: used to collect the construction time, dam height, installed capacity, reservoir capacity and geographical location information of the dams in the target basin, and calculate the tree connectivity index (DCI) of the target basin; (5) Connectivity assessment module: used to superimpose the habitat suitability distribution (HS) of target river fish with the dendritic connectivity index (DCI) to generate the habitat connectivity evaluation index (FRCI) of target river fish.

7. The river fish habitat connectivity assessment system according to claim 6, characterized in that: The data processing module further includes: (1) Factor screening submodule: used to calculate the variance inflation factor (VIF) of environmental factors, eliminate variables with strong collinearity, and retain bioclimatic factors that have a greater impact on the distribution of target fish habitats; (2) Modeling factor optimization submodule: Based on the submodel operation results, the climate factors and environmental factors with relatively high contributions are screened to form an optimized modeling factor data set.

8. The river fish habitat connectivity assessment system according to claim 6, characterized in that: The model evaluation module further includes: (1) Model running submodule: used to run each model in the selected model set multiple times, and divide the training set and test set by randomly generating background point and distribution point data; (2) Performance evaluation submodule: used to evaluate model performance based on the area under the receiver operating characteristic curve (AUC) and the true skill statistic (TSS), and to determine the sub-model set with superior performance; (3) Habitat suitability simulation submodule: Generates the habitat suitability distribution (HS) of target river fish based on the preferred submodel set.

9. The river fish habitat connectivity assessment system according to claim 6, characterized in that: The river network connectivity calculation module further includes: (1) Data integration submodule: used to integrate the construction time, dam height, installed capacity, reservoir capacity and geographical location information of dams in the target basin; (2) River network segmentation submodule: segment the river network of the target basin according to the geographical location of the dam; (3) Connectivity index calculation submodule: Based on the river segmentation data and dam pass rate, the tree connectivity index (DCI) of the target basin is calculated.

10. The river fish habitat connectivity assessment system according to claim 6, characterized in that: The connectivity assessment module further includes: (1) Data overlay submodule: used to overlay the habitat suitability distribution (HS) of target river fish with the dendritic connectivity index (DCI); (2) Connectivity assessment submodule: The habitat connectivity assessment index (FRCI) of the target river fish is generated based on the superposition results to quantify the connectivity status of the habitat of the target fish in the assessment basin.

Images