Method for identifying growth quality of transplanted arbor in alpine region

Abstract

The invention discloses an alpine region transplanted arbor growth quality identification method, and relates to the technical field of forestry. The method comprises the following steps: taking unhealthy, sub-healthy and healthy picea crassifolia transplanted for 1-2 years in the alpine region as objects, collecting plant branch and leaf samples and soil samples, measuring 15 parameters as characterization indexes of the health condition, and analyzing through a random forest model, a proportional dominance model and factor difference. According to the method, six parameters sensitive to the health quality condition of the arbor are obtained, the judgment threshold value of each factor on the growth quality is obtained, health indexes are constructed based on the six parameters and used for comprehensively evaluating the health condition of the plant, the comprehensive judgment threshold value of the growth condition of the plant is obtained, and the survival condition of the transplanted arbor is accurately judged. The adaptability of the method to the transplanting environment is analyzed, and a scientific basis is provided for later site environment improvement and arbor protection.

Description

Technical Field

[0001] This invention relates to the field of forestry technology, and more specifically to a method for identifying the growth quality of transplanted trees in high-altitude and cold regions. Background Technology

[0002] The Qinghai-Tibet Plateau is rich in hydropower resources, and the construction of large-scale hydropower stations has strongly supported local power generation and energy storage needs. However, the inundation areas created by these projects will significantly alter the original ecological environment, posing a risk of habitat loss and even extinction for centuries-old trees. Scientific transplantation to suitable habitats can effectively ensure the survival and reproduction of these plants, preventing the loss of species resources caused by engineering development. Scientific transplantation not only minimizes the damage to the ecosystem caused by hydropower station construction but also maintains the integrity of the regional ecosystem's structure and function, effectively safeguarding ecological balance and laying the foundation for the stability and sustainable development of the ecosystem.

[0003] However, high-altitude and cold regions often face harsh environments such as extreme low temperatures, strong radiation, strong winds, and short growing seasons, posing a severe challenge to the survival and growth of transplanted trees. Transplanted trees generally suffer from low survival rates, long recovery periods, and delayed growth, and latent stresses such as root frost damage and physiological disorders are difficult to detect. Currently, assessing the growth quality of transplanted trees mainly relies on visual inspection, which is highly subjective and inefficient. Especially for tall trees with long recovery periods and delayed growth, visual inspection alone is insufficient to effectively identify their true growth quality, making it difficult to accurately assess the effectiveness of transplantation. However, there is currently no relevant indicator system available to determine the growth quality of transplanted trees in high-altitude and cold regions, seriously affecting the high-quality implementation of transplantation work in these areas.

[0004] Therefore, there is an urgent need to develop a scientific monitoring and identification method for the growth quality of transplanted trees that is specifically suitable for high-altitude and cold environments, so as to achieve rapid and accurate identification of the growth status of transplanted trees and provide key technical support for improving the survival rate of transplanted trees and building a stable ecological barrier. Summary of the Invention

[0005] In view of this, the present invention provides a method for identifying the growth quality of transplanted trees in high-altitude and cold regions.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] A method for identifying the growth quality of transplanted trees in high-altitude and cold regions, the method comprising the following steps:

[0008] Step 1: In the high-altitude and cold regions above 3200m on the Qinghai-Tibet Plateau, Qinghai spruce that has been transplanted for 1-2 years was used as the research object. In summer, several trees of three health quality categories (unhealthy, sub-healthy, and healthy) were selected by manual identification. Fresh leaves and branches were obtained, and soil samples from the 0-100cm soil layer at 50cm from the base of the tree were collected.

[0009] Step 2: Determine the total carbon (TC) in tree leaves. L ), total nitrogen (TN) L Total phosphorus (TP) L Content, calculate the leaf stoichiometry parameters C:N L C:P L N:P L The values ​​were determined, and the differences between the parameters under healthy, sub-healthy, and unhealthy conditions were identified.

[0010] Step 3: Determine the total carbon (TC) of tree branches. B ), total nitrogen (TN) B Total phosphorus (TP) B Content, calculate the stoichiometric parameters C:N of the branches. B C:P B N:P B The values ​​were determined, and the differences between the parameters under healthy, sub-healthy, and unhealthy conditions were identified.

[0011] Step 4: Measure the chlorophyll content (CHL) of plant leaves and the malondialdehyde (MDA) content of plant branches. Malondialdehyde is an important indicator of plant stress physiology. The lower the plant growth quality, the higher the MDA content. Also, determine the differences of each parameter under healthy, sub-healthy, and unhealthy conditions.

[0012] Step 5: Measure the hydrogen and oxygen isotope concentrations of plant water and soil water, and analyze the proportion of water utilization by plants in different soil layers; the higher the health of the plant, the higher the proportion of water utilization in deeper soil layers. Therefore, the proportion of water absorbed by the plant in the 40-100cm soil layer (WU) is used as an indicator of health status. The higher the WU value, the better the health status. And identify the differences in parameters under healthy, sub-healthy and unhealthy conditions.

[0013] Step 6: Using random forest and proportional dominance models, combined with factor variance analysis, determine the sensitivity of plant growth quality to various factors; and the impact of plant health status on WU, MDA, CHL, and C:N. B N:P B TC B The six parameters showed the highest sensitivity;

[0014] Step 7: Using three types of machine learning models—grid search, genetic algorithm, and Bayesian optimization algorithm—to comprehensively identify the threshold for judging plant growth status, and to screen rapid indicators for identifying plant growth quality;

[0015] Step 8: Based on the above 6 parameters, construct a health index to comprehensively assess the health status of plants. Use three types of machine learning models—grid search, genetic algorithm, and Bayesian optimization algorithm—to identify health index thresholds under different health conditions, which serve as the criteria for judging plant growth status.

[0016] Furthermore, step 1 is as follows:

[0017] In the high-altitude and cold regions of the Qinghai-Tibet Plateau above 3200m, Qinghai spruce trees transplanted for 1-2 years were selected as the research object. Ten trees of each of three categories of growth quality were selected: unhealthy, sub-healthy, and healthy. At the same time, another 30 trees were selected to test the applicability of the identification method. Fresh leaves and branches of Qinghai spruce were obtained by pruning shears, and soil samples were collected from the 0-100cm soil layer 50cm away from the base of the plant.

[0018] Furthermore, step 2 is as follows:

[0019] The collected plant leaf samples were blanched in an electrically heated drying oven at 65°C; the total carbon (TC) was determined using an elemental analyzer. L ) and total nitrogen (TN) L The total phosphorus (TP) content was determined using a fully automated discrete chemical analyzer. L Content; calculation of leaf stoichiometry parameters C:N L C:P L N:P L Values; differences in parameters under healthy, sub-healthy, and unhealthy conditions were analyzed using one-way ANOVA; TN values ​​under healthy and sub-healthy conditions. L The differences were not significant (P>0.05), but highly significant (P<0.01) were observed between healthy and unhealthy individuals, and between those in a sub-healthy state and those in an unhealthy state. Other parameters showed no significant differences. This indicates that TN... L It can be used as a characterization indicator of plant health quality.

[0020] Furthermore, step 3 is as follows:

[0021] The collected plant branch samples were blanched in an electrically heated drying oven at 65°C; after grinding using a ball mill, the total carbon (TC) was determined using an elemental analyzer. B ) and total nitrogen (TN) B The total phosphorus (TP) content was determined using a fully automated discrete chemical analyzer. B Content; calculation of stoichiometric parameters C:N in branches B C:P B N:PB Values; Factor differences among healthy, sub-healthy, and unhealthy conditions were analyzed using one-way ANOVA; Total N / T (TN) values ​​were analyzed under healthy and sub-healthy conditions. B C:N B N:P B The differences were not significant (P>0.05), but highly significant differences were found between healthy and unhealthy individuals, and between those in a sub-healthy state and those in an unhealthy state (P<0.01). Other parameters showed no significant differences. This indicates that TN... B C:N B N:P B It can be used as a characterization indicator of plant health quality.

[0022] Furthermore, step 4 is as follows:

[0023] Chlorophyll content (CHL) in plant leaves was determined by spectrophotometry, and malondialdehyde (MDA) content was determined by enzyme-linked immunosorbent assay (ELISA). MDA is an important indicator of plant stress physiology; higher MDA content indicates lower plant growth quality. One-way ANOVA showed highly significant differences in CHL and MDA between healthy and unhealthy, and between sub-healthy and unhealthy plants (P<0.01), indicating that CHL and MDA can be used as indicators of plant health quality.

[0024] Furthermore, step 5 is as follows:

[0025] Liquid water was extracted from collected plant and soil samples using a vacuum condensation extraction system, and hydrogen and oxygen isotopes were measured using a liquid water isotope analyzer. The proportion of water utilized by plants in each soil layer was analyzed using the Isource model. Since higher plant health levels correlate with better water utilization efficiency in deeper soil layers, the proportion of water absorbed by plants in the 40-100cm soil layer (WU) was used as a characterization indicator of plant health status. One-way ANOVA showed highly significant differences in WU between healthy and unhealthy, and between sub-healthy and unhealthy plants (P<0.01), indicating that WU can serve as a characterization indicator of plant health quality.

[0026] Furthermore, step 6 is as follows:

[0027] The sensitivity of plant growth quality to various factors was determined using a random forest model; the importance of each factor was as follows: WU (17.67) > MDA (7.80) > CHL (3.39) > N:P B (2.50)>TN B (1.86)>C:N B (1.57)>TN L (1.29)>TC L (1.25)>C:N L (0.49)>TP L (0.21)>TC B(0.20)>N:P L (0.19)>C:P B (0.13)>TP L (0.11)>C:P L (0.0018); Overall, WU, MDA, CHL, N:P B TN B C:N B TN L Most sensitive;

[0028] Factor effect analysis using the proportional advantage model yielded the following effect values: WU (0.8839***) > MDA (0.8260***) > CHL (0.6798***) > C:N B (0.5769***)>N:P B (0.5126***)>TN B (0.4721***)>TN L (0.3499**)>TC L (0.2847**)>C:N L (0.1399)>N:P L (0.0441)>TP L (0.0369)>TC B (0.0191)>C:P B (-0.0206)>TP L (-0.0356)>C:P L (-0.0671); Overall, WU, MDA, CHL, C:N B N:P B TN B Most sensitive.

[0029] Based on steps 2-5, the TN was obtained through differential analysis of various parameters under healthy, sub-healthy, and unhealthy conditions. L TN B C:N B N:P B CHL, MDA, and WU can be used as indicators of plant health quality;

[0030] The water absorption ratio, malondialdehyde, chlorophyll, and N:P were optimized using the three methods described above. B N:P B TN B Six parameters are used to characterize the health status of transplanted trees.

[0031] Further, step 7 is as follows: Through three machine learning models, namely the grid search method, the genetic algorithm, and the Bayesian optimization algorithm, through repeated iterative operations, the recognition thresholds of each factor under three health conditions of healthy, sub-healthy, and unhealthy are obtained; it is obtained that when WU > 0.54, the plant is in a healthy state, when 0.32 < WU ≤ 0.54, it is in a sub-healthy state, and when WU ≤ 0.32, it is in an unhealthy state, with a recognition accuracy rate of 100% and a verification accuracy rate of 80.00%; when MDA ≤ 4.26 μg / g, the plant is in a healthy state, when 4.26 μg / g < MDA ≤ 5.45 μg / g, it is in a sub-healthy state, and when MDA > 5.45 μg / g, it is in an unhealthy state, with a recognition accuracy rate of 86.67% and a verification accuracy rate of 83.33%; when CHL > 8.33 mg / g, the plant is in a healthy state, when 5.85 mg / g < CHL ≤ 8.33 mg / g, it is in a sub-healthy state, and when CHL ≤ 5.85 mg / g, it is in a dead state, with a recognition accuracy rate of 76.67% and a verification accuracy rate of 83.33%; when C:N B ≤ 41.41, the plant is in a healthy state, when 41.41 < C:N B ≤ 60.39, it is in a sub-healthy state, and when C:N B > 60.39, it is in a dead state, with a recognition accuracy rate of 70% and a verification accuracy rate of 83.33%, and a verification accuracy rate of 90%; for N:P B , the thresholds obtained through the grid search method are 4.75 and 4.24 respectively. When N:P B > 4.755, the plant is in a healthy state, when 4.245 < N:P B ≤ 4.755, it is in a sub-healthy state, and when N:P B ≤ 4.245, it is in an unhealthy state, with a recognition accuracy rate of 70% and a verification accuracy rate of 76.67%; when TN B > 10.96 g / kg, the plant is in a healthy state, when 9.56 g / kg < TN B ≤ 10.96 g / kg, it is in a sub-healthy state, and when TN B ≤ 9.56 g / kg, it is in an unhealthy state, with a recognition accuracy rate of 70% and a verification accuracy rate of 63.33%; Since chlorophyll can be directly measured in the field by a chlorophyll meter, it can be used as an indicator for rapid monitoring of the growth quality of arbors, and the thresholds are 8.33 mg / g and 5.85 mg / g respectively.

[0032] Further, step 8 is as follows: Based on the water absorption ratio, malondialdehyde, chlorophyll, N:P B , N:P B , TN BSix parameters are used to construct a health index (Formula 1 and Formula 2) for comprehensively evaluating the health status of plants. Based on three machine learning models: grid search method, genetic algorithm, and Bayesian optimization algorithm, the recognition threshold of the health index in the healthy state is obtained as a comprehensive evaluation index for the growth quality of plants. It is obtained that when HI > 0.4, the plant is in a healthy state; when -0.40 < CHL ≤ 0.4, the plant is in a sub-healthy state; when CHL ≤ -0.4, the plant is in an unhealthy state. The recognition accuracy rate is 100%, and the verification accuracy rate is 96.67%.

[0033]

[0034] In the formula, HI is the health index, is the index weight (calculated from the results of the Kruskal-Wallis test), Xi is the parameter value, μ i is the parameter mean, σ i is the standard deviation of the parameter, H is the statistic of the Kruskal-Wallis test, k is the number of groups, and n is the number of samples.

[0035] As can be seen from the above technical solutions, compared with the prior art, the beneficial effects of the present invention are as follows:

[0036] The present invention provides a method for identifying the growth quality of transplanted arbors in alpine regions to solve the problem that it is difficult to accurately identify the survival situation of transplanted arbors during the engineering construction process in alpine regions. Taking Picea crassifolia in alpine regions with three growth qualities of unhealthy, sub-healthy, and healthy transplanted for 1 - 2 years as the object, plant leaf samples and soil samples are collected, and the total carbon, total nitrogen, total phosphorus contents, C:N L , C:P L , N:P L in the leaves, and the total carbon, total nitrogen, total phosphorus, C:N B , C:P B , N:P BFifteen parameters, including chlorophyll content, malondialdehyde (MDA) content in branches, and the proportion of water absorbed by the plant from the 40-100cm soil layer, were used as indicators of tree health. Through random forest model, proportional dominance model, and factor variance analysis, six parameters sensitive to the health quality of trees were identified, and the discrimination thresholds for each factor on growth quality were determined. Based on these six parameters, a health index was constructed to comprehensively assess plant health and obtain a threshold for comprehensive evaluation of plant growth status, thus enabling precise monitoring of tree growth quality. This method, on the one hand, obtains sensitive factors and related discrimination thresholds that can characterize plant growth quality, enabling rapid monitoring of tree growth quality through the detection of single indicator parameters; on the other hand, it constructs a comprehensive health index to accurately identify the survival rate of alpine trees with long seedling establishment periods and delayed growth, thereby helping to accurately assess the effectiveness of tree transplantation and providing stronger guidance for engineering projects. Attached Figure Description

[0037] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0038] Figure 1 A threshold division diagram for water absorption ratio under different health conditions;

[0039] Figure 2 A threshold classification chart for malondialdehyde (MDA) under different health conditions;

[0040] Figure 3 A threshold division diagram for chlorophyll under different health conditions;

[0041] Figure 4 C:N under different health conditions B Threshold partitioning diagram;

[0042] Figure 5 N:P under different health conditions B Threshold partitioning diagram;

[0043] Figure 6 TN under different health conditions B Threshold partitioning diagram;

[0044] Figure 7 A threshold division chart for health indices under different health conditions. Detailed Implementation

[0045] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0046] Example 1

[0047] A method for identifying the growth quality of transplanted trees in high-altitude and cold regions, the method comprising the following steps:

[0048] Step 1: In the high-altitude and cold regions of the Qinghai-Tibet Plateau above 3200m, Qinghai spruce trees transplanted for 1-2 years were selected as the research object. Ten trees of each of the three categories of growth quality (healthy, sub-healthy, and unhealthy) were selected for a total of 30 trees for indicator screening and method construction. At the same time, 30 trees were selected to test the applicability of the identification method.

[0049] Obtain fresh leaves and branches, and collect soil samples from the 0-100cm soil layer at 50cm intervals from the base of the plant.

[0050] Further, step 1 is as follows:

[0051] In the high-altitude, cold regions of the Qinghai-Tibet Plateau (above 3200m), Qinghai spruce trees transplanted 1-2 years ago were used as the research subject. Through manual identification, 10 healthy, 10 sub-healthy, and 10 unhealthy trees were selected, totaling 30 trees, for indicator screening and identification method construction. An additional 30 trees were selected to test the applicability of the identification method. The selected trees ranged in height from 4.0 to 7.0m, diameter at breast height from 13.5 to 53.8cm, and age from 60 to 100 years.

[0052] During the summer when the plant is growing vigorously, fresh leaves and branches of Qinghai spruce are obtained by pruning shears and placed in kraft paper bags; soil samples from the 0-100cm soil layer are collected at a depth of 20cm from 50cm away from the base of the plant using a soil drill with a diameter of 5cm and placed in brown glass bottles.

[0053] Step 2: Determine the total carbon (TC) in tree leaves L ), total nitrogen (TN) L Total phosphorus (TP) L Content, calculate the leaf stoichiometric parameters C:N L C:P L N:P L The values ​​were analyzed using one-way ANOVA to determine the differences in various parameters under healthy, sub-healthy, and unhealthy conditions.

[0054] Further, step 2 is as follows:

[0055] Plant leaf samples were collected and blanched in an electrically heated drying oven (WGL-625B, Tester, China) at 65℃. The samples were then ground and crushed using a ball mill, and the total carbon (TC) was determined using an elemental analyzer. L ) and total nitrogen (TN) L The total phosphorus (TP) content was determined using a fully automated discrete chemical analyzer. L Content; calculation of leaf stoichiometry parameters C:N L C:P L N:P L Values; factor differences were analyzed under healthy, sub-healthy, and unhealthy conditions using one-way ANOVA. TN values ​​under healthy and sub-healthy conditions. L The differences were not significant (P>0.05), but highly significant (P<0.01) were observed between healthy and unhealthy individuals, and between those in a sub-healthy state and those in an unhealthy state. Other parameters showed no significant differences across different health conditions. This indicates that TN... L It can be used as a characterization indicator of plant health quality.

[0056] Step 3: Determine the total carbon (TC) of tree branches B ), total nitrogen (TN) B Total phosphorus (TP) B Content, calculate the stoichiometric parameters C:N of the branches. B C:P B N:P B The values ​​of these parameters were analyzed using one-way ANOVA to determine the differences between healthy, sub-healthy, and unhealthy conditions.

[0057] Furthermore, step 3 is as follows:

[0058] Plant branch samples were collected and blanched in an electrically heated drying oven at 65°C. They were then ground and crushed using a ball mill, and the total carbon (TC) was determined using an elemental analyzer. B ) and total nitrogen (TN) B The total phosphorus (TP) content was determined using a fully automated discrete chemical analyzer. B Content; calculation of stoichiometric parameters C:N in branches B C:P B N:P B The values ​​were analyzed; one-way ANOVA was used to analyze the differences in parameters under healthy, sub-healthy, and unhealthy conditions. TN values ​​under healthy and sub-healthy conditions were also analyzed. B C:N B N:P BThe differences were not significant (P>0.05), but highly significant differences were observed between healthy and deceased individuals, and between those in a sub-healthy and unhealthy state (P<0.01). Other parameters showed no significant differences across different health conditions. This indicates that TN... B C:N B N:P B It can be used as a characterization indicator of plant health quality.

[0059] Step 4: Measure the chlorophyll content (CHL) of plant leaves and the malondialdehyde content (MDA) of plant branches. The higher the MDA content, the lower the plant growth quality. Use one-way ANOVA to identify the differences in parameters under healthy, sub-healthy and unhealthy conditions.

[0060] Furthermore, step 4 is as follows:

[0061] Chlorophyll content (CHL) in plant leaves was determined by spectrophotometry, and malondialdehyde (MDA) content was determined by enzyme-linked immunosorbent assay (ELISA). MDA is an important indicator of plant stress physiology; higher MDA content indicates lower plant growth quality. The differences in various parameters under healthy, sub-healthy, and unhealthy conditions were identified. One-way ANOVA showed highly significant differences in CHL and MDA between healthy and unhealthy, and between sub-healthy and unhealthy conditions (P<0.01), indicating that CHL and MDA can serve as indicators of plant health quality.

[0062] Step 5: Determine the hydrogen and oxygen isotope concentrations of plant water and soil moisture, and analyze the proportion of water utilized by plants in different soil layers. The higher the plant health, the better the efficiency of water utilization in deeper soil layers. Therefore, the proportion of water absorbed by plants in the 40-100cm soil layer (WU) is used as an indicator of plant health status. One-way ANOVA is used to identify the differences in parameters under healthy, sub-healthy, and unhealthy conditions.

[0063] Furthermore, step 5 is as follows:

[0064] Liquid water was extracted from all plant and soil samples using a vacuum condensation extraction system (L1-2100EP). After filtration, hydrogen and oxygen isotopes were measured using a liquid water isotope analyzer (LGR DLT-100). The Isource model was used to analyze the proportion of water utilized by plants in each soil layer. Since higher plant health correlates with better water utilization efficiency in deeper soil layers, the proportion of water absorbed by plants in the 40-100cm soil layer (WU) was used as an indicator of plant health status. One-way ANOVA showed highly significant differences in WU between healthy and unhealthy, and between sub-healthy and unhealthy plants (P<0.01), indicating that WU can serve as an indicator of plant health quality.

[0065] Step 6: Using a random forest model and a proportional dominance model, combined with factor variance analysis, determine the sensitivity of plant growth quality to various factors. Plant health status affects WU, MDA, CHL, and C:N. B N:P B TC B The six factors showed the highest sensitivity.

[0066] Furthermore, step 6 is as follows:

[0067] Using a random forest model, the importance features of factors were analyzed to determine the sensitivity of plant growth quality to each factor. The importance of each factor was as follows: WU (17.67) > MDA (7.80) > CHL (3.39) > N:P B (2.50)>TN B (1.86)>C:N B (1.57)>TN L (1.29)>TC L (1.25)>C:N L (0.49)>TP L (0.21)>TC B (0.20)>N:P L (0.19)>C:P B (0.13)>TP L (0.11)>C:P L (0.0018). Overall, WU, MDA, MDA, N:P B TN B C:N B TN L Most sensitive.

[0068] Factor effect analysis using the proportional advantage model yielded the following results: WU (0.8839***) > MDA (0.8260***) > CHL (0.6798***) > C:N B (0.5769***)>N:P B (0.5126***)>TN B (0.4721***)>TN L (0.3499**)>TC L (0.2847**)>C:N L (0.1399)>N:P L (0.0441)>TP L (0.0369)>TC B (0.0191)>C:P B (-0.0206)>TP L (-0.0356)>C:PL (-0.0671). Generally, it is most sensitive to WU, MDA, CHL, C:N B , N:P B , TN B Most sensitive.

[0069] On the basis of completing Steps 2 - 5, through the differential analysis of each parameter under healthy and unhealthy, sub - healthy and unhealthy conditions, TN L , TN B , C:N B , N:P B , CHL, MDA, WU can be used as the characterization indicators of plant health quality.

[0070] Through the above three methods, the water absorption ratio (WU), malondialdehyde (MDA), chlorophyll (CHL), N:P B , N:P B , TN B 6 parameters are used to characterize the health quality of transplanted arbors;

[0071] Step 7: Through three types of machine learning models, namely grid search method, genetic algorithm, and Bayesian optimization algorithm, through repeated iterations, obtain the scientific discrimination thresholds for plant growth conditions; Since chlorophyll (CHL) can also be directly measured, chlorophyll can be used as an index for rapid monitoring of plant growth quality, and the thresholds are 8.33 μg / g and 5.85 μg / g respectively, and the recognition rate is above 70%.

[0072] Furthermore, in Step 7: Three machine learning models, namely grid search method, genetic algorithm, and Bayesian optimization algorithm, are used. Through repeated iterative operations, obtain the recognition thresholds of each factor under three health conditions of healthy, sub - healthy, and unhealthy;

[0073] For the water absorption ratio (WU) (see Figure 1 ), through the grid search method, the thresholds are 0.45 and 0.32 respectively, that is, when WU > 0.45, the plant is in a healthy state, when 0.32 < WU ≤ 0.45, it is in a sub - healthy state, and when WU ≤ 0.32, it is in an unhealthy state, and the recognition accuracy rate is 100%; Through the genetic algorithm, the thresholds are 0.54 and 0.32 respectively, and the recognition accuracy rate is 100%; Through the Bayesian optimization algorithm, the thresholds are 0.51 and 0.31 respectively, and the recognition accuracy rate is 96.67%;

[0074] The results of the verification group show that the recognition accuracies obtained by the three machine learning models of grid search method, genetic algorithm, and Bayesian optimization algorithm are 70.00%, 80.00%, and 73.33% respectively. Combining the results of the experimental group and the verification group, taking the thresholds obtained by the genetic algorithm as the standard, that is, the recognition thresholds of WU are 0.54 and 0.32.

[0075] For malondialdehyde (MDA) (see Figure 2 ), the thresholds obtained by the grid search method are 4.25 μg / g and 5.30 μg / g respectively. That is, when MDA ≤ 4.25 μg / g, the plant is in a healthy state; when 4.25 μg / g < MDA ≤ 5.30 μg / g, it is in a sub-healthy state; when MDA > 5.30 μg / g, it is in an unhealthy state, and the recognition accuracy rate is 86.67%; the thresholds obtained by the genetic algorithm are 4.26 μg / g and 5.45 μg / g respectively, and the recognition accuracy rate is 87.1%; the thresholds obtained by the Bayesian optimization algorithm are 4.26 μg / g and 5.45 μg / g respectively, and the recognition accuracy rate is 87.1%.

[0076] The results of the verification group show that the recognition accuracies obtained by the three machine learning models of the grid search method, genetic algorithm, and Bayesian optimization algorithm are 83.33%, 83.33%, and 83.33% respectively. Combining the results of the experimental group and the verification group, taking the values obtained by the genetic algorithm and the Bayesian optimization algorithm as the standard, that is, the recognition thresholds of MDA are 4.26 μg / g and 5.45 μg / g respectively.

[0077] For chlorophyll (CHL) (see Figure 3 ), the thresholds obtained by the grid search method are 7.11 mg / g and 5.81 mg / g respectively. That is, when CHL > 7.11 mg / g, the plant is in a healthy state; when 5.81 mg / g < CHL ≤ 7.11 mg / g, it is in a sub-healthy state; when CHL ≤ 5.81 mg / g, it is in a dead state, and the recognition accuracy rate is 76.67%; the thresholds obtained by the genetic algorithm are 5.85 μg / g and 8.33 μg / g respectively, and the recognition accuracy rate is 76.67%; the thresholds obtained by the Bayesian optimization algorithm are 5.87 μg / g and 7.43 μg / g respectively, and the recognition accuracy rate is 76.67%.

[0078] The results of the verification group show that the recognition accuracies obtained by the three machine learning models of the grid search method, genetic algorithm, and Bayesian optimization algorithm are 70.00%, 83.33%, and 76.67% respectively. Combining the results of the experimental group and the verification group, the genetic algorithm has a higher threshold recognition accuracy rate. Therefore, the thresholds of CHL are 8.33 μg / g and 5.85 μg / g respectively.

[0079] For C:N B (see Figure 4 ), the thresholds obtained by the grid search method are 39.5 and 41.5 respectively. That is, when C:N B ≤ 39.5, the plant is in a healthy state; when 39.5 < C:N B ≤ 41.5, it is in a sub-healthy state; when C:NB >41.5 indicates a dead state, with an accuracy rate of 70%; using a genetic algorithm, thresholds of 41.69 and 62.05 were obtained, with an accuracy rate of 66.67%; using a Bayesian optimization algorithm, thresholds of 41.41 and 60.39 were obtained, with an accuracy rate of 66.67%.

[0080] The results of the verification group show that the recognition accuracy obtained by the three machine learning models, namely grid search, genetic algorithm and Bayesian optimization algorithm, is 66.67%, 86.67% and 90.00%, respectively. Combining the results of the experimental group and the verification group, the result obtained by Bayesian optimization algorithm is taken as the standard, and the thresholds are 41.41 and 60.39, respectively.

[0081] For N:P B (See Figure 5 The thresholds obtained through grid search were 4.75 and 4.24, respectively, when N:P B >4.75, the plant is in a healthy state; when 4.24... <N:P B ≤4.75 indicates a sub-healthy state, when N:P B ≤4.24 indicates an unhealthy state, with an accuracy rate of 70%; thresholds obtained through genetic algorithm are 4.76 and 4.25, with an accuracy rate of 70%; thresholds obtained through Bayesian optimization algorithm are 4.78 and 4.33, with an accuracy rate of 70%.

[0082] The results of the verification group show that the recognition accuracy obtained by the three machine learning models, namely grid search, genetic algorithm and Bayesian optimization algorithm, is 76.67%, 76.67% and 70.00%, respectively. Combining the results of the experimental group and the verification group, the threshold recognition accuracy obtained by grid search and genetic algorithm is higher and the results are similar. The average of the two is taken as the threshold, which is 4.245 and 4.755, respectively.

[0083] For TN B (See Figure 6 The threshold values ​​obtained through grid search were 10.91 and 9.21 g / kg, respectively. When TN B >10.91g / kg, the plant is in a healthy state; when it is 9.21g / kg... <TN B ≤10.91g / kg indicates a sub-healthy state, when TN B ≤9.21g / kg indicates an unhealthy state, with an accuracy rate of 70%; the thresholds obtained through genetic algorithm are 9.93g / kg and 10.97g / kg, with an accuracy rate of 70%; the thresholds obtained through Bayesian optimization algorithm are 11.00g / kg and 9.54g / kg, with an accuracy rate of 70%.

[0084] The results of the verification group show that the recognition accuracies obtained by the three machine learning models, namely the grid search method, the genetic algorithm, and the Bayesian optimization algorithm, are 63.33%, 63.33%, and 63.33% respectively. Considering the results of the comprehensive test group and the verification group, the accuracies obtained by the three methods are the same. The average values obtained by the three methods are taken as the thresholds, which are 10.96 g / kg and 9.56 g / kg respectively.

[0085] In summary, through the three machine learning models, six relatively scientific recognition thresholds of parameters are obtained, and the genetic algorithm can effectively identify the thresholds of most parameters. Considering the water absorption ratio, malondialdehyde, N:P B 、N:P B 、TN B The results of 5 parameters are measured by indoor instruments, while chlorophyll can be measured on-site by a portable chlorophyll meter. The experimental recognition accuracy and verification recognition accuracy of the discrimination threshold are both above 70%, which can meet the requirements for the rapid discrimination of the growth quality of transplanted trees.

[0086] Step 8: Based on the above six parameters, construct a tree growth health index to comprehensively evaluate the health status of plants. Through three types of machine learning models, namely the grid search method, the genetic algorithm, and the Bayesian optimization algorithm, through repeated iteration, obtain the health index recognition thresholds of the plant growth status. The thresholds are 0.40 and -0.41 respectively. The recognition rate of the experimental group is 100%, and the recognition rate of the test group can reach more than 96.67%. When the health index < -0.40, the cell membrane system is severely damaged, the cell contents leak out, resulting in a decrease in chlorophyll content and mitochondrial content. The photosynthesis and respiration of plants are affected, which in turn affects the absorption of water and fertility by plants, and further affects the growth of plants, and may ultimately lead to the death of plants.

[0087] Furthermore, in the said Step 8: Based on six parameters of water absorption ratio, malondialdehyde, chlorophyll, N:P B 、N:P B 、TN B construct a tree growth health index to comprehensively evaluate the health quality status of plants; the thresholds obtained by the grid search method are 0.4 and -0.4 respectively. When HI > 0.4, the plant is in a healthy state. When -0.40 < HI ≤ 0.40, it is in a sub-healthy state. When HI ≤ -0.4, it is in a dead state, and the recognition accuracy is 100%; the thresholds obtained by the genetic algorithm are -0.38 and 0.50 respectively, and the recognition accuracy is 100%; the thresholds obtained by the Bayesian optimization algorithm are 0.40 and -0.41 respectively, and the recognition accuracy is 100%;

[0088] The results of the verification group show that the recognition accuracies obtained by the grid search method and the genetic algorithm are 96.67%, 86.67%, and 96.67%, respectively. Considering the results of both the experimental and verification groups, the accuracy obtained by the grid search method and the Bayesian optimization algorithm is similar, and their thresholds are also similar. The average of the two is taken as the recognition threshold, which is -0.40 and 0.41, respectively (see...). Figure 7 ).

[0089]

[0090] In the formula, HI represents the health index. Xi is the indicator weight (calculated using the results of the Kruskal-Wallis test), μ is the parameter value, and X is the parameter weight. i Let σ be the mean of the parameters. i denoted as the standard deviation of the parameter, H is the Kruskal-Wallis test statistic, k is the number of groups, and n is the sample size.

[0091] By comparing the effects of the above six factors on tree growth quality, it was found that WU (0.884) > MDA (0.826) > CHL (0.680) > C:N B (0.577)>N:P B (0.513)>TN B (0.472), with WU, MDA, and CHL having a significant impact on the tree health index; considering the physiological significance of the three parameters, it is indicated that when HI < -0.40, the cell membrane system is severely damaged, cell contents leak out, leading to a decrease in chlorophyll and mitochondrial content, affecting the plant's photosynthesis and respiration, thus impacting the plant's absorption of water and nutrients, consequently affecting plant growth, and ultimately potentially leading to plant death. C:N B N:P B N B All three parameters indicate that the growth quality of trees deteriorates when nitrogen is deficient, further illustrating that the ability of cells to absorb and utilize nitrogen is reduced under low HI conditions.

[0092] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for identifying the growth quality of transplanted trees in high-altitude and cold regions, characterized in that, The method includes the following steps: Step 1: In the high-altitude and cold regions above 3200m on the Qinghai-Tibet Plateau, Qinghai spruce that has been transplanted for 1-2 years was used as the research object. In summer, several trees of three health quality categories (unhealthy, sub-healthy, and healthy) were selected by manual identification. Fresh leaves and branches were obtained, and soil samples from the 0-100cm soil layer at 50cm from the base of the tree were collected. Step 2: Determine the total carbon (TC) in tree leaves. L ), total nitrogen (TN) L Total phosphorus (TP) L Content, calculate the leaf stoichiometric parameters C:N L C:P L N:P L The values ​​were determined, and the differences between the parameters under healthy, sub-healthy, and unhealthy conditions were identified. Step 3: Determine the total carbon (TC) of tree branches. B ), total nitrogen (TN) B Total phosphorus (TP) B Content, calculate the stoichiometric parameters C:N of the branches. B C:P B N:P B The values ​​were determined, and the differences between the parameters under healthy, sub-healthy, and unhealthy conditions were identified. Step 4: Measure the chlorophyll content (CHL) of plant leaves and the malondialdehyde (MDA) content of plant branches. Malondialdehyde is an important indicator of plant stress physiology. The lower the plant growth quality, the higher the MDA content. Also, determine the differences of each parameter under healthy, sub-healthy, and unhealthy conditions. Step 5: Measure the hydrogen and oxygen isotope concentrations of plant water and soil water, and analyze the proportion of water utilization by plants in different soil layers; the higher the health of the plant, the higher the proportion of water utilization in deeper soil layers. Therefore, the proportion of water absorbed by the plant in the 40-100cm soil layer (WU) is used as an indicator of health status. The higher the WU value, the better the health status. And identify the differences in parameters under healthy, sub-healthy and unhealthy conditions. Step 6: Using random forest and proportional dominance models, combined with factor variance analysis, determine the sensitivity of plant growth quality to various factors; and the impact of plant health status on WU, MDA, CHL, and C:N. B N:P B TC B The six parameters showed the highest sensitivity; Step 7: Using three types of machine learning models—grid search, genetic algorithm, and Bayesian optimization algorithm—to comprehensively identify the threshold for judging plant growth status, and to screen rapid indicators for identifying plant growth quality; Step 8: Based on the above 6 parameters, construct a health index to comprehensively assess the health status of plants. Use three types of machine learning models—grid search, genetic algorithm, and Bayesian optimization algorithm—to identify health index thresholds under different health conditions, which serve as the criteria for judging plant growth status.

2. The method for identifying the growth quality of transplanted trees in high-altitude and cold regions according to claim 1, characterized in that, Step 1 is as follows: In the high-altitude and cold regions of the Qinghai-Tibet Plateau above 3200m, Qinghai spruce trees transplanted for 1-2 years were selected as the research object. Ten trees of each of three categories of growth quality were selected: unhealthy, sub-healthy, and healthy. At the same time, another 30 trees were selected to test the applicability of the identification method. Fresh leaves and branches of Qinghai spruce were obtained by pruning shears, and soil samples were collected from the 0-100cm soil layer 50cm away from the base of the plant.

3. The method for identifying the growth quality of transplanted trees in high-altitude and cold regions according to claim 1, characterized in that, Step 2 is as follows: The collected plant leaf samples were blanched in an electrically heated drying oven at 65°C; the total carbon (TC) was determined using an elemental analyzer. L ) and total nitrogen (TN) L The total phosphorus (TP) content was determined using a fully automated discrete chemical analyzer. L Content; calculation of leaf stoichiometry parameters C:N L C:P L N:P L Values; the differences in parameters under healthy, sub-healthy, and unhealthy conditions were analyzed using one-way ANOVA; TN under healthy and sub-health conditions L The differences were not significant (P>0.05), but highly significant (P<0.01) were observed between healthy and unhealthy individuals and between those in a sub-healthy state and those in an unhealthy state. Other parameters showed no significant differences. Explanation of TN L It can be used as a characterization indicator of plant health quality.

4. The method for identifying the growth quality of transplanted trees in high-altitude and cold regions according to claim 1, characterized in that, Step 3 is as follows: The collected plant branch samples were blanched in an electrically heated drying oven at 65°C; after grinding using a ball mill, the total carbon (TC) was determined using an elemental analyzer. B ) and total nitrogen (TN) B The total phosphorus (TP) content was determined using a fully automated discrete chemical analyzer. B Content; calculation of stoichiometric parameters C:N in branches B C:P B N:P B Values; Factor differences under healthy, sub-healthy, and unhealthy conditions were analyzed using one-way ANOVA; TN under healthy and sub-health conditions B C:N B N:P B The differences were not significant (P>0.05), but there were extremely significant differences between healthy and unhealthy, and between sub-healthy and unhealthy (P<0.01). The differences in other parameters were not significant. Explanation of TN B C:N B N:P B It can be used as a characterization indicator of plant health quality.

5. The method for identifying the growth quality of transplanted trees in high-altitude and cold regions according to claim 1, characterized in that, Step 4 is as follows: Chlorophyll content (CHL) in plant leaves was determined by spectrophotometry, and malondialdehyde (MDA) content was determined by enzyme-linked immunosorbent assay (ELISA). MDA is an important indicator of plant stress physiology; higher MDA content indicates lower plant growth quality. One-way ANOVA showed highly significant differences in CHL and MDA between healthy and unhealthy, and between sub-healthy and unhealthy plants (P<0.01), indicating that CHL and MDA can be used as indicators of plant health quality.

6. The method for identifying the growth quality of transplanted trees in high-altitude and cold regions according to claim 1, characterized in that, Step 5 is as follows: Liquid water was extracted from collected plant and soil samples using a vacuum condensation extraction system, and hydrogen and oxygen isotopes were measured using a liquid water isotope analyzer. The proportion of water utilized by plants in each soil layer was analyzed using the Isource model. Since higher plant health levels correlate with better water utilization efficiency in deeper soil layers, the proportion of water absorbed by plants in the 40-100cm soil layer (WU) was used as a characterization indicator of plant health status. One-way ANOVA showed highly significant differences in WU between healthy and unhealthy, and between sub-healthy and unhealthy plants (P<0.01), indicating that WU can serve as a characterization indicator of plant health quality.

7. The method for identifying the growth quality of transplanted trees in high-altitude and cold regions according to any one of claims 3-6, characterized in that, Step 6 is as follows: The sensitivity of plant growth quality to various factors was determined using a random forest model; the importance of each factor was as follows: WU (17.67) > MDA (7.80) > CHL (3.39) > N:P B (2.50)>TN B (1.86)>C:N B (1.57)>TN L (1.29)>TC L (1.25)>C:N L (0.49)>TP L (0.21)>TC B (0.20)>N:P L (0.19)>C:P B (0.13)>TP L (0.11)>C:P L (0.0018); Overall, WU, MDA, CHL, N:P B TN B C:N B TN L Most sensitive; Factor effect analysis using the proportional advantage model yielded the following effect values: WU (0.8839***) > MDA (0.8260***) > CHL (0.6798***) > C:N B (0.5769***)>N:P B (0.5126***)>TN B (0.4721***)>TN L (0.3499**)>TC L (0.2847**)>C:N L (0.1399)>N:P L (0.0441)>TP L (0.0369)>TC B (0.0191)>C:P B (-0.0206)>TP L (-0.0356)>C:P L (-0.0671); Overall, WU, MDA, CHL, C:N B N:P B TN B Most sensitive; Based on steps 2-5, the TN was obtained through differential analysis of various parameters under healthy, sub-healthy, and unhealthy conditions. L TN B C:N B N:P B CHL, MDA, and WU can be used as indicators of plant health quality; After comparing the three methods mentioned above, the final selection was based on water absorption ratio (WU), malondialdehyde content (MDA), chlorophyll content (CHL), and N:P ratio. B N:P B TN B A total of 6 parameters are used to characterize the health status of transplanted trees.

8. The method for identifying the growth quality of transplanted trees in high-altitude and cold regions according to claim 1, characterized in that, Step 7 is as follows: Through three machine learning models, namely the grid search method, genetic algorithm, and Bayesian optimization algorithm, through repeated iterative operations, the recognition thresholds of each factor under three health conditions of healthy, sub-healthy, and unhealthy are obtained; it is obtained that when WU > 0.54, the plant is in a healthy state, when 0.32 < WU ≤ 0.54, it is in a sub-healthy state, and when WU ≤ 0.32, it is in an unhealthy state, with a recognition accuracy of 100% and a verification accuracy of 80.00%; when MDA ≤ 4.26 μg / g, the plant is in a healthy state, when 4.26 μg / g < MDA ≤ 5.45 μg / g, it is in a sub-healthy state, and when MDA > 5.45 μg / g, it is in an unhealthy state, with a recognition accuracy of 86.67% and a verification accuracy of 83.33%; when CHL > 8.33 mg / g, the plant is in a healthy state, when 5.85 mg / g < CHL ≤ 8.33 mg / g, it is in a sub-healthy state, and when CHL ≤ 5.85 mg / g, it is in a dead state, with a recognition accuracy of 76.67% and a verification accuracy of 83.33%; when C:N B ≤ 41.41, the plant is in a healthy state, when 41.41 < C:N B ≤ 60.39, it is in a sub-healthy state, and when C:N B > 60.39, it is in a dead state, with a recognition accuracy of 70% and a verification accuracy of 83.33%, and a verification accuracy of 90%; for N:P B , the thresholds obtained by the grid search method are 4.75 and 4.24 respectively. When N:P B > 4.755, the plant is in a healthy state, when 4.245 < N:P B ≤ 4.755, it is in a sub-healthy state, and when N:P B ≤ 4.245, it is in an unhealthy state, with a recognition accuracy of 70% and a verification accuracy of 76.67%; when TN B > 10.96 g / kg, the plant is in a healthy state, when 9.56 g / kg < TN B ≤ 10.96 g / kg, it is in a sub-healthy state, and when TN B ≤ 9.56 g / kg, it is in an unhealthy state, with a recognition accuracy of 70% and a verification accuracy of 63.33%; Since chlorophyll can be directly measured in the field by a chlorophyll meter, it can be used as an index for rapid monitoring of the growth quality of arbors, and the thresholds are 8.33 mg / g and 5.85 mg / g respectively.

9. The method for identifying the growth quality of transplanted trees in high-altitude and cold regions according to claim 1, characterized in that, Step 8 is as follows: Based on six parameters including water absorption ratio, malondialdehyde, chlorophyll, N:P B , N:P B , TN B A health index (Formulas 1 and 2) is constructed to comprehensively evaluate the health status of plants. Based on three machine learning models, namely the grid search method, genetic algorithm, and Bayesian optimization algorithm, the recognition threshold of the health index under healthy conditions is obtained as a comprehensive evaluation index for plant growth quality. It is obtained that when HI > 0.4, the plant is in a healthy state; when -0.40 < CHL ≤ 0.4, the plant is in a sub-healthy state; when CHL ≤ -0.4, the plant is in an unhealthy state, with a recognition accuracy of 100% and a verification accuracy of 96.67%. In Formulas 1 and 2, HI represents the health index. Xi is the indicator weight (calculated using the results of the Kruskal-Wallis test), μ is the parameter value, and X is the parameter weight. i Let σ be the mean of the parameters. i denoted as the standard deviation of the parameter, H is the Kruskal-Wallis test statistic, k is the number of groups, and n is the sample size.

Images