A Disease Early Warning and Ecological Control Method for a Live Stock Bank of Rare Fish
By constructing a database of rare fish species and deploying multi-dimensional monitoring equipment, and combining it with a BP neural network for three-level progressive early warning and targeted prevention and control, the problems of delayed early warning and inaccurate prevention and control in rare fish farming have been solved. This has enabled early warning and precise prevention and control of rare fish diseases, reducing the risk of disease and management costs.
Patent Information
- Application Number
- CN202611130129.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-07-28
- Publication Date
- 2026-11-17
AI Technical Summary
Current disease control practices in rare fish farming suffer from problems such as delayed early warning, limited monitoring dimensions, lack of precision in control measures, and low data utilization, making it difficult to meet the high-precision and low-damage requirements of live rare fish banks.
By constructing a database of the normal survival status of rare fish species, deploying microbial eDNA monitoring, water quality sensors, and individual behavior video monitoring equipment, a three-level progressive early warning system is achieved. Anomaly detection is performed by combining a BP neural network, and targeted prevention and control measures, such as probiotic droplet addition and bacteriophage delivery, are adopted. A unified database is established for data optimization.
It has enabled early warning of diseases in rare fish species, improved the accuracy of early warning and the effectiveness of prevention and control, reduced the morbidity and mortality of rare fish species, protected the micro-ecological balance of aquatic bodies, and reduced management costs.
Smart Images

Figure CN123116121A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of rare aquatic organism protection and aquaculture technology, specifically to a method for disease early warning and ecological control in a live stock bank of rare fish. Background Technology
[0002] As a key component of aquatic ecosystems, the conservation and population recovery of rare fish species have become crucial tasks for global biodiversity conservation. Live fish banks, serving as the core carriers for ex-situ conservation of rare fish, play vital roles in species preservation, artificial breeding, and germplasm resource protection. However, rare fish typically exhibit characteristics such as sensitivity to environmental changes, weak immunity, and poor adaptability. During live fish bank aquaculture, the risk of disease outbreaks is extremely high. Once a disease occurs, it can not only lead to the death of individual rare fish but also potentially cause the extinction of the entire population.
[0003] Existing methods for disease control in the aquaculture of rare fish species have the following drawbacks:
[0004] First, the early warning system is highly delayed. Traditional disease monitoring relies heavily on manual observation or laboratory testing after fish show obvious symptoms, which fails to detect risks in the early stages of disease outbreaks and misses the best opportunity for prevention and control. For example, current technologies for detecting fish pathogens typically require collecting fish tissue samples for culture and identification, a process that takes 24-72 hours and poses a risk of damage to rare fish species, making it difficult to meet the protection needs of live fish banks.
[0005] Secondly, the monitoring dimensions are limited. Existing technologies often focus on monitoring single indicators, such as water quality parameters or specific pathogens, lacking comprehensive analysis of multi-dimensional data including microbial communities, water quality environment, and fish behavior. This results in low accuracy of early warnings and a high risk of misjudgment or missed detection. For example, some patents use water quality sensors to monitor parameters such as water temperature and pH, but fail to consider that changes in the microbial community are an early signal of disease occurrence, thus failing to achieve early warning.
[0006] Third, the prevention and control measures lack precision. Existing methods mostly employ broad-spectrum bactericides or antibiotics, which not only easily disrupt the aquatic microecological balance but may also cause stress reactions in rare fish species, affecting their health. Furthermore, there is a lack of targeted prevention and control measures for different risk types, resulting in poor control effectiveness.
[0007] Finally, the data utilization rate is low. The data from various monitoring devices in the existing aquaculture system are mostly isolated, and a unified database has not been established for data comparison and analysis. This makes it impossible to trace, analyze, and optimize the data, which makes it difficult to continuously improve the prevention and control plan.
[0008] Currently, microbial monitoring methods based on eDNA technology have been initially applied in aquaculture. For example, patent CN202411418028.0 discloses a method for identifying pathogens in shrimp farming water sources, which uses eDNA technology to achieve early detection of pathogens. However, this technology is only applicable to shrimp farming and does not combine water quality parameters with biological behavior data for comprehensive early warning, failing to meet the high-precision and low-damage requirements of live stock banks for rare fish. Patent CN106561532A discloses a method for monitoring fish activity, which uses image recognition technology to monitor fish movement speed. However, this method only focuses on a single behavioral indicator and does not integrate with microbial and water quality data, resulting in limited accuracy in early warning. Summary of the Invention
[0009] To address the problems existing in the prior art, this invention provides a method for disease early warning and ecological control in a live stock of rare fish species. This invention can reduce the lag in early warning and the risk of fish damage, and can also improve the accuracy of early warning and continuously optimize the control plan.
[0010] To achieve the above objectives, the technical solution provided by this invention is a method for disease early warning and ecological control in a live stock bank of rare fish species, comprising the following steps:
[0011] (1) Basic data collection and database construction: Collect basic data on rare fish under normal living conditions over the years. The basic data includes microbial eDNA data, water quality environmental data, and individual fish behavior data; in collecting individual fish behavior data G nmsrk Simultaneously, microbial eDNA data E from the corresponding time and region are collected. nmspk and water quality environmental data F nmsqk n represents the year, m represents the season, s represents the fish species, r represents the type of individual fish behavior data parameters, p represents the type of microbial eDNA data parameters, q represents the type of environmental data parameters, and k represents the number of samples. n, m, s, r, p, and k are all natural numbers. A database of the normal survival status of rare fish species is established, which includes a microbial community benchmark database, a water quality parameter benchmark database, and a fish behavior benchmark characteristic database.
[0012] Among them, the data parameters in the microbial community benchmark library include at least the relative abundance E of the core dominant bacterial community. nms1k Relative abundance of potential pathogenic bacteria (E) nms2k Microbial Shannon-Wiener Diversity Index E nms3k The ratio of the abundance of core dominant bacterial groups to potential pathogenic bacterial genera is E. nms4k Abundance of nitrifying bacteria functional groups E nms5k and the evenness index of the microbial community E nms6k The water quality parameter benchmark database includes at least the oxidation-reduction potential (ORP). nms1kAmmonia nitrogen (NH3-N) concentration F nms2k Nitrite NO2 - -N concentration F nms3k Dissolved oxygen (DO) content nms4k and pH value F nms5k The data parameters in the fish behavior benchmark feature database should at least include swimming speed G per unit time. nms1k Feeding frequency G nms2k and the dispersion of the motion trajectory G nms3k ;
[0013] (2) High-frequency monitoring equipment setup: A live animal culture unit was established, and a real-time monitoring device consisting of a microbial eDNA monitoring unit, a high-precision water quality sensor array, and an individual behavior video monitoring unit was deployed in the culture unit to obtain daily microbial eDNA data E. 0mspk Water quality environmental data F 0msqk and fish individual behavior data G 0msrk The microbial eDNA monitoring unit is the core monitoring unit. It collects water samples daily from each independent aquaculture unit using an automated water sample collection and filtration device, extracts eDNA from the samples, and performs rapid 16S rRNA gene sequencing using a portable nanopore sequencer to analyze the microbial community structure. The high-precision water quality sensor array monitors the temperature, pH, dissolved oxygen content, oxidation-reduction potential (ORP), ammonia nitrogen concentration, and nitrite concentration of the aquaculture water. The individual behavior video monitoring unit monitors the movement trajectory and intensity of individual fish in real time through a combination of image recognition and distance sensing. The temperature error is no greater than ±0.2℃, pH error no greater than ±0.2, dissolved oxygen error no greater than ±0.2 mg / L, ORP error no greater than ±5mV, ammonia nitrogen error no greater than ±5%, and nitrite error no greater than ±5% of full scale.
[0014] (3) Abnormal data early warning: Based on the real-time monitoring data obtained in step (2), the real-time monitoring data obtained in step (2) is processed to obtain the daily average microbial eDNA data E. 0msp Average water quality data F 0msq and mean data of individual fish behaviors G 0msr Then, the normal data from the database in step (1) is extracted and processed to obtain the average value data E of normal microbial eDNA data over the years. nmsp Mean data of normal water quality environment F nmsq The mean data G of normal individual fish behavior data nmsr The processed real-time monitoring data is then compared with the processed normal data. Based on the difference in the compared data, a three-level progressive early warning system is implemented, specifically including:
[0015] 1) Level 1 warning, i.e., microbial imbalance warning: This involves using the daily average microbial eDNA data (E...) 0msp Compared with the average value of normal microbial eDNA data over the years E nmsp Comparison was performed, and if the daily average microbial eDNA data E 0msp Compared with the average value of normal microbial eDNA data over the years E nmsp If the average data for one year exceeds the threshold, a Level 1 risk warning is issued, namely a warning of microbial imbalance.
[0016] The threshold mentioned above specifically refers to the mean relative abundance data E of the core dominant bacterial groups in the real-time microbial community structure. 0ms1 Deviation from the mean data of normal data E nms1 The amplitude exceeded the historical fluctuation baseline and the mean relative abundance data of potential pathogenic bacteria E. 0ms2 E, the average data compared to normal data nm2 An increase of more than 20% or a microbial diversity index E 0m3 E, the average data compared to normal data nm3 The reduction should be no less than 20%; that is, (|E nms1 -E 0ms1 |) / E nms1 >30%, (E 0ms2 -E nms2 ) / E nms2 >20% or (E nms3 -E 0ms3 ) / E nms3 ≥20%;
[0017] 2) Level II Early Warning, i.e., Abnormal Behavior Confirmation Early Warning: After a Level I early warning is triggered, individual fish behavior data will be collected through individual behavior video monitoring units, along with the average data G of normal individual fish behavior data over the years. nmsr By comparison, if an individual fish is found to have a feeding frequency that is at least 30% lower than the mean of normal data or a swimming speed that is at least 30% lower than the mean of normal data, i.e. (G) nms2 -G 0ms2 ) / G nms2 ≥30% or (G nms1 -G 0ms1 ) / G nms1 If the risk level is ≥30%, the system will raise the risk level to Level 2 and preliminarily determine the risk type as microecological imbalance, potential pathogen proliferation, or organic load stress based on the characteristics of microbial community changes.
[0018] The aforementioned microecological imbalance is specifically manifested in the mean relative abundance data E of the core dominant bacterial community. 0m1 E, the average data compared to normal data nm1Decline; the proliferation of potential pathogens is specifically manifested in the mean relative abundance data of potential pathogen genera E. 0m2 E, the average data compared to normal data nm2 An increase; the organic load stress type is specifically manifested in the mean relative abundance data E of the core dominant bacterial community. 0m1 E, the average data compared to normal data nm1 rise;
[0019] 3) Level III Early Warning, i.e., Water Quality Anomaly Evidence Warning: After a Level II early warning is triggered, based on monitoring data from a high-precision water quality sensor array, if the mean oxidation-reduction potential (ORP) data F is detected... 0ms1 The mean data F compared to normal data nms1 Reduce by at least 50mV, average ammonia nitrogen concentration data F 0ms2 The mean data F compared to normal data nms2 An increase of not less than 0.3 mg / L or the average nitrite concentration data F 0ms3 The mean data F compared to normal data nms3 If the increase is not less than 0.1 mg / L, the risk level will be raised to level three, and a comprehensive diagnosis will be completed.
[0020] (4) Implementation of ecological prevention and control measures: Based on the warning level and risk type, corresponding prevention and control measures shall be initiated, including:
[0021] 1) Precise regulation of microecology: If the risk type is determined to be microecological imbalance, activate the probiotic liquid precise dripping device, add nitrifying bacteria and Bacillus compound bacterial liquid at a ratio of 0.05-0.1% of the aquaculture water volume, control the dripping rate at 50-200mL / min, and continue dripping for 8-12 hours; if the risk type is determined to be potential pathogen proliferation, activate the phage targeted delivery device, deliver specific phages matching the pathogen species, at a concentration of 105-107 PFU / mL, and deliver in three evenly distributed doses with an interval of 4-6 hours between each dose; if the risk type is determined to be organic load stress, reduce the daily feed amount by 30-50%, increase the frequency of circulating water filtration, and continuously monitor for 6-12 hours; if the dominant bacterial population does not return to the normal range within 24 hours, upgrade to a level three warning and initiate physical isolation.
[0022] 2) Physical isolation and emergency response: When a Level 3 warning occurs, the water circulation isolation system is triggered, the water connection pipes between high-risk aquaculture units and other units are shut off, and an independent water circulation mode is started; at the same time, the management personnel are notified through audible and visual alarms, and emergency response suggestions are generated;
[0023] Among the emergency response recommendations, at least one is to use underwater high-definition cameras to conduct non-destructive observation of high-risk individuals, or to transfer high-risk individuals to an independent quarantine unit within 2 hours.
[0024] (5) Construction of disease early warning and ecological control knowledge base: Record all data in steps (2)-(4) in real time, including microbial community data, water quality monitoring data, fish behavior data, early warning information, control measures parameters and implementation effect data, form individual files for each rare fish and environmental files for each aquaculture unit, update to the database established in step (1), and construct a live animal database data information and disease control knowledge base. The live animal database data information can map the physical state of the aquaculture unit in real time. The knowledge base is used to optimize the judgment threshold of subsequent early warning and the parameter settings of control measures.
[0025] The further improvement to the above technical solution is as follows:
[0026] The method for collecting microbial eDNA data in step (1) includes: collecting water samples from the normal living area of rare fish under normal living conditions, with a collection volume of 500 mL each time, filtering through a 0.22 μm filter membrane to extract eDNA, performing PCR amplification using universal primers for the 16S rRNA gene, and obtaining basic data of the microbial community after second-generation sequencing of the amplified products.
[0027] The steps for establishing a database of the normal survival status of rare fish species in step (1) are as follows:
[0028] 1) The year is used as the root node, which is the starting point of the tree structure database. All data is organized based on the root node.
[0029] 2) Add specific years as second-level nodes under the root node;
[0030] 3) Under each second-level node, add spring, summer, autumn, and winter as third-level nodes. These nodes represent different seasons in different years.
[0031] 4) Under each third-level node, add fish species as fourth-level nodes. These nodes represent the fish species studied in different years and seasons.
[0032] 5) Under each fourth-level node, further add fish individual behavior data parameter types, microbial eDNA data parameter types, and environmental data parameter types as fifth-level nodes. These nodes represent different parameter directions for different fish species in different years and seasons.
[0033] 6) Under the fifth layer node, further add data on individual fish behavior parameters, microbial eDNA data parameters and environmental data parameters under normal survival conditions as the sixth layer node. The data type of this layer node is an array, and these nodes record the data information under the corresponding category.
[0034] 7) Store the relevant data in the corresponding nodes.
[0035] The live animal culture unit mentioned in step (2) is an independent closed culture unit with a volume of 1-10 m³, a water depth of 0.8-1.5 m, and a culture density of 1-5 fish / m³. The automated water sample collection and filtration device includes a water sample collection pump, a multi-way valve, a filter membrane assembly, and a sample preservation module. The flow control accuracy of the water sample collection pump is ±1 mL / min. The multi-way valve can automatically switch between 16-32 culture units. The filter membrane assembly uses a polyethersulfone filter membrane with a filtration pressure of 0.1-0.3 MPa. The temperature of the sample preservation module is controlled at 4℃ to ensure the stability of the eDNA sample within 24 hours.
[0036] The method and steps for monitoring the movement intensity of individual fish using the individual behavior video monitoring unit in step (2) are as follows:
[0037] 1) Reference object setting: Three reference objects of known size are fixedly set in each breeding unit. The reference objects are cubes with a side length of 10cm and high-contrast markings on the surface. They are installed at the front, middle and rear of the breeding unit respectively.
[0038] 2) Image acquisition: A 4K high-definition underwater camera is used, with the frame rate set to 30fps. A continuous 10-second video clip is acquired every 5 minutes. At the same time, the distance L between the individual fish and the camera is measured in real time through a distance sensor.
[0039] 3) Image processing: Frame analysis is performed on the acquired video segments, and the outlines of individual fish and reference objects are extracted using edge detection algorithms to obtain the pixel length N of the individual fish and the pixel length n of the reference object in the image;
[0040] 4) Calculation of actual motion parameters: The actual movement length S of the individual fish is calculated using the formula S=Y×N×L / (n×I), where Y is the actual side length of the reference object and I is the fixed distance between the reference object and the camera; the movement speed V is calculated using the formula V=S / Δt, where Δt is the time interval between two adjacent frames; the movement intensity M per unit time is calculated using the formula M=V×T, where T is the monitoring period.
[0041] In step (3), the abnormal data early warning adopts a BP neural network model. The input layer of the model includes microbial community characteristic parameters, water quality parameters, and fish behavior parameters. The hidden layer is set with 3 layers, and the number of neurons in each layer is 32, 16, and 8, respectively. The output layer is the early warning level and risk type. The training error is controlled within 5% during the model training process.
[0042] The probiotic liquid precise dripping system in step (4) includes a bacterial liquid storage tank, a peristaltic pump, a flow sensor, a dripping pipeline and an atomizing nozzle. The flow rate adjustment accuracy of the peristaltic pump is ±0.1 mL / min. The flow sensor provides real-time feedback on the dripping amount. The PLC controller realizes closed-loop control of the dripping amount and dripping rate. The phage targeted delivery system includes a phage storage module, a concentration detection unit and a quantitative delivery pump. The concentration detection unit monitors the phage concentration using real-time fluorescence quantitative PCR technology to ensure that the delivery concentration error is no greater than 10%.
[0043] The water circulation isolation system in step (4) includes an electric valve, an independent circulating water pump, and an ultraviolet disinfection module. The response time of the electric valve is no more than 0.5 seconds, the flow rate of the independent circulating water pump is 5-10 m³ / h, the power of the ultraviolet disinfection module is 30-50W, and the disinfection dose is no less than 30 mJ / cm², ensuring that the pathogen removal rate of the circulating water is no less than 95%.
[0044] The abnormal data early warning in step (3) also includes a self-learning optimization mechanism. After each control measure is implemented and the early warning conditions are met, the system automatically adjusts the dynamic threshold range of the microbial community benchmark database, the abnormal behavior judgment criteria, and the water quality parameter early warning threshold. The adjustment range is ±2-5% each time to ensure the adaptability and accuracy of the early warning results.
[0045] The rare fish species mentioned include those under national first- or second-class protection.
[0046] As can be seen from the above technical solution, the present invention provides a method for disease early warning and ecological control in a live stock bank of rare fish. This method first constructs a standardized benchmark database by collecting three types of basic data: microbial eDNA, water quality environment, and fish behavior. Then, a real-time monitoring system consisting of eDNA monitoring, water quality sensing, and video surveillance is deployed. Anomaly detection is achieved based on a three-level progressive early warning rule combined with a BP neural network. Different methods are used to address different types of diseases, such as microecological imbalance, potential pathogen proliferation, and organic load stress. Physical isolation and emergency treatment are initiated at the third-level early warning. Data is recorded throughout the process to form individual and environmental profiles, and the system self-learns to optimize thresholds and parameters, achieving early detection, precise control, and low-damage protection of diseases in a live stock bank of rare fish. The present invention has the following advantages over existing technologies:
[0047] (1) The technical solution adopted in this invention captures early signals of disease occurrence through microbial eDNA monitoring unit, and can issue an early warning before fish show obvious symptoms. The early warning time is 24-48 hours earlier than the traditional method, which buys valuable time for the implementation of prevention and control measures and effectively reduces the morbidity and mortality of rare fish.
[0048] (2) The technical solution adopted in this invention integrates multi-dimensional data of microbial community, water quality parameters and fish behavior, and significantly improves the accuracy of early warning through three-level progressive early warning cross-verification, avoiding misjudgment and omission of single indicator monitoring.
[0049] (3) The technical solution adopted in this invention uses targeted prevention and control measures for different risk types. When the microbial community is imbalanced, probiotics are added, and when pathogens proliferate, specific bacteriophages are released. This avoids the use of broad-spectrum bactericides and antibiotics, and causes no stress or damage to rare fish. At the same time, it protects the microecological balance of the water body. Physical isolation measures effectively prevent the spread of pathogens and ensure the safety of the entire live fish bank.
[0050] (4) The technical solution adopted in this invention establishes a unified database and self-learning mechanism to realize full-process data traceability and closed-loop optimization, break data silos, continuously improve early warning accuracy and prevention and control effect. The whole process is automated, reducing manual intervention and management costs, while avoiding the subjectivity and error of manual operation. Attached Figure Description
[0051] Figure 1 Overall flowchart of disease early warning and ecological control methods for rare fish live stock banks;
[0052] Figure 2 A schematic diagram illustrating the abnormal data early warning logic and the implementation of ecological prevention and control measures. Detailed Implementation
[0053] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments, but the scope of protection of the present invention is not limited to the following embodiments.
[0054] The present invention provides a method for disease early warning and ecological control in a live stock bank of rare fish species, such as... Figure 1 and Figure 2 As shown, it includes the following steps:
[0055] (1) Basic data collection and database construction: Collect basic data on rare fish under normal living conditions over the years. The basic data includes microbial eDNA data, water quality environmental data, and individual fish behavior data; in collecting individual fish behavior data G nmsrk Simultaneously, microbial eDNA data E from the corresponding time and region are collected. nmspk and water quality environmental data F nmsqkn represents the year, m represents the season, s represents the fish species, r represents the type of individual fish behavior data parameters, p represents the type of microbial eDNA data parameters, q represents the type of environmental data parameters, and k represents the number of samples. n, m, s, r, p, and k are all natural numbers. A database of the normal survival status of rare fish species is established, which includes a microbial community benchmark database, a water quality parameter benchmark database, and a fish behavior benchmark characteristic database.
[0056] Among them, the data parameters in the microbial community benchmark library include at least the relative abundance E of the core dominant bacterial community. nms1k Relative abundance of potential pathogenic bacteria (E) nms2k Microbial Shannon-Wiener Diversity Index E nms3k The ratio of the abundance of core dominant bacterial groups to potential pathogenic bacterial genera is E. nms4k Abundance of nitrifying bacteria functional groups E nms5k and the evenness index of the microbial community E nms6k The water quality parameter benchmark database includes at least the oxidation-reduction potential (ORP). nms1k Ammonia nitrogen (NH3-N) concentration F nms2k Nitrite NO2 - -N concentration F nms3k Dissolved oxygen (DO) content nms4k and pH value F nms5k The data parameters in the fish behavior benchmark feature database should at least include swimming speed G per unit time. nms1k Feeding frequency G nms2k and the dispersion of the motion trajectory G nms3k ;
[0057] The method for collecting microbial eDNA data includes: collecting water samples from the normal living area of rare fish under normal living conditions, with a collection volume of 500 mL each time; extracting eDNA after filtration through a 0.22 μm filter membrane; performing PCR amplification using universal primers for the 16S rRNA gene; and obtaining basic data of the microbial community after second-generation sequencing of the amplified products.
[0058] The steps for establishing a database of the normal survival status of rare fish species are as follows:
[0059] 1) The year is used as the root node, which is the starting point of the tree structure database. All data is organized based on the root node.
[0060] 2) Add specific years as second-level nodes under the root node;
[0061] 3) Under each second-level node, add spring, summer, autumn, and winter as third-level nodes. These nodes represent different seasons in different years.
[0062] 4) Under each third-level node, add fish species as fourth-level nodes. These nodes represent the fish species studied in different years and seasons.
[0063] 5) Under each fourth-level node, further add fish individual behavior data parameter types, microbial eDNA data parameter types, and environmental data parameter types as fifth-level nodes. These nodes represent different parameter directions for different fish species in different years and seasons.
[0064] 6) Under the fifth layer node, further add data on individual fish behavior parameters, microbial eDNA data parameters and environmental data parameters under normal survival conditions as the sixth layer node. The data type of this layer node is an array, and these nodes record the data information under the corresponding category.
[0065] 7) Store the relevant data in the corresponding nodes.
[0066] (2) High-frequency monitoring equipment setup: A live animal culture unit was established, and a real-time monitoring device consisting of a microbial eDNA monitoring unit, a high-precision water quality sensor array, and an individual behavior video monitoring unit was deployed in the culture unit to obtain daily microbial eDNA data E. 0mspk Water quality environmental data F 0msqk and fish individual behavior data G 0msrk The microbial eDNA monitoring unit is the core monitoring unit. It collects water samples daily from each independent aquaculture unit using an automated water sample collection and filtration device, extracts eDNA from the samples, and performs rapid 16S rRNA gene sequencing using a portable nanopore sequencer to analyze the microbial community structure. The high-precision water quality sensor array monitors the temperature, pH, dissolved oxygen content, oxidation-reduction potential (ORP), ammonia nitrogen concentration, and nitrite concentration of the aquaculture water. The individual behavior video monitoring unit monitors the movement trajectory and intensity of individual fish in real time through a combination of image recognition and distance sensing. The temperature error is no greater than ±0.2℃, pH error no greater than ±0.2, dissolved oxygen error no greater than ±0.2 mg / L, ORP error no greater than ±5mV, ammonia nitrogen error no greater than ±5%, and nitrite error no greater than ±5% of full scale.
[0067] The live animal culture unit is an independent, enclosed culture unit with a volume of 1-10 m³, a water depth of 0.8-1.5 m, and a stocking density of 1-5 fish / m³. The automated water sample collection and filtration device includes a water sample collection pump, a multi-way valve, a filter membrane assembly, and a sample preservation module. The flow rate control accuracy of the water sample collection pump is ±1 mL / min. The multi-way valve can automatically switch between 16-32 culture units. The filter membrane assembly uses a polyethersulfone filter membrane with a filtration pressure controlled at 0.1-0.3 MPa. The temperature of the sample preservation module is controlled at 4℃ to ensure the stability of the eDNA sample within 24 hours.
[0068] The specific method and steps for the individual behavior video monitoring unit to monitor the movement intensity of individual fish are as follows:
[0069] 1) Reference object setting: Three reference objects of known size are fixedly set in each breeding unit. The reference objects are cubes with a side length of 10cm and high-contrast markings on the surface. They are installed at the front, middle and rear of the breeding unit respectively.
[0070] 2) Image acquisition: A 4K high-definition underwater camera is used, with the frame rate set to 30fps. A continuous 10-second video clip is acquired every 5 minutes. At the same time, the distance L between the individual fish and the camera is measured in real time through a distance sensor.
[0071] 3) Image processing: Frame analysis is performed on the acquired video segments, and the outlines of individual fish and reference objects are extracted using edge detection algorithms to obtain the pixel length N of the individual fish and the pixel length n of the reference object in the image;
[0072] 4) Calculation of actual motion parameters: The actual movement length S of the individual fish is calculated according to the formula S=Y×N×L / (n×I), where Y is the actual side length of the reference object and I is the fixed distance between the reference object and the camera; the movement speed V is calculated according to the formula V=S / Δt, where Δt is the time interval between two adjacent frames; the movement intensity M per unit time is calculated according to the formula M=V×T, where T is the monitoring period.
[0073] (3) Abnormal data early warning: Based on the real-time monitoring data obtained in step (2), the real-time monitoring data obtained in step (2) is processed to obtain the daily average microbial eDNA data E. 0msp Average water quality data F 0msq and mean data of individual fish behaviors G 0msr Then, the normal data from the database in step (1) is extracted and processed to obtain the average value data E of normal microbial eDNA data over the years. nmsp Mean data of normal water quality environment F nmsq The mean data G of normal individual fish behavior datanmsr The processed real-time monitoring data is then compared with the processed normal data. Based on the difference in the compared data, a three-level progressive early warning system is implemented, specifically including:
[0074] 1) Level 1 warning, i.e., microbial imbalance warning: This involves using the daily average microbial eDNA data (E...) 0msp Compared with the average value of normal microbial eDNA data over the years E nmsp Comparison was performed, and if the daily average microbial eDNA data E 0msp Compared with the average value of normal microbial eDNA data over the years E nmsp If the average data for one year exceeds the threshold, a Level 1 risk warning is issued, namely a warning of microbial imbalance.
[0075] The threshold mentioned above specifically refers to the mean relative abundance data E of the core dominant bacterial groups in the real-time microbial community structure. 0ms1 Deviation from the mean data of normal data E nms1 The amplitude exceeded the historical fluctuation baseline and the mean relative abundance data of potential pathogenic bacteria E. 0ms2 E, the average data compared to normal data nm2 An increase of more than 20% or a microbial diversity index E 0m3 E, the average data compared to normal data nm3 The reduction should be no less than 20%; that is, (|E nms1 -E 0ms1 |) / E nms1 >30%, (E 0ms2 -E nms2 ) / E nms2 >20% or (E nms3 -E 0ms3 ) / E nms3 ≥20%;
[0076] 2) Level II Early Warning, i.e., Abnormal Behavior Confirmation Early Warning: After a Level I early warning is triggered, individual fish behavior data will be collected through individual behavior video monitoring units, along with the average data G of normal individual fish behavior data over the years. nmsr By comparison, if an individual fish is found to have a feeding frequency that is at least 30% lower than the mean of normal data or a swimming speed that is at least 30% lower than the mean of normal data, i.e. (G) nms2 -G 0ms2 ) / G nms2 ≥30% or (G nms1 -G 0ms1 ) / G nms1 If the risk level is ≥30%, the system will raise the risk level to Level 2 and preliminarily determine the risk type as microecological imbalance, potential pathogen proliferation, or organic load stress based on the characteristics of microbial community changes.
[0077] The aforementioned microecological imbalance is specifically manifested in the mean relative abundance data E of the core dominant bacterial community. 0m1 E, the average data compared to normal data nm1 A decrease reflects the decline of beneficial functional groups; the proliferation of potential pathogens is specifically manifested in the mean relative abundance data E of potential pathogen genera. 0m2 E, the average data compared to normal data nm2 An increase indicates an increase in potential pathogenic microbial communities; the organic load stress type is specifically manifested by the mean relative abundance data E of the core dominant microbial community. 0m1 E, the average data compared to normal data nm1 An increase in water levels reflects the accumulation of organic matter in the water or an increase in fish shed material due to stress.
[0078] 3) Level III Early Warning, i.e., Water Quality Anomaly Evidence Warning: After a Level II early warning is triggered, based on monitoring data from a high-precision water quality sensor array, if the mean oxidation-reduction potential (ORP) data F is detected... 0ms1 The mean data F compared to normal data nms1 Reduce by at least 50mV, average ammonia nitrogen concentration data F 0ms2 The mean data F compared to normal data nms2 An increase of not less than 0.3 mg / L or the average nitrite concentration data F 0ms3 The mean data F compared to normal data nms3 If the increase is not less than 0.1 mg / L, the risk level will be raised to level three, and a comprehensive diagnosis will be completed.
[0079] The abnormal data early warning adopts a BP neural network model. The input layer of the model includes microbial community characteristic parameters, water quality parameters, and fish behavior parameters. There are 3 hidden layers with 32, 16, and 8 neurons in each layer, respectively. The output layer is the early warning level and risk type. The training error is controlled within 5% during the model training process.
[0080] The abnormal data early warning also includes a self-learning optimization mechanism. After each control measure is implemented and the conditions for lifting the early warning are met, the system automatically adjusts the dynamic threshold range of the microbial community benchmark database, the criteria for judging abnormal behavior, and the early warning thresholds for water quality parameters. The adjustment range is ±2-5% each time to ensure the adaptability and accuracy of the early warning results.
[0081] (4) Implementation of ecological prevention and control measures: Based on the warning level and risk type, corresponding prevention and control measures shall be initiated, including:
[0082] 1) Precise regulation of microecology: If the risk type is determined to be microecological imbalance, activate the probiotic liquid precise dripping device, add nitrifying bacteria and Bacillus compound bacterial liquid at a ratio of 0.05-0.1% of the aquaculture water volume, control the dripping rate at 50-200mL / min, and continue dripping for 8-12 hours; if the risk type is determined to be potential pathogen proliferation, activate the phage targeted delivery device, deliver specific phages matching the pathogen species, at a concentration of 105-107 PFU / mL, and deliver in three evenly distributed doses with an interval of 4-6 hours between each dose; if the risk type is determined to be organic load stress, reduce the daily feed amount by 30-50%, increase the frequency of circulating water filtration, and continuously monitor for 6-12 hours; if the dominant bacterial population does not return to the normal range within 24 hours, upgrade to a level three warning and initiate physical isolation.
[0083] 2) Physical isolation and emergency response: When a Level 3 warning occurs, the water circulation isolation system is triggered, the water connection pipes between high-risk aquaculture units and other units are shut off, and an independent water circulation mode is started; at the same time, the management personnel are notified through audible and visual alarms, and emergency response suggestions are generated;
[0084] Among the emergency response recommendations, at least one is to use underwater high-definition cameras to conduct non-destructive observation of high-risk individuals, or to transfer high-risk individuals to an independent quarantine unit within 2 hours.
[0085] The probiotic liquid precise dripping system includes a bacterial liquid storage tank, a peristaltic pump, a flow sensor, a dripping pipeline, and an atomizing nozzle. The flow rate adjustment accuracy of the peristaltic pump is ±0.1 mL / min, and the flow sensor provides real-time feedback on the dripping volume. A PLC controller enables closed-loop control of the dripping volume and dripping rate. The phage targeted delivery system includes a phage storage module, a concentration detection unit, and a quantitative delivery pump. The concentration detection unit monitors the phage concentration using real-time fluorescence quantitative PCR technology to ensure that the delivery concentration error is no greater than 10%.
[0086] The water circulation isolation system includes an electric valve, an independent circulating water pump, and an ultraviolet disinfection module. The response time of the electric valve is no more than 0.5 seconds, the flow rate of the independent circulating water pump is 5-10 m³ / h, the power of the ultraviolet disinfection module is 30-50W, and the disinfection dosage is no less than 30 mJ / cm², ensuring that the pathogen removal rate of the circulating water is no less than 95%.
[0087] (5) Construction of disease early warning and ecological control knowledge base: Record all data in steps (2)-(4) in real time, including microbial community data, water quality monitoring data, fish behavior data, early warning information, control measures parameters and implementation effect data, form individual files for each rare fish and environmental files for each aquaculture unit, update to the database established in step (1), and construct a live animal database data information and disease control knowledge base. The live animal database data information can map the physical state of the aquaculture unit in real time. The knowledge base is used to optimize the judgment threshold of subsequent early warning and the parameter settings of control measures.
[0088] In this embodiment, to verify the feasibility and effectiveness of the rare fish live-stock disease early warning and ecological control method of the present invention, rare protected fish species are selected as the target protected species of the live-stock. The core dominant bacterial community with high abundance and stable fluctuations under the microecological stability of the aquaculture system is used as the indicator group of dominant bacterial community stability. The core dominant bacterial community refers to the genera with the top 5 relative abundance and a fluctuation coefficient (standard deviation / mean) of less than 10% under the healthy state of the aquaculture system. Their abundance changes reflect the physical stability of the aquatic microecology. In this embodiment, *Acinetobacter* is used as an example, but it is not limited to this. At the same time, known potential conditional pathogen genera (*Aeromonas* is used as an example in this embodiment) are used as the key risk groups for monitoring. A standardized, independent, closed-loop aquaculture unit system is constructed. Systematic simulation and functional verification are carried out from five dimensions: database construction, real-time monitoring deployment, early warning logic verification, ecological control implementation, and knowledge base iteration. The specific steps are as follows:
[0089] Step 1: Collect microbial eDNA data, water quality data, and individual fish behavior data:
[0090] Microbial eDNA Data Collection: After the rare fish species adapted to the live stock culture environment and were in good health (no abnormal behavior and normal physiological indicators observed for 15 consecutive days), water samples were collected daily from each culture unit for 30 consecutive days, with each sample consisting of 500 mL. The collection point was located at a water depth of 50 cm in the middle of the culture unit. The samples were filtered using a 0.22 μm polyethersulfone membrane and stored at -8℃. eDNA was extracted within 24 hours using a DNA extraction kit (e.g., Qiagen, #80204, Germany). PCR amplification was performed using universal primers for the 16S rRNA gene (forward primer: 341F 5′-CCTACGGGNGGCWGCAG-3′, reverse primer: 805R 5′-GACTACHVGGGTATCTAATCC-3′). The amplified products were then subjected to next-generation sequencing to obtain microbial community data. Through bioinformatics analysis, the baseline abundance ranges of indicator genera and potential pathogen genera were determined, the baseline values of the microbial diversity index (Shannon-Wiener index) were calculated, and a baseline database of microbial communities was established.
[0091] Water quality environmental data acquisition: Water quality parameters were collected synchronously for 30 days, including temperature, pH value, dissolved oxygen content, oxidation-reduction potential (ORP), ammonia nitrogen concentration, and nitrite concentration. Each parameter was collected once per hour. High-precision sensors (such as the HachHQ40d multi-parameter water quality analyzer) were used for detection. The average value and standard deviation of each parameter were recorded to determine the benchmark range of water quality parameters under normal living conditions.
[0092] Fish individual behavior data collection: Through individual behavior video monitoring units, fish movement trajectories, movement speeds, feeding frequencies and other behavioral parameters are recorded continuously for 30 days. Ten minutes of video data are collected every hour, and the baseline values and fluctuation ranges of each behavioral parameter are analyzed to establish a fish behavior baseline feature database.
[0093] Step 2: Establish a real-time monitoring system consisting of a microbial eDNA monitoring unit, a high-precision water quality sensor array, and an individual behavior video monitoring unit to achieve real-time acquisition of multi-dimensional data.
[0094] Microbial eDNA Monitoring Unit: This is the core monitoring unit, comprising an automated water sample collection and filtration device, an eDNA extraction module, and a portable nanopore sequencer (such as the Oxford Nanopore MinION). The automated water sample collection and filtration device, controlled by a PLC, collects 500 mL of water from each aquaculture unit at a fixed time daily (e.g., 9:00 AM). After switching through a multi-port valve, the sample is filtered through a membrane filtration assembly. The filter membrane is automatically transferred to the eDNA extraction module, where a rapid extraction kit (extraction time no more than 30 minutes) is used to extract eDNA. The extracted eDNA sample is then directly fed into the portable nanopore sequencer for rapid 16S rRNA gene sequencing. The sequencing time is no more than 6 hours. The sequencing data is transmitted in real-time to the data processing center to analyze the microbial community structure, including parameters such as the abundance and diversity index of each genera.
[0095] High-precision water quality sensor array: Three sensor monitoring points (front, middle, and rear) are set up in each aquaculture unit. Each monitoring point is equipped with sensors for temperature, pH, dissolved oxygen, ORP, ammonia nitrogen, and nitrite. In this embodiment, the monitoring accuracy of the sensors meets the following requirements: temperature error ≤ ±0.1℃, pH error ≤ ±0.05, dissolved oxygen error ≤ ±0.1mg / L, ORP error ≤ ±5mV, ammonia nitrogen error ≤ ±0.05mg / L, and nitrite error ≤ ±0.02mg / L, which is higher than the monitoring accuracy of ordinary aquaculture. Sensor data is collected every 30 minutes and uploaded to the data processing center in real time.
[0096] Individual Behavior Video Monitoring Unit: Three 4K high-definition underwater cameras are installed above and to the side of each aquaculture unit, along with three fixed reference objects, to achieve precise monitoring of individual fish behavior. The reference objects are 10cm cubes with high-contrast black and white markings, installed at the front, middle, and rear of the aquaculture unit respectively, for image scaling. The camera frame rate is set to 30fps, capturing a continuous 10-second video clip every 5 minutes. Simultaneously, a distance sensor (such as a laser rangefinder) measures the distance L between the individual fish and the camera in real time. The video data is analyzed by the image processing module to extract the fish's outline and motion parameters, using the following method:
[0097] 1. Image preprocessing: Denoising and enhancing video frames to improve contour recognition accuracy;
[0098] 2. Contour Extraction: The Canny edge detection algorithm is used to extract the contours of individual fish and the reference object.
[0099] 3. Pixel length measurement: The pixel length N of an individual fish in the image and the pixel length n of a reference object are measured using image recognition software;
[0100] 4. Calculation of actual motion parameters: The actual motion length S is calculated according to the formula S=Y×N×L / (n×I) (Y is the actual side length of the reference object, 10cm, and I is the fixed distance between the reference object and the camera). The motion speed V is calculated according to the formula V=S / Δt (Δt is the time interval between adjacent frames, 1 / 30s). The motion intensity and feeding frequency are calculated by the motion distance and the number of feedings per unit time.
[0101] 5. Data Upload: The calculated behavioral parameters are uploaded to the data processing center in real time.
[0102] Step 3: Based on the real-time monitoring data obtained in step (2), compare the real-time monitoring data with the normal data in the database in step (1), and implement a three-level progressive early warning based on the difference in the compared data:
[0103] A backpropagation (BP) neural network-based early warning model was constructed. The input layer included six microbial community characteristic parameters (relative abundance of *Acinetobacter* and *Aeromonas*, Shannon-Wiener diversity index, abundance ratio of dominant bacteria to potential pathogens, abundance of nitrifying bacteria functional groups, and microbial community evenness index), five water quality parameters, and three fish behavior parameters. Three hidden layers were used. The output layer displayed the early warning level and risk type (microecological imbalance, potential pathogen proliferation, and organic load stress). Gradient descent was used for model training, with the training error controlled within 5% to ensure the model's predictive accuracy.
[0104] Level 1 Warning (Microbial Imbalance Warning): This involves using the daily average microbial eDNA data (E...) 0msp Compare the data with the mean value of normal microbial eDNA data in step 1. If the daily mean value of microbial eDNA data E 0msp If the value exceeds the threshold compared to the mean value of normal microbial eDNA data, a Level 1 risk warning is issued, namely a warning of microbial imbalance.
[0105] The threshold mentioned above specifically refers to the mean relative abundance data (E) of *Acinetobacter* genus in the real-time microbial community structure. 0ms1 E, the average data compared to normal data nms1 Deviations exceeding ±30% in the mean relative abundance data of Aeromonas spp. 0ms2 E, the average data compared to normal data nm2 An increase of more than 20% or a microbial diversity index E 0m3 E, the average data compared to normal data nm3 The reduction should be no less than 20%; that is, (|E nms1 - E 0ms1 |) / E nms1>30%, (E 0ms2 - E nms2 ) / E nms2 >20% or (E nms3 - E 0ms3 ) / E nms3 ≥20%; among them, a decrease of more than 30% in Acinetobacter triggers a warning of microecological decline, and an increase of more than 30% triggers a warning of organic load stress.
[0106] The aforementioned warning thresholds (deviation ±30%, increase >30%, decrease ≥20%) are dynamically determined based on the standard deviation (SD) of each parameter within a healthy aquaculture cycle, specifically ranging from the mean to two times the standard deviation. The percentages given in this embodiment are typical application values, and can be adjusted according to the specific rare fish species, stocking density, and season in actual implementation.
[0107] At this time, the fish may behave completely normally, but the data can identify early signs of disease and issue a level one risk warning, prompting managers to pay attention to changes in the microecology of the aquaculture unit.
[0108] Level 2 Warning (Behavioral Anomaly Confirmation Warning): After a Level 1 warning is triggered, individual fish behavior data will be collected via individual behavior video monitoring units, along with the average data G of normal individual fish behavior data over the years. nmsr By comparison, if an individual fish is found to have a feeding frequency that is at least 30% lower than the mean of normal data or a swimming speed that is at least 30% lower than the mean of normal data, i.e. (G) nms2 - G 0ms2 ) / G nms2 ≥30% or (G nms1 - G 0ms1 ) / G nms1 If the risk level is ≥30%, the system will raise the risk level to Level 2 and preliminarily determine the risk type as either microecological imbalance or potential pathogen proliferation based on the characteristics of microbial community changes.
[0109] Simultaneously, based on the characteristics of microbial community changes, the risk type was preliminarily determined; the microecological imbalance type was specifically manifested by the mean relative abundance data of Acinetobacter spp. E. 0m1 E, the average data compared to normal data nm1 Decline; the proliferative form of potential pathogens is specifically manifested in the mean relative abundance data of Aeromonas spp. E 0m2 E, the average data compared to normal data nm2 Increase; organic load stress type is specifically manifested in the mean relative abundance data of Acinetobacter E 0m1 E, the average data compared to normal data nm1 rise;
[0110] Level 3 Warning (Water Quality Anomaly Evidence Warning): After a Level 2 warning is triggered, based on monitoring data from a high-precision water quality sensor array, if the mean oxidation-reduction potential (ORP) data F is detected... 0ms1 The mean data F compared to normal data nms1 Reduce by at least 50mV, average ammonia nitrogen concentration data F 0ms2 The mean data F compared to normal data nms2 An increase of not less than 0.3 mg / L or the average nitrite concentration data F 0ms3 The mean data F compared to normal data nms3 If the level rises by at least 0.1 mg / L, the risk level will be raised to level three, and a comprehensive diagnosis will be completed to confirm the authenticity and severity of the disease risk.
[0111] Step 4: Based on the warning level and risk type, activate the corresponding prevention and control measures to ensure the accuracy and safety of prevention and control:
[0112] Differentiated ecological control plans are implemented based on the risk type identified in the early warning. For risks related to microecological imbalance, a precise probiotic dripping device is activated, supplementing the aquaculture water with a compound bacterial solution of nitrifying bacteria and Bacillus at a ratio of 0.05-0.1% of the water volume. The dripping rate is controlled at 50-200 mL / min, and dripping is continued for 8-12 hours to regulate the microecological balance of the water. For risks related to potential pathogen proliferation, a phage targeted delivery device is activated, delivering specific phages matching the pathogen species at a concentration of 10⁵-10⁷ PFU / mL. The delivery is performed in three evenly distributed doses, with an interval of 4-6 hours between each dose, precisely inhibiting pathogen proliferation and avoiding stress damage to rare fish and disruption of the aquatic microecology caused by broad-spectrum drugs. For risks related to organic load stress, the daily feeding amount is reduced by 30-50%, the frequency of circulating water filtration is increased, and continuous monitoring is conducted for 6-12 hours. If the potential pathogen genera or dominant bacterial groups do not return to normal within 24 hours, the warning level is upgraded to Level III. When a Level 3 warning is triggered, the water circulation isolation system is activated to quickly isolate high-risk aquaculture units, initiate independent circulation and ultraviolet disinfection to prevent the spread of pathogens, and simultaneously push out warning information and provide emergency response suggestions such as non-destructive observation and individual isolation to ensure the overall safety of the live animal storage facility.
[0113] Step 5, Knowledge Base Construction and Continuous Optimization:
[0114] The entire process records monitoring data, early warning information, control parameters, and implementation effects, creating individual breeding files for rare fish species and environmental files for breeding units. These files are continuously updated to a baseline database, constructing a knowledge base for disease early warning and ecological control. Based on a self-learning optimization mechanism of the knowledge base, the dynamic adjustment capability of early warning thresholds and control parameters is verified. After each control measure achieves its target, the early warning thresholds related to microorganisms, water quality, and behavior, as well as the parameters for bacterial solution / phage administration, are fine-tuned based on the actual results. This gradually improves the accuracy of early warnings and the adaptability of control measures, forming a closed-loop management system integrating monitoring, early warning, control, and optimization.
[0115] This embodiment, through a systematic research and development approach, fully verifies the feasibility of applying the method of this invention in a live stock bank of rare fish species. It achieves early disease warning, precise ecological control, and continuous optimization driven by data, with no damage or antibiotic use throughout the process. This meets the low-interference protection requirements for rare aquatic organisms and provides a reliable technical paradigm and implementation path for the ex-situ conservation of national first- and second-class protected fish species.
Claims
1. A method for disease early warning and ecological control in a live stock bank of rare fish species, characterized in that, Includes the following steps: (1) Basic data collection and database construction: Collect basic data on rare fish under normal living conditions over the years. The basic data includes microbial eDNA data, water quality environmental data, and individual fish behavior data; in collecting individual fish behavior data G nmsrk Simultaneously, microbial eDNA data E from the corresponding time and region are collected. nmspk and water quality environmental data F nmsqk n represents the year, m represents the season, s represents the fish species, r represents the type of individual fish behavior data parameters, p represents the type of microbial eDNA data parameters, q represents the type of environmental data parameters, and k represents the number of samples. n, m, s, r, p, and k are all natural numbers. A database of the normal survival status of rare fish species is established, which includes a microbial community benchmark database, a water quality parameter benchmark database, and a fish behavior benchmark characteristic database. Among them, the data parameters in the microbial community benchmark library include at least the relative abundance E of the core dominant bacterial community. nms1k Relative abundance of potential pathogenic bacteria (E) nms2k Microbial Shannon-Wiener Diversity Index E nms3k The ratio of the abundance of core dominant bacterial groups to potential pathogenic bacterial genera is E. nms4k Abundance of nitrifying bacteria functional groups E nms5k and the evenness index of the microbial community E nms6k The water quality parameter benchmark database includes at least the oxidation-reduction potential (ORP). nms1k Ammonia nitrogen (NH3-N) concentration F nms2k Nitrite NO2 - -N concentration F nms3k Dissolved oxygen (DO) content nms4k and pH value F nms5k The data parameters in the fish behavior benchmark feature database should at least include swimming speed G per unit time. nms1k Feeding frequency G nms2k and the dispersion of the motion trajectory G nms3k ; (2) High-frequency monitoring equipment setup: A live animal culture unit was established, and a real-time monitoring device consisting of a microbial eDNA monitoring unit, a high-precision water quality sensor array, and an individual behavior video monitoring unit was deployed in the culture unit to obtain daily microbial eDNA data E. 0mspk Water quality environmental data F 0msqk and fish individual behavior data G 0msrk The microbial eDNA monitoring unit is the core monitoring unit. It collects water samples daily from each independent aquaculture unit using an automated water sample collection and filtration device, extracts eDNA from the samples, and performs rapid 16S rRNA gene sequencing using a portable nanopore sequencer to analyze the microbial community structure. The high-precision water quality sensor array monitors the temperature, pH, dissolved oxygen content, oxidation-reduction potential (ORP), ammonia nitrogen concentration, and nitrite concentration of the aquaculture water. The individual behavior video monitoring unit monitors the movement trajectory and intensity of individual fish in real time through a combination of image recognition and distance sensing. The temperature error is no greater than ±0.2℃, pH error no greater than ±0.2, dissolved oxygen error no greater than ±0.2 mg / L, ORP error no greater than ±5mV, ammonia nitrogen error no greater than ±5%, and nitrite error no greater than ±5% of full scale. (3) Abnormal data early warning: Based on the real-time monitoring data obtained in step (2), the real-time monitoring data obtained in step (2) is processed to obtain the daily average microbial eDNA data E. 0msp Average water quality data F 0msq and mean data of individual fish behavior G 0msr Then, the normal data from the database in step (1) is extracted and processed to obtain the average value data E of normal microbial eDNA data over the years. nmsp Mean data of normal water quality environment F nmsq The mean data G of normal individual fish behavior data nmsr The processed real-time monitoring data is then compared with the processed normal data. Based on the difference in the compared data, a three-level progressive early warning system is implemented, specifically including: 1) Level 1 warning, i.e., microbial imbalance warning: This involves using the daily average microbial eDNA data (E...) 0msp Compared with the average value of normal microbial eDNA data over the years E nmsp Comparison was performed, and if the daily average microbial eDNA data E 0msp Compared with the average value of normal microbial eDNA data over the years E nmsp If the average data for one year exceeds the threshold, a Level 1 risk warning is issued, namely a warning of microbial imbalance. The threshold mentioned above specifically refers to the mean relative abundance data E of the core dominant bacterial groups in the real-time microbial community structure. 0ms1 Deviation from the mean data of normal data E nms1 The amplitude exceeded the historical fluctuation baseline and the mean relative abundance data of potential pathogenic bacteria E. 0ms2 E, the average data compared to normal data nm2 An increase of more than 20% or a microbial diversity index E 0m3 E, the average data compared to normal data nm3 The reduction should be no less than 20%; that is, (|E nms1 -E 0ms1 |) / E nms1 >30%, (E 0ms2 -E nms2 ) / E nms2 >20% or (E nms3 -E 0ms3 ) / E nms3 ≥20%; 2) Level II Early Warning, i.e., Abnormal Behavior Confirmation Early Warning: After a Level I early warning is triggered, individual fish behavior data will be collected through individual behavior video monitoring units, along with the average data G of normal individual fish behavior data over the years. nmsr By comparison, if an individual fish is found to have a feeding frequency that is at least 30% lower than the mean of normal data or a swimming speed that is at least 30% lower than the mean of normal data, i.e. (G) nms2 -G 0ms2 ) / G nms2 ≥30% or (G nms1 -G 0ms1 ) / G nms1 If the risk level is ≥30%, the system will raise the risk level to Level 2 and preliminarily determine the risk type as microecological imbalance, potential pathogen proliferation, or organic load stress based on the characteristics of microbial community changes. The aforementioned microecological imbalance is specifically manifested in the mean relative abundance data E of the core dominant bacterial community. 0m1 E, the average data compared to normal data nm1 Decline; the proliferation of potential pathogens is specifically manifested in the mean relative abundance data of potential pathogen genera E. 0m2 E, the average data compared to normal data nm2 An increase; the organic load stress type is specifically manifested in the mean relative abundance data E of the core dominant bacterial community. 0m1 E, the average data compared to normal data nm1 rise; 3) Level III Early Warning, i.e., Water Quality Anomaly Evidence Warning: After a Level II early warning is triggered, based on monitoring data from a high-precision water quality sensor array, if the mean oxidation-reduction potential (ORP) data F is detected... 0ms1 The mean data F compared to normal data nms1 Reduce by at least 50mV, average ammonia nitrogen concentration data F 0ms2 The mean data F compared to normal data nms2 An increase of not less than 0.3 mg / L or the average nitrite concentration data F 0ms3 The mean data F compared to normal data nms3 If the increase is not less than 0.1 mg / L, the risk level will be raised to level three, and a comprehensive diagnosis will be completed. (4) Implementation of ecological prevention and control measures: Based on the warning level and risk type, corresponding prevention and control measures shall be initiated, including: 1) Precise regulation of the microecology: If the risk type is determined to be microecological imbalance, activate the precise dripping device for probiotic solution, adding nitrifying bacteria and Bacillus compound bacterial solution at a ratio of 0.05-0.1% of the aquaculture water volume, with the dripping rate controlled at 50-200 mL / min, and continuously dripping for 8-12 hours; if the risk type is determined to be potential pathogen proliferation, activate the phage targeted delivery device, delivering specific phages matching the pathogen species at a concentration of 105-107 PFU / mL, delivered evenly in three portions, with an interval of 4-6 hours between each delivery; if the risk type is determined to be organic load stress, reduce the daily feeding amount by 30-50%, increase the frequency of circulating water filtration, and continuously monitor for 6-12 hours; if the dominant bacterial population does not return to the normal range within 24 hours, the alert level is upgraded to Level 3. 2) Physical isolation and emergency response: When a Level 3 warning occurs, the water circulation isolation system is triggered, the water connection pipes between high-risk aquaculture units and other units are shut off, and an independent water circulation mode is started; at the same time, the management personnel are notified through audible and visual alarms, and emergency response suggestions are generated; Among the emergency response recommendations, at least one is to use underwater high-definition cameras to conduct non-destructive observation of high-risk individuals, or to transfer high-risk individuals to an independent quarantine unit within 2 hours. (5) Construction of disease early warning and ecological control knowledge base: Record all data in steps (2)-(4) in real time, including microbial community data, water quality monitoring data, fish behavior data, early warning information, control measures parameters and implementation effect data, form individual files for each rare fish and environmental files for each aquaculture unit, update to the database established in step (1), and construct a live animal database data information and disease control knowledge base. The live animal database data information can map the physical state of the aquaculture unit in real time. The knowledge base is used to optimize the judgment threshold of subsequent early warning and the parameter settings of control measures.
2. The method for disease early warning and ecological control in a live stock bank of rare fish according to claim 1, characterized in that: The method for collecting microbial eDNA data in step (1) includes: collecting water samples from the normal living area of rare fish under normal living conditions, with a collection volume of 500 mL each time, filtering through a 0.22 μm filter membrane to extract eDNA, performing PCR amplification using universal primers for the 16S rRNA gene, and obtaining basic data of the microbial community after second-generation sequencing of the amplified products.
3. The method for disease early warning and ecological control in a live stock bank of rare fish according to claim 1, characterized in that... The steps for establishing a database of the normal survival status of rare fish species in step (1) are as follows: 1) The year is used as the root node, which is the starting point of the tree structure database. All data is organized based on the root node. 2) Add specific years as second-level nodes under the root node; 3) Under each second-level node, add spring, summer, autumn, and winter as third-level nodes. These nodes represent different seasons in different years. 4) Under each third-level node, add fish species as fourth-level nodes. These nodes represent the fish species studied in different years and seasons. 5) Under each fourth-level node, further add fish individual behavior data parameter types, microbial eDNA data parameter types, and environmental data parameter types as fifth-level nodes. These nodes represent different parameter directions for different fish species in different years and seasons. 6) Under the fifth layer node, further add data on individual fish behavior parameters, microbial eDNA data parameters and environmental data parameters under normal survival conditions as the sixth layer node. The data type of this layer node is an array, and these nodes record the data information under the corresponding category. 7) Store the relevant data in the corresponding nodes.
4. The method for disease early warning and ecological control in a live stock bank of rare fish according to claim 1, characterized in that: The live animal culture unit mentioned in step (2) is an independent closed culture unit with a volume of 1-10 m³, a water depth of 0.8-1.5 m, and a culture density of 1-5 fish / m³. The automated water sample collection and filtration device includes a water sample collection pump, a multi-way valve, a filter membrane assembly, and a sample preservation module. The flow control accuracy of the water sample collection pump is ±1 mL / min. The multi-way valve can automatically switch between 16-32 culture units. The filter membrane assembly uses a polyethersulfone filter membrane with a filtration pressure of 0.1-0.3 MPa. The temperature of the sample preservation module is controlled at 4℃ to ensure the stability of the eDNA sample within 24 hours.
5. The method for disease early warning and ecological control in a live stock bank of rare fish according to claim 1, characterized in that: The method and steps for monitoring the movement intensity of individual fish using the individual behavior video monitoring unit in step (2) are as follows: 1) Reference object setting: Three reference objects of known size are fixedly set in each breeding unit. The reference objects are cubes with a side length of 10cm and high-contrast markings on the surface. They are installed at the front, middle and rear of the breeding unit respectively. 2) Image acquisition: A 4K high-definition underwater camera is used, with the frame rate set to 30fps. A continuous 10-second video clip is acquired every 5 minutes. At the same time, the distance L between the individual fish and the camera is measured in real time through a distance sensor. 3) Image processing: Frame analysis is performed on the acquired video segments, and the outlines of individual fish and reference objects are extracted using edge detection algorithms to obtain the pixel length N of the individual fish and the pixel length n of the reference object in the image; 4) Calculation of actual motion parameters: The actual movement length S of the individual fish is calculated according to the formula S=Y×N×L / (n×I), where Y is the actual side length of the reference object and I is the fixed distance between the reference object and the camera; the movement speed V is calculated according to the formula V=S / Δt, where Δt is the time interval between two adjacent frames. The intensity of motion per unit time, M, is calculated using the formula M=V×T, where T is the monitoring period.
6. The method for disease early warning and ecological control in a live stock bank of rare fish according to claim 1, characterized in that: In step (3), the abnormal data early warning adopts a BP neural network model. The input layer of the model includes microbial community characteristic parameters, water quality parameters, and fish behavior parameters. The hidden layer is set with 3 layers, and the number of neurons in each layer is 32, 16, and 8, respectively. The output layer is the early warning level and risk type. The training error is controlled within 5% during the model training process.
7. The method for disease early warning and ecological control in a live stock bank of rare fish according to claim 1, characterized in that: The probiotic liquid precise dripping system in step (4) includes a bacterial liquid storage tank, a peristaltic pump, a flow sensor, a dripping pipeline and an atomizing nozzle. The flow rate adjustment accuracy of the peristaltic pump is ±0.1 mL / min. The flow sensor provides real-time feedback on the dripping amount. The PLC controller realizes closed-loop control of the dripping amount and dripping rate. The phage targeted delivery system includes a phage storage module, a concentration detection unit and a quantitative delivery pump. The concentration detection unit monitors the phage concentration using real-time fluorescence quantitative PCR technology to ensure that the delivery concentration error is no greater than 10%.
8. The method for disease early warning and ecological control in a live stock bank of rare fish according to claim 1, characterized in that: The water circulation isolation system in step (4) includes an electric valve, an independent circulating water pump, and an ultraviolet disinfection module. The response time of the electric valve is no more than 0.5 seconds, the flow rate of the independent circulating water pump is 5-10 m³ / h, the power of the ultraviolet disinfection module is 30-50W, and the disinfection dose is no less than 30 mJ / cm², ensuring that the pathogen removal rate of the circulating water is no less than 95%.
9. The method for disease early warning and ecological control in a live stock bank of rare fish according to claim 1, characterized in that: The abnormal data early warning in step (3) also includes a self-learning optimization mechanism. After each control measure is implemented and the early warning conditions are met, the system automatically adjusts the dynamic threshold range of the microbial community benchmark database, the abnormal behavior judgment criteria, and the water quality parameter early warning threshold. The adjustment range is ±2-5% each time to ensure the adaptability and accuracy of the early warning results.
Citation Information
Patent Citations
Method and device for monitoring activity of fish
CN106561532A
Prawn culture water source pathogen identification primer group, identification method, kit and application
CN119265328A
