High-oil and high-yield soybean breeding method and system based on remote sensing assistance of unmanned aerial vehicle

Abstract

The invention discloses a high-oil and high-yield soybean breeding method and system based on remote sensing assistance of an unmanned aerial vehicle, and the method comprises the following steps: selecting high-oil and high-yield germplasm and a local main cultivated variety as parents, and carrying out large-scale hybridization; estimating the soybean yield by using the hyperspectral remote sensing image reflectivity of the unmanned aerial vehicle; wherein an empirical regression analysis algorithm is configured in an unmanned aerial vehicle remote sensing algorithm, and based on unmanned aerial vehicle hyperspectral remote sensing data of different time sequences, the accuracy of a soybean material yield estimation model and the accuracy of mutual verification are compared; and screening sensitive vegetation indexes, optimal growth periods, optimal growth period combinations and material types with relatively good determination coefficients of the yield estimation models, and performing verification, comparison and analysis on the effects of different yield estimation models in practical application in different levels to generate the yield estimation model for assisting breeding.

Description

Technical Field

[0001] The present application relates to the field of breeding technology, and in particular to a high-oil and high-yield soybean breeding method and system based on unmanned aerial vehicle remote sensing assistance. Background Art

[0002] At present, domestic soybeans have basically met the demand for food. To further develop soybean production, high yield is the prerequisite. The key is to meet the demand for oil for pressing and take into account the demand for soybean meal for feed. At present, soybean breeding is still mainly based on traditional breeding technology, which lags behind the world's advanced level in breeding efficiency and precise improvement of traits. The degree of automation, engineering and intelligence of breeding is very low.

[0003] With the development of modern remote sensing technology applications in agriculture, the introduction of remote sensing technology to assist in the cultivation and creation of new high-yield, density-resistant, high-oil, and multi-resistant soybean varieties is of great significance for promoting the rapid increase in my country's soybean yield level, ensuring the security of my country's soybean supply, and improving the competitiveness of domestic soybeans. Summary of the invention

[0004] The present application provides a high-oil and high-yield soybean breeding method and system based on UAV remote sensing assistance to solve the above problems.

[0005] On the one hand, the present application provides a high-oil and high-yield soybean breeding method assisted by unmanned aerial vehicle remote sensing, the method comprising the following steps: selecting high-oil and high-yield germplasm and local main cultivated varieties as parents, and implementing large-scale hybridization; estimating soybean yield by using the reflectivity of unmanned aerial vehicle hyperspectral remote sensing images; wherein, the unmanned aerial vehicle remote sensing algorithm is equipped with an empirical regression analysis algorithm, based on unmanned aerial vehicle hyperspectral remote sensing data of different time series, by comparing the accuracy of soybean material yield estimation models and the accuracy of mutual verification, the sensitive vegetation index, optimal growth period and optimal growth period combination and material type with better determination coefficient of the yield estimation model are screened, and the effects of different yield estimation models in practical applications are verified and compared at different levels to generate a yield estimation model for assisted breeding.

[0006] In one implementation of the present application, the selection of parents is specifically as follows: selecting varieties with better salt tolerance indexes during the germination and seedling stages as parents for preparing hybrid combinations.

[0007] In one implementation of the present application, hybridization specifically includes:

[0008] Select plants that are growing well and strong for hybridization;

[0009] Select the flower buds in the upper middle part of the mother plant, where the inflorescence has not yet completely extended from the calyx, but the color of the corolla can be seen, and remove all the flowers and young buds around it. Use tweezers to remove the sepals of the buds that are about to bloom, and use the thumb and index finger of the left hand to gently pinch the pedicel and the base of the flower. Use tweezers in the right hand to first remove the two sepals under the keel petal, then pinch the corolla part and gently lift it along the direction of the petals, remove the corolla and all anthers, and only keep the complete stigma;

[0010] When the male parent begins to shed pollen, remove the sepals and petals, gently apply the anthers to the stigma of the emasculated female parent, and mark the pedicel;

[0011] Hang labels on the branches corresponding to the hybrid flowers, indicate the names of the parents of the hybrid combination, the pollination date, and keep records;

[0012] Check the survival rate one week after hybridization and remove the newly grown flower buds around the successfully hybridized pods.

[0013] In one implementation of the present application, the method further includes:

[0014] After the hybrid pods mature, they are harvested in time, stored in mesh bags together with labels according to the combination, placed in a drying room to dry naturally, and then placed in seed bags for safekeeping;

[0015] Record the names of the parents, the name of the hybridizer, the hybridization date, the harvest date, the number of harvested pods and the number of harvested grains in detail, and keep an electronic file.

[0016] In one implementation of the present application, the method further includes: selecting each generation, specifically:

[0017] Sow all the harvested hybrid seeds with a row spacing of 0.5m and a plant spacing of 0.2m;

[0018] Investigate and record the growth, uniformity of growth, disease resistance and seed yield of the F1 generation plants, and preliminarily remove false hybrids;

[0019] Screen F2 generation strains with excellent comprehensive field traits and high heritability of yield and quality for self-pollination and propagation; sow the selected high-oil and excellent F2 generation single plant seeds in rows at equal intervals to form F3 generation plant rows; select F3 generation single plants or plant rows with uniform and outstanding traits for self-pollination and propagation; sow the selected F3 generation single plant seeds in rows at equal intervals to form F4 generation plant rows; select F4 and later generations with uniform and outstanding traits and plant rows for mixed harvesting, test them indoors at the same time, use near-infrared analyzer to screen and retain high-oil offspring strains; conduct yield identification of the retained excellent strains in plots of different areas, and for yield identification, introduce unmanned aerial vehicle remote sensing technology to quickly screen new high-yield and high-oil soybean varieties.

[0020] The present application also provides a high-oil and high-yield soybean breeding system based on UAV remote sensing assistance, the system comprising: a high-resolution hyperspectral image module installed on the UAV, used to analyze phenotypic information such as crop yield and other agronomic traits, and at the same time found that the best growth period for establishing a soybean yield estimation model is the beginning of soybean grain filling.

[0021] The present application provides a high-oil and high-yield soybean breeding method and system based on unmanned aerial vehicle remote sensing assistance. Soybean varieties from different sources are used as research objects. Based on unmanned aerial vehicle hyperspectral remote sensing data of different time series, the accuracy of soybean material yield estimation models and the accuracy of mutual verification are compared to screen sensitive vegetation indices, optimal growth periods and optimal growth period combinations, and material types with better determination coefficients of yield estimation models. Verification and comparative analysis of the effects of different yield estimation models in practical applications are carried out at different levels, in order to find a yield estimation model that can assist breeding, reduce the workload of manual yield counting in the field, and improve the efficiency of soybean breeding. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 A flow chart of a high-oil and high-yield soybean breeding method based on drone remote sensing assistance provided in an embodiment of the present application. DETAILED DESCRIPTION

[0023] In order to make the purpose, technical solution and advantages of the present application clearer, the technical solution of the present application will be clearly and completely described below in combination with the specific embodiments of the present application and the corresponding drawings. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application.

[0024] The present application embodiment provides a method and system for breeding high-oil and high-yield soybeans based on remote sensing assistance of unmanned aerial vehicles, such as Figure 1 As shown in the figure, the method mainly includes the following steps: S1: select high-oil and high-yield germplasm and local main cultivated varieties as parents, and implement large-scale hybridization; S2: use the reflectance of UAV hyperspectral remote sensing images to estimate soybean yield; wherein, the algorithm of UAV remote sensing is equipped with an empirical regression analysis algorithm, which is based on UAV hyperspectral remote sensing data of different time series, and by comparing the accuracy of soybean material yield estimation models and the accuracy of mutual verification, screens sensitive vegetation indices, optimal growth periods and optimal growth period combinations, and material types with better determination coefficients of yield estimation models, and conducts verification and comparative analysis of the effects of different yield estimation models in practical applications at different levels, so as to generate a yield estimation model for assisted breeding.

[0025] In one implementation of the present application, the selection of parents is specifically: selecting varieties with better salt tolerance indicators during the germination and seedling stages as parents to prepare hybrid combinations.

[0026] In one implementation of the present application, hybridization specifically includes:

[0027] Select plants that are growing well and strong for hybridization;

[0028] Select the flower buds in the upper middle part of the mother plant, where the inflorescence has not yet completely extended from the calyx, but the color of the corolla can be seen, and remove all the flowers and young buds around it. Use tweezers to remove the sepals of the buds that are about to bloom, and use the thumb and index finger of the left hand to gently pinch the pedicel and the base of the flower. Use tweezers in the right hand to first remove the two sepals under the keel petal, then pinch the corolla part and gently lift it along the direction of the petals, remove the corolla and all anthers, and only keep the complete stigma;

[0029] When the male parent begins to shed pollen, remove the sepals and petals, gently apply the anthers to the stigma of the emasculated female parent, and mark the pedicel;

[0030] Hang labels on the branches corresponding to the hybrid flowers, indicate the names of the parents of the hybrid combination, the pollination date, and keep records;

[0031] Check the survival rate one week after hybridization and remove the newly grown flower buds around the successfully hybridized pods.

[0032] In one implementation of the present application, the method further includes:

[0033] After the hybrid pods mature, they are harvested in time, stored in mesh bags together with labels according to the combination, placed in a drying room to dry naturally, and then placed in seed bags for safekeeping;

[0034] Record the names of the parents, the name of the hybridizer, the hybridization date, the harvest date, the number of harvested pods and the number of harvested grains in detail, and keep an electronic file.

[0035] In one implementation of the present application, the method further includes: selecting each generation, specifically:

[0036] Sow all the harvested hybrid seeds with a row spacing of 0.5m and a plant spacing of 0.2m;

[0037] Investigate and record the growth, uniformity of growth, disease resistance and seed yield of the F1 generation plants, and preliminarily remove false hybrids;

[0038] Screen F2 generation strains with excellent comprehensive field traits and high heritability of yield and quality for self-pollination and propagation; sow the selected high-oil and excellent F2 generation single plant seeds in rows at equal intervals to form F3 generation plant rows; select F3 generation single plants or plant rows with uniform and outstanding traits for self-pollination and propagation; sow the selected F3 generation single plant seeds in rows at equal intervals to form F4 generation plant rows; select F4 and later generations with uniform and outstanding traits and plant rows for mixed harvesting, test them indoors at the same time, use near-infrared analyzer to screen and retain high-oil offspring strains; conduct yield identification of the retained excellent strains in plots of different areas, and for yield identification, introduce unmanned aerial vehicle remote sensing technology to quickly screen new high-yield and high-oil soybean varieties.

[0039] The present application also provides a high-oil and high-yield soybean breeding system based on UAV remote sensing assistance, the system comprising: a high-resolution hyperspectral image module installed on the UAV, used to analyze phenotypic information such as crop yield and other agronomic traits, and at the same time found that the best growth period for establishing a soybean yield estimation model is the beginning of soybean grain filling.

[0040] There are two main methods for estimating soybean yield using the reflectance of UAV hyperspectral remote sensing images: physics-based models and empirical regression analysis. Physics-based models are based on physical principles and require canopy growth physical parameters, soil parameters and some external parameters to simulate canopy reflectance, but these models are often not easy to obtain. In contrast, empirical regression techniques can provide a direct relationship between spectral features and vegetation parameters. Previous studies have used many effective empirical regression techniques, and the study of spectral indices has made full use of narrow bands on the hyperspectral spectrum and even some different types of sensor data. The main purpose of the model construction method study is to find a method for building a yield estimation model with higher accuracy, and to apply remote sensing technology to the large-scale crop breeding process more efficiently and conveniently. Based on the hyperspectral data of UAVs from different periods, a simple and easy-to-operate yield prediction model is obtained using a large number of breeding materials as research samples.

[0041] The present application example proves that the high-resolution hyperspectral image installed on the drone can analyze the phenotypic information such as crop yield and other agronomic traits. At the same time, it is found that the best growth period for establishing a soybean yield estimation model is the soybean grain filling stage (R5). In the R5 growth period, the sensitive bands screened by different varieties in the 2015-2016 strain identification and strain comparison test are distributed between 478-638nm and 570-922nm, proving that the infrared and near-infrared regions on the hyperspectral band are sensitive areas of crop canopy reflectivity. The present application example finds that the top two vegetation indices in the soybean R2 to R6 stages using the spectral index ranking method of the R-square ranking method are both RVI and NDVI. Improving the accuracy and adaptability of the yield estimation model is the premise of applying drone remote sensing technology. The overall research goal of the present application example is to develop a better model for estimating soybean yield based on canopy spectral information, reduce the workload of manual harvesting, save breeding costs, and improve soybean breeding efficiency. Taking soybean varieties from different sources as the research objects, by comparing the accuracy of soybean material yield estimation models, the sensitive vegetation index and the optimal growth period of different test materials for yield estimation are determined. This chapter mainly compares the model accuracy, and at the same time selects materials with better yield estimation models through mutual verification.

[0042] Based on the analysis of sensitive bands and maximum determination coefficients of breeding materials in the R5 growth period, models based on the sensitive bands NDVI (514, 606) and RVI (514, 606) were constructed. At the same time, the breeding materials were evenly divided into two groups, A and B, with 266 materials in each group. The sensitive bands and maximum determination coefficients of the two groups of materials were analyzed based on the R5 period. The yield estimation models based on the sensitive bands NDVI (514, 606) and RVI (514, 606) were constructed using the same method. The results of the three models are summarized in Table 1.

[0043] Table 1 Comparison of high spectral sensitivity bands and model determination coefficients based on experimental strains at R5 stage

[0044]

[0045] Where, P: p-value of the model significance test.

[0046] The results showed that the sensitive bands of the three groups of materials were consistent. In 2015, the maximum determination coefficient of group A materials was the largest, and the maximum determination coefficient of group B materials was the smallest. The yield estimation model based on NDVI (638, 674) and RVI (638, 674) had the largest average R2 value for modeling and verification of group A materials, and the smallest for group B materials.

[0047] According to the yield standards of line elimination, retention, and upgrading, we compared and classified them one by one, and used the model established for group A materials to verify the yield of group B materials. In the same way, we used the model established for group B materials to verify the yield of group A materials. We established a model based on all varieties to predict the yield of this test material. The results are summarized in Table 2.

[0048] Table 2 Mutual validation of hyperspectral yield models based on experimental lines at R5 stage

[0049]

[0050] Among them, Eli: indicates that the output is less than 3.00t ha -1 Elimination type materials, Pro: indicates more than 3.75t ha -1 Upgrade type material, Res: indicates between 3.00t ha -1 and 3.75t ha -1 The retention type material between; M A 、M B 、M A+B The models are constructed based on three groups of materials: A, B, and A+B.

[0051] The results showed that the model constructed by group B materials predicted group A materials, and the screening accuracy of the type materials of group A below 3.00t ha-1 was the highest, reaching 89.81%, and the screening accuracy of the type materials of group A above 3.750t ha-1 was the lowest, only 22.22%. The overall accuracy of the model constructed by group A materials in screening group B materials was high, at 61.65%, and the overall accuracy of the model constructed by all materials was the highest, at 65.04%. Compared with the models constructed by group A and group B materials, the actual application effect of the model constructed based on all breeding materials was slightly better. The use of this model can assist in the rapid screening of new soybean varieties with both high yield and high oil content.

[0052] Each embodiment in this application is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0053] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0054] The above is only an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. A high-oil and high-yield soybean breeding method based on unmanned aerial vehicle remote sensing assistance, characterized in that: The method comprises the following steps: selecting high-oil and high-yield germplasm and local main cultivated varieties as parents, and implementing large-scale hybridization; estimating soybean yield by using the reflectance of unmanned aerial vehicle (UAV) hyperspectral remote sensing images; wherein the algorithm of the UAV remote sensing is equipped with an empirical regression analysis algorithm, and the accuracy of the soybean material yield estimation model and the accuracy of mutual verification are compared by using UAV hyperspectral remote sensing data based on different time series, and sensitive vegetation indexes, optimal growth periods and optimal growth period combinations, and material types with better determination coefficients of the yield estimation models are screened, and verification and comparative analysis of the effects of different yield estimation models in practical applications are carried out at different levels to generate a yield estimation model for auxiliary breeding.

2. The method for breeding high-oil and high-yield soybeans based on remote sensing assisted by unmanned aerial vehicles according to claim 1, characterized in that: The selection of parents is specifically as follows: selecting varieties whose salt tolerance indexes at the germination and seedling stages meet the preset requirements as parents for preparing hybrid combinations.

3. The high-oil and high-yield soybean breeding method based on drone remote sensing assistance according to claim 1 is characterized in that: Hybridization specifically includes: Select plants that are growing well and strong for hybridization; Select the flower buds in the upper middle part of the mother plant, where the inflorescence has not yet completely extended from the calyx, but the color of the corolla can be seen, and remove all the flowers and young buds around it. Use tweezers to remove the sepals of the buds that are about to bloom, and use the thumb and index finger of the left hand to gently pinch the pedicel and the base of the flower. Use tweezers in the right hand to first remove the two sepals under the keel petal, then pinch the corolla part and gently lift it along the direction of the petals, remove the corolla and all anthers, and only keep the complete stigma; When the male parent begins to shed pollen, remove the sepals and petals, gently apply the anthers to the stigma of the emasculated female parent, and mark the pedicel; Hang labels on the branches corresponding to the hybrid flowers, indicate the names of the parents of the hybrid combination, the pollination date, and keep records; Check the survival rate one week after hybridization and remove the newly grown flower buds around the successfully hybridized pods.

4. The high-oil and high-yield soybean breeding method based on drone remote sensing assistance according to claim 3 is characterized in that: The method further comprises: After the hybrid pods mature, they are harvested in time, stored in mesh bags together with labels according to the combination, placed in a drying room to dry naturally, and then placed in seed bags for safekeeping; Record the names of the parents, the name of the hybridizer, the hybridization date, the harvest date, the number of harvested pods and the number of harvested grains in detail, and keep an electronic file.

5. The method for breeding high-oil and high-yield soybeans based on drone remote sensing assistance according to claim 1, characterized in that: The method further comprises: selection of each generation, specifically: Sow all the harvested hybrid seeds with a row spacing of 0.5m and a plant spacing of 0.2m; Investigate and record the growth, uniformity of growth, disease resistance and seed yield of the F1 generation plants, and preliminarily remove false hybrids; Screen F2 generation strains with excellent comprehensive field traits and high heritability of yield and quality for self-pollination and propagation; sow the selected high-oil and excellent F2 generation single plant seeds in rows at equal intervals to form F3 generation plant rows; select F3 generation single plants or plant rows with uniform and outstanding traits for self-pollination and propagation; sow the selected F3 generation single plant seeds in rows at equal intervals to form F4 generation plant rows; select F4 and later generations with uniform and outstanding traits and plant rows for mixed harvesting, test them indoors at the same time, use near-infrared analyzer to screen and retain high-oil offspring strains; conduct yield identification of the retained excellent strains in plots of different areas, and for yield identification, introduce unmanned aerial vehicle remote sensing technology to quickly screen new high-yield and high-oil soybean varieties.

6. A high-oil and high-yield soybean breeding system based on UAV remote sensing assistance, characterized in that: The system includes: a high-resolution hyperspectral image module installed on a drone, which is used to analyze phenotypic information such as crop yield and other agronomic traits, and it is found that the best growth period for establishing a soybean yield estimation model is the beginning of soybean grain filling.

Images