Method and program for estimating irrigation conditions, and method for controlling the cultivation environment.

Abstract

Appropriate fluid supply conditions can be estimated in a short time. [Solution] The method for estimating irrigation conditions includes: a first step of generating a model showing the relationship between the irrigation conditions and the rate of damaged fruit in the crop, using a first relationship between the irrigation conditions when cultivating the crop and the expression level of genes related to the rate of damage in the crop, and a second relationship between the expression level of genes related to the rate of damage in the crop and the rate of damaged fruit in the crop; and a second step of using the model to estimate the irrigation conditions necessary to cultivate the crop so that it achieves a desired rate of damaged fruit.

Description

Technical Field

[0001] The present invention relates to a liquid supply condition estimation method, a program, and a cultivation environment control method.

Background Art

[0002] Conventionally, in agriculture, improvements in techniques according to varieties and geographical conditions have been attempted through trial and error in repeated cultivation. For example, when improving the cultivation method of tomatoes or the like, it is common to try and error to find a suitable cultivation method by comparing a new technique with the conventional method through cultivation tests. Therefore, when starting cultivation in a new area or introducing a new cultivation method or variety, it is necessary to conduct a number of cultivation tests and try and error to establish an appropriate cultivation method, which requires a long time.

[0003] Also, for example, in order to hydroponically cultivate high-value-added plants such as high-sugar tomatoes, it is known that it is effective to appropriately adjust the liquid supply amount of the nutrient solution to give moderate water stress to the crops.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Patent Document 2

Summary of the Invention

Problems to be Solved by the Invention

[0005] However, since water stress varies depending on the evapotranspiration amount and the change in the moisture of the medium, producers need to adjust the timing and amount of liquid supply while observing the wilting of the crops at any time, which is a skill-required operation. Also, if the liquid supply amount is limited too much in high-sugar tomato cultivation, damage effects are likely to occur. Therefore, it is necessary to construct a liquid supply method so that the incidence rate of damage effects is suppressed as much as possible.

[0006] The present invention aims to provide a method and program for estimating appropriate irrigation conditions in a short time, as well as a method for controlling the cultivation environment to an appropriate level. [Means for solving the problem]

[0007] The present invention provides a method for estimating water supply conditions, comprising: a first step of generating a model showing the relationship between the water supply conditions and the rate of damaged fruit occurrence in the crop, using a first relationship between the water supply conditions when cultivating the crop and the expression level of genes related to the incidence of damage in the crop, and a second relationship between the expression level of genes related to the incidence of damage in the crop and the rate of damaged fruit occurrence in the crop; and a second step of using the model to estimate the water supply conditions necessary to cultivate the crop so that it achieves a desired rate of damaged fruit occurrence. [Effects of the Invention]

[0008] The irrigation condition estimation method and program of the present invention have the effect of being able to estimate appropriate irrigation conditions in a short time. Furthermore, the cultivation environment control method of the present invention has the effect of being able to control the cultivation environment appropriately. [Brief explanation of the drawing]

[0009] [Figure 1] Figure 1(a) is a graph showing the daily changes in the amount of liquid supplied (ml / plant) under the "high-very low" and "low-low" conditions, and Figure 1(b) is a graph showing the effect of the liquid supply conditions on the incidence of defective fruit in the first flower cluster. [Figure 2] Figures 2(a) to 2(c) show the average expression levels of chitinase genes involved in cell wall degradation in response to ethylene-jasmonic acid, average expression levels of fruit weight-related genes, and average expression levels of fruit development-related genes in undeveloped fruits sampled at 42, 55, 69, and 84 days after transplanting. [Figure 3] Figure 3 shows the expression ratio of chitinase genes and fruit weight-correlated genes as a function of the number of days after transplanting. [Figure 4] Figure 4 is a graph showing the gene expression of chitinase genes and fruit weight-correlated genes (10 genes) in underdeveloped fruits and the subsequent incidence of damaged fruit. [Figure 5] Figure 5(a) shows a simulation model (first model) for estimating the incidence of damaged fruit in cultivated crops based on cultivation conditions, and Figure 5(b) shows a simulation model (second model) for estimating the sugar content of cultivated crops based on cultivation conditions. [Figure 6] Figure 6 shows the configuration of an automatic nutrient solution supply system used for hydroponic cultivation of tomatoes. [Figure 7] Figure 7 is a flowchart showing the automatic nutrient solution control and hydroponic cultivation method using a controller. [Figure 8] Figure 8 is a graph showing the timing and amount of liquid supply in the process shown in Figure 7. [Figure 9] Figure 9 shows the hardware configuration of the server shown in Figure 6. [Figure 10] Figure 10 is a functional block diagram of the server. [Figure 11] Figure 11 is a graph showing the relationship between the measured maximum weight (Wm) and the estimated maximum weight (estimated Wm) calculated using a model based on plant height and stem diameter. [Figure 12] Figure 12(a) is a graph showing the relationship between maximum weight Wm and stem diameter, and Figure 12(b) is a graph showing the relationship between maximum weight Wm and plant height. [Figure 13] Figure 13 shows an example of a decision tree model. [Figure 14] Figure 14 is a flowchart showing the server's processing. [Figure 15] Figure 15 is a graph showing the average sugar content of cultivated tomatoes, comparing the case where the threshold was determined using the decision tree model in Figure 14 (automatic) and the case where a skilled grower manually changed the threshold every day (manual). [Modes for carrying out the invention]

[0010] Hereinafter, an embodiment will be described in detail based on FIGS. 1 to 15. This embodiment aims to reduce the incidence of defective fruits during the cultivation of high-sugar tomatoes, and relates to a method for estimating cultivation conditions for enabling such cultivation, and a cultivation method using such cultivation conditions.

[0011] Regarding the cultivation technology of high-sugar tomatoes, the following is known. (1) Examples of conventional cultivation technologies for high-sugar tomatoes For example, it is known that high-sugar tomatoes with a sugar content of 8 or more can be cultivated by combining the methods (a) to (c) shown below and other methods. (a) Restriction of water absorption amount (lowering the watering frequency, controlling the soil moisture rate, inhibiting water absorption by increasing the osmotic pressure by adding salts). (b) Using varieties with a high sugar content originally. (c) Root zone restriction (using small pots, using a root barrier sheet / film).

[0012] (2) Problems in high-sugar tomato cultivation If appropriate water management is not carried out with high precision, the following problems (a) to (c) will occur. (a) Disorders such as blossom end rot (fruit tip browning) occur in the fruits. (b) In the case of close planting cultivation, diseases are likely to occur. (c) If the watering is too little on a fine day, the plants will wilt, and conversely, on a day with high humidity, the water supply is likely to be excessive and the sugar content will not increase.

[0013] Conventionally, in order to explore the liquid supply conditions under which high-sugar tomatoes (for example, with a sugar content of 8 or more) can be harvested while maintaining a low incidence of disorders such as blossom end rot (for example, 10% or less), it was necessary to repeat cultivation tests and try by trial and error. On the other hand, in recent years, so-called omics research has been carried out using molecular biological techniques to analyze the dynamics of gene expression and metabolites in plants, and a lot of knowledge has been accumulating. However, at present, these findings have not been linked to the improvement of cultivation technology.

[0014] Based on this background, the inventors have developed a method for quickly exploring optimal irrigation conditions for crop varieties, types, and cultivation environments, based on physiological mechanisms elucidated from molecular fluctuations within plants. To implement this method, a simulation model (first model) for estimating the rate of fruit damage and a simulation model (second model) for estimating sugar content are used.

[0015] (Regarding the first model) As mentioned above, the incidence of damaged fruit (fruit apex browning in tomatoes) is known to be high in high-sugar content cultivation. Furthermore, it is known that damaged fruit occurs frequently during the hot summer months and after fluctuations in irrigation water supply. To confirm these findings, tomato cultivation was conducted during the summer for four years, combining cooling technology (heat pump, fog and fan) with two basic conditions: one where the irrigation water supply changed from a high level to a very low level (high-very low) and another where the irrigation water supply remained low for an extended period (low-low). Figure 1(a) is a graph showing the daily changes in irrigation water supply (ml / plant) under each condition. Figure 1(b) is a graph showing the effect of irrigation water supply conditions on the incidence of damaged fruit in the first flower cluster.

[0016] As shown in Figure 1(a), the amount of irrigation water supplied from September 18 to October 7 was less than the amount supplied under the "low-low" condition, from October 8 to November 4 the amount supplied under the "low-low" condition was greater than or equal to the "high-low" condition, and from November 5 to December 7 the amount supplied under the "low-low" condition was greater than the "high-low" condition. In the example in Figure 1(a), the grower adjusted the amount of irrigation water supplied according to the growth of the tomatoes and the weather. As a result, as shown in Figure 1(b), the incidence of defective fruit in the first flower cluster (first to third fruit) was about 3% under the "low-low" condition, while it was over 30% under the "high-low" condition.

[0017] The inventors obtained the following data during the cultivation experiment shown in Figures 1(a) and 1(b): (a) cultivation environment measurement data, (b) crop trait measurement data, and (c) comprehensive molecular variation information data. (a) Cultivation environment measurement data Cultivator, cultivation location, cultivation purpose, variety, planting density, number of cultivation stages, sowing date, transplanting date, planting date, survey end date, photosynthetic effective solar radiation, indoor temperature, humidity, CO2 concentration, nutrient solution volume, nutrient solution EC (b) Crop trait measurement data Weekly yield of ripe fruit (including damaged fruit), plant height, number of unfolded leaves, number of leaves bearing fruit, stem diameter directly below cotyledons, date of flowering of the first inflorescence, node at which the first inflorescence is formed, leaf length of the first fruit cluster, leaf width of the first fruit cluster, stem diameter below the first fruit cluster, number of fruits per plant, size of ripe fruit (fruit weight, vertical dimension (cm), horizontal dimension (maximum diameter (cm), minimum diameter (cm))), fruit components (sugar content, acidity, lycopene concentration, fresh weight), and incidence of damaged fruit (fruit with browning at the apex). (c) Comprehensive molecular variation information data Samples for comprehensive molecular variation acquisition (along with recording the date and time of collection, gene expression (transcriptome) data for all genes, plant hormones (hormonome), and secondary metabolites (metabolome) data are obtained for developing fruits).

[0018] The inventors decided to extract gene expression in underdeveloped fruits that is related to the rate of damaged fruit development from this data and to create a simulation model using the mechanism of damaged fruit development.

[0019] Figures 2(a) to 2(c) show the average expression levels of various genes in underdeveloped fruits sampled at 42, 55, 69, and 84 days after transplanting. Note that since damage had already occurred by day 84, data that could not be used to control the rate of damaged fruit (expression levels at day 84) were not considered.

[0020] Figure 2(a) shows the average expression levels of chitinase genes involved in cell wall degradation in response to ethylene-jasmonic acid in undeveloped fruits sampled at 42, 55, 69, and 84 days after transplanting. This group of chitinase genes includes Solyc01g060020.2.1, Solyc10g074390.1.1, Solyc10g068350.1.1, Solyc10g074360.1.1, and Solyc10g017980.1.1. From Figure 2(a), it can be seen that the expression level of chitinase genes is higher under conditions with a high incidence of damaged fruit ("high-very low" conditions) than under conditions with a low incidence of damaged fruit ("low-low" conditions). However, as shown in Figure 2(a), gene expression levels increase with developmental stage, making it difficult to create a relationship with the incidence of damaged fruit.

[0021] Other gene groups include genes correlated with fruit weight (fruit weight-correlated genes) and genes correlated with fruit development stages (fruit development-related genes). Examples of fruit weight-correlated genes include Solyc01g096970.2.1, Solyc02g062140.2.1, Solyc05g016190.2.1, Solyc11g062350.1.1, and Solyc10g050120.1.1. Examples of fruit development-related genes include Solyc03g115340.1.1, Solyc03g115270.1.1, and Solyc03g115300.1.1. Figure 2(b) shows the average expression levels of fruit weight-correlated genes in undeveloped fruits sampled at 42, 55, 69, and 84 days after planting. Figure 2(c) shows the average expression levels of fruit development-related genes in undeveloped fruits sampled at 42, 55, 69, and 84 days after planting. For example, the expression level of fruit weight-related genes in Figure 2(b) is lower under conditions with a high incidence of damaged fruit ("high-very low") than under conditions with a low incidence of damaged fruit ("low-low" condition). However, since the expression level increases with developmental stage, it is difficult to create a relationship with the incidence of damaged fruit, similar to the chitinase gene. Similarly, for the fruit development-related genes shown in Figure 2(c), the expression level changes according to the rate of increase in fruit weight, making it difficult to create a relationship with the incidence of damaged fruit.

[0022] In contrast, by focusing on the expression ratio of chitinase genes and fruit weight-correlated genes obtained from Figures 2(a) and 2(b), as shown in Figure 3, it was found that an expression ratio of approximately 1.3 could be used as a threshold to roughly distinguish between conditions with a high rate of damaged fruit occurrence and conditions with a low rate of damaged fruit occurrence. Furthermore, it was found that the subsequent rate of damaged fruit occurrence could be estimated from the balance between chitinase, which increases with the expansion of the outer skin, and the increase in fruit weight.

[0023] Furthermore, it was found that the relationship between chitinase genes, fruit weight correlation genes (10 genes), and the incidence of damaged fruit can be expressed as a simple linear relationship as shown in equation (1) below. Damaged fruit incidence rate = 193.209 + 10.9 × Solyc01g060020.2.1 +(-7.7)×Solyc10g074390.1.1+36.9×Solyc10g068350.1.1 +(-28.2)×Solyc10g074360.1.1+(-4.2)×Solyc10g017980.1.1 +(-46)×Solyc01g096970.2.1+14.4×Solyc02g062140.2.1 +(-7.5)×Solyc05g016190.2.1+15×Solyc11g062350.1.1 +(-2.9) × Solyc10g050120.1.1 …(1)

[0024] Furthermore, as shown in Figure 4, it was found that the subsequent incidence of defective fruit correlates with the gene expression levels at any developmental stage.

[0025] Based on Figure 3, the inventors aimed to suppress the rate of damaged fruit by ensuring that the expression level of the chitinase gene does not become excessively high relative to the expression level of the fruit weight-correlated gene. To achieve this, it was necessary to calculate the expression levels of these genes under different cultivation conditions. Therefore, the inventors created a machine learning model to calculate gene expression from the cultivation environment. Note that gene expression levels change depending on the amount of irrigation and developmental stage, as shown in Figures 2(a) to 2(c), but they also change depending on solar radiation, humidity, and temperature.

[0026] The inventors then generated a simulation model (first model) as shown in Figure 5(a) using the relationship between past cultivation condition data and the expression level data of specific genes (in this embodiment, the chitinase gene and fruit weight correlation gene) obtained in past cultivation (first relationship), and the relationship between the expression level data of specific genes and the rate of defective fruit occurrence (second relationship). This simulation model uses cultivation conditions as the explanatory variable and the rate of defective fruit occurrence of cultivated tomatoes as the objective variable.

[0027] (Regarding the second model) The inventors have created a simulation model (second model) as shown in Figure 5(b) to estimate the sugar content of crops from cultivation conditions. The simulation model in Figure 5(b) includes the relationship between cultivation conditions and gene expression levels (third relationship), and the relationship between gene expression levels and the sugar content of cultivated crops (fourth relationship). This simulation model in Figure 5(b) can be said to be a model in which cultivation conditions are the explanatory variable and sugar content is the dependent variable. Details on which genes should be selected as genes, and which gene expression levels at which time should be used, are disclosed in Japanese Patent Publication No. 2023-122061, so a detailed explanation is omitted here.

[0028] The simulation model in Figure 5(b) allows us to determine what sugar content can be achieved under different cultivation conditions (including irrigation conditions). In other words, the simulation model in Figure 5(b) can be used to estimate the cultivation conditions (including irrigation conditions) necessary to grow crops with a predetermined sugar content.

[0029] (Regarding the automatic fluid supply system 100) The automatic liquid supply system 100 of this embodiment will be described in detail below.

[0030] Figure 6 shows the configuration of an automatic nutrient solution supply system 100 used for hydroponic cultivation of tomatoes. The automatic nutrient solution supply system 100 comprises a growing container 105, a support frame 102 suspended and supported by a wire 104 from a pivot point 103 and capable of holding the growing container 105, a weight sensor 106 such as a load cell that converts the suspension load applied to the pivot point 103 into an electrical signal, and a controller 107 to which the electrical signal from the weight sensor 106 is input. The automatic nutrient solution supply system 100 also comprises a tank 108 that stores nutrient solution, a nutrient solution supply pump 109 installed inside the tank 108, and piping 110 that supplies the nutrient solution from the tank 108 to the growing container 105.

[0031] The growing container 105 is filled with a growing medium such as coconut fiber, and the tomato plants P are planted in it.

[0032] The controller 107 has a CPU 111 that controls the entire automatic liquid supply system 100 and performs necessary calculations. The CPU 111 receives electrical signals from the weight sensor 106 and can output control signals to the liquid supply pump 109. The controller 107 also has a memory unit 112 that stores the maximum weight obtained from the weight sensor 106, the liquid supply time period (e.g., 7am to 4pm), the first and second ratios R1 and R2 for determining the liquid supply amount for the first and subsequent supplyes, and the control program. The liquid supply time period, the first and second ratios R1 and R2, etc., are input via the input unit 113 and can be changed as needed. Furthermore, the controller 107 has a timer 114 for counting the liquid supply time, etc. The controller 107 also has a communication unit 115 that can communicate with external devices.

[0033] The controller 107 is connected to the server 10 via a network 80 such as the Internet. The server 10 generates a decision tree model for determining thresholds used in automatic liquid supply control by the controller 107, and provides the decision tree model to the controller 107. The controller 107 uses the decision tree model to determine thresholds and uses them for the automatic liquid supply control described below.

[0034] Figure 7 is a flowchart illustrating the automatic nutrient solution control and hydroponic cultivation method using controller 107. Figure 8 is a graph showing the timing and amount of nutrient solution supplied during the process shown in Figure 7.

[0035] When the process shown in Figure 7 begins, first, in step S10, the controller 107 starts measuring the plant weight Wx, including the growing container 105, using the weight sensor 106 and storing it in the memory unit 112. Next, in step S12, the controller 107 waits until the nutrient supply time begins.

[0036] When the liquid supply time begins, the process moves to step S14, where the controller 107 determines whether the previous day's maximum weight Wm is stored in the memory unit 112. If the determination in step S14 is denied (i.e., the previous day's maximum weight Wm is not stored), the process moves to step S16, where the controller 107 operates the liquid supply pump 109 to supply liquid to the cultivation container 105 for the first time. The amount of liquid supplied at this time is assumed to be a predetermined fixed amount.

[0037] On the other hand, if the judgment in step S14 is affirmed (i.e., the maximum weight Wm from the previous day was remembered), the process proceeds to step S18, where the controller 107 sets the difference between the maximum weight Wm from the previous day and the plant weight Wx at that time as the minimum hydration volume Smin. Then, the controller 107 multiplies the minimum hydration volume Smin by a predetermined first ratio R1 (for example, 1.1 or 1.2) and uses that value as the first hydration volume to supply hydration (see T1 in Figure 8).

[0038] After step S16 or S18, when the system proceeds to step S20, the controller 107 measures the plant weight immediately after hydration and stores it in the memory unit 112 as the maximum weight Wm for the day.

[0039] Next, in step S22, the controller 107 takes into account the decrease in plant weight Wx due to evapotranspiration of the tomato plant P and starts calculating the relative weight Rw (Rw = Wx / Wm × 100), which is the percentage of the plant weight Wx relative to the maximum weight Wm. From here on, the controller 107 repeats the calculation of the relative weight Rw at predetermined intervals.

[0040] Next, in step S24, the controller 107 waits until the calculated relative weight Rw falls to a threshold calculated by the method described later. When the relative weight Rw falls to the threshold, the process moves to step S26, and the controller 107 performs the second hydration (T2 in Figure 8). The amount of hydration at this time is determined by multiplying the maximum hydration amount Smax, which is the difference between the maximum weight Wm for the day and the plant weight Wx when the threshold is reached, by a predetermined second ratio R2 (for example, 0.5). The amount of hydration at this time may be a predetermined amount (a fixed amount).

[0041] Then, moving to step S28, the controller 107 determines whether the hydration time has ended. If the determination in step S28 is negative, the process returns to step S24. If step S24 is affirmed again, the controller 107 performs the third hydration in step S26, using an amount calculated by multiplying the difference between the maximum weight Wm and the plant weight Wx corresponding to the relative weight Rw at that time by a second ratio R2, which is used as the maximum hydration amount (T3 in Figure 8). In this case as well, the hydration amount may be a predetermined amount (a fixed amount).

[0042] The fluid supply process is repeated in this manner, and once the judgment in step S28 is confirmed, the fluid supply control for the day ends, and the controller 107 returns to step S10.

[0043] (Regarding Server 10) Server 10 generates a model (simulation model) for simulating tomato cultivation, and from this simulation model, generates a decision tree model used by controller 107 to determine thresholds.

[0044] Figure 9 shows the hardware configuration of server 10. As shown in Figure 9, server 10 includes a CPU (Central Processing Unit) 90, ROM (Read Only Memory) 92, RAM (Random Access Memory) 94, storage (here, HDD (Hard Disk Drive) or SSD (Solid State Drive)) 96, network interface 97, and a portable storage medium drive 99, etc. Each of these components of server 10 is connected to bus 98. In server 10, the CPU 90 executes programs (including a liquid supply condition estimation program) stored in ROM 92 or storage 96, or programs read from the portable storage medium 91 by the portable storage medium drive 99, thereby realizing the functions of each component shown in Figure 10. Note that the functions of each component in Figure 10 may also be realized by integrated circuits such as ASIC (Application Specific Integrated Circuit) or FPGA (Field Programmable Gate Array).

[0045] Figure 10 shows a functional block diagram of server 10. In server 10, the CPU 90 executes a program, thereby realizing the functions of the simulation model generation unit 32, the decision tree model generation unit 34, and the output unit 36, as shown in Figure 10.

[0046] The simulation model generation unit 32 identifies the relationship (first relationship) between past cultivation condition data stored in the cultivation condition DB 50 and the expression level data of specific genes (in this embodiment, the chitinase gene and fruit weight correlation gene mentioned above) stored in the gene expression level DB 52. The simulation model generation unit 32 also identifies the relationship (second relationship) between the expression level data of specific genes stored in the gene expression level DB 52 and the rate of defective fruit occurrence stored in the crop trait DB 54. Then, the simulation model generation unit 32 uses the first and second relationships to generate a simulation model (the first model in Figure 5(a)) for performing a cultivation simulation. This simulation model is a model that shows the relationship between cultivation conditions and the rate of defective fruit occurrence of tomatoes grown under those cultivation conditions.

[0047] Here, the cultivation conditions DB50 stores the cultivation environment measurement data mentioned above, which represents the cultivation conditions when the cultivation experiment was actually conducted. The gene expression level DB52 stores the comprehensive molecular variation information data mentioned above, linked to the cultivation conditions stored in the cultivation conditions DB50. The crop trait DB54 stores the crop trait measurement data mentioned above, linked to the cultivation conditions stored in the cultivation conditions DB50.

[0048] In addition to the simulation model described above, the simulation model generation unit 32 also generates the following models. (A) A model showing the relationship between cultivation conditions and the sugar content of tomatoes grown under those conditions (Figure 5(b), second model). (B) A model that calculates the maximum weight (Wm) for a given day from the plant height and stem diameter, or the daily variation in plant height, based on crop trait measurement data. (C) A model that calculates the daily threshold for relative weight Rw from the maximum weight (Wm) for that day, water intake, and environmental data.

[0049] Model (A), as shown in Figure 5(b), includes the relationship between cultivation conditions and the expression level of genes related to sugar content (the third relationship), and the relationship between the expression level of genes related to sugar content and the sugar content of the cultivated crops (the fourth relationship).

[0050] Model (B) is a model that estimates the maximum weight (Wm) from plant height and stem diameter, or from plant height, in cases where the maximum weight (Wm) for the day cannot be measured. Figure 11 is a graph showing the relationship between the measured maximum weight (Wm) and the maximum weight (estimated Wm) estimated from plant height and stem diameter using the model. As shown in Figure 11, the correlation coefficient R is 0.97, indicating a high correlation between Wm and estimated Wm. Furthermore, there is a correlation between maximum weight Wm and stem diameter as shown in Figure 12(a), and there is also a correlation between maximum weight Wm and plant height as shown in Figure 12(b), so model (b) is feasible. Note that if the maximum weight Wm for each day can be measured, it is not necessary to generate model (B).

[0051] Model (C) is, for example, the model shown in equation (2) below. The threshold for relative weight Rw = f(solar radiation, temperature, humidity, nitrogen supply, maximum weight Wm, incidence of damaged fruit (fruit apex browning), days after planting, date of flowering of the first fruit cluster) ... (2) Note that 'f' represents a function whose variable is the value specified in the parentheses.

[0052] By combining these models, the subsequent rate of defective fruit can be determined from the maximum weight (Wm) (or plant height or stem diameter), the threshold of relative weight Rw, and cultivation environment measurements. For example, if approximately 10,000 patterns of environmental measurements and subsequent environmental measurements are generated and input into the model, it is possible to calculate the threshold of relative weight Rw that is likely to result in a high rate of defective fruit and the threshold of relative weight Rw that is likely to result in a low rate of defective fruit. However, calculating the thresholds each day would result in an enormous amount of computation. Therefore, in this embodiment, the decision tree model generation unit 34 of the server 10 generates a decision tree model using the above-described models.

[0053] The decision tree model generation unit 34 learns patterns (e.g., approximately 10,000 patterns) where the defect rate is below a certain level (e.g., 10% or less) and the sugar content is above a certain level (e.g., 8 or higher), simulated using the model described above, and generates a decision tree model that can calculate thresholds with less computation. Figure 13 shows an example of a decision tree model. If the maximum weight (Wm) and cultivation environment measurement values ​​are input to this decision tree model in Figure 13, the daily thresholds for keeping the defect rate below a certain level and the sugar content above a certain level will be output. In the original simulation model, humidity and temperature have a high correlation with the subsequent defect rate, while the correlation between solar radiation and the defect rate is low. However, since both humidity and temperature have a high correlation with solar radiation, in order to reduce the amount of computation, Figure 13 uses only solar radiation without using humidity or temperature. In this embodiment, such improvements make the decision tree model simpler. Note that "Wm.1day" represents the average daily maximum weight (Wm) up to the previous day, "Wm.7day" represents the average weekly maximum weight (Wm), "solar radiation.7day" represents the average weekly solar radiation, and "solar radiation.30day" represents the average monthly solar radiation. In the decision tree model in Figure 13, the process starts from the top state and branches repeatedly by determining whether the conditions shown below the rectangular box are met (yes) or not (no). When the process reaches the bottom rectangular box (end node), the numerical value written in the upper part of the rectangular box is determined to be the threshold for that day. According to the decision tree model in Figure 13, for example, if the average daily maximum weight (Wm) up to the previous day is less than 437g, the average weekly maximum weight Wm is 746g or more, and the average monthly solar radiation is 175 (μmol m -2 s -1 If the value is greater than or equal to ), the threshold for relative weight Rw becomes "77". Note that the value of n and the percentage in the lower row of the rectangular frame in Figure 13 represent the number of simulations out of 10,000 environmental condition simulations that met this condition and the percentage of the total.

[0054] The output unit 36 ​​outputs the decision tree generated by the decision tree model generation unit 34 to the communication unit 115 of the controller 107.

[0055] (Processing by Server 10) Figure 14 shows the processing performed by server 10. It should be noted that prior to the start of the processing shown in Figure 14, cultivation trials were conducted under various conditions, and the cultivation environment measurement data, comprehensive molecular variation information data, and crop trait measurement data obtained from each cultivation trial are stored in cultivation condition DB50, gene expression level DB52, and crop trait DB54.

[0056] When the processing shown in Figure 14 begins, in step S50, the simulation model generation unit 32 first identifies the relationship (first relationship) between past cultivation condition data stored in the cultivation condition DB 50 and the expression level data of specific genes (in this embodiment, chitinase genes and fruit weight correlation genes) stored in the gene expression level DB 52. This process uses, for example, multiple regression analysis, generalized linear (GLM) models, and machine learning models such as support vector machines, random forests (RF), orientation boosting (XGBoost, LightGBM), and neural networks (CNN).

[0057] Furthermore, in the next step S52, the simulation model generation unit 32 identifies the relationship (second relationship) between the expression level data of a specific gene stored in the gene expression level DB52 and the rate of defective fruit occurrence stored in the crop trait DB54. The same method as in step S50 is used for this process as well.

[0058] Next, in step S54, the simulation model generation unit 32 generates a model (simulation model) for performing cultivation simulations using the first relationship and the second relationship.

[0059] Next, in step S55, the simulation model generation unit 32 generates the simulation models (A) to (C) described above.

[0060] Next, in step S56, the decision tree model generation unit 34 uses the simulation models generated in steps S54 and S55 to perform a simulation using a large number of environmental measurement patterns and generates a decision tree model. As a result, a decision tree model like the one shown in Figure 13 is obtained.

[0061] Next, in step S58, the output unit 36 ​​outputs the decision tree model to the controller 107.

[0062] Controller 107 calculates a threshold at which the rate of defective fruit occurrence falls below a certain level by inputting, for example, the maximum weight (Wm) (or plant height or stem diameter) and cultivation environment measurements into the decision tree model once a day (e.g., early in the morning).

[0063] Figure 15 is a graph showing the average sugar content of cultivated tomatoes under two conditions: when the threshold was determined using the decision tree model of this embodiment (automatic) and when a skilled grower manually changed the threshold daily (manual). For both the automatic and manual conditions, cultivation was carried out under two conditions: without nighttime cooling and with nighttime cooling.

[0064] As shown in Figure 15, even in the automated system, it was possible to obtain tomatoes with a sugar content comparable to that obtained in the manual system (cultivated by a skilled grower). Furthermore, in all cases—no cooling (manual), no cooling (automatic), night cooling (manual), and night cooling (automatic)—the incidence of damaged fruit was kept below 10%.

[0065] As described in detail above, according to this embodiment, the simulation model generation unit 32 of the server 10 generates a simulation model showing the relationship between the irrigation conditions and the rate of damaged fruit occurrence using a first relationship between the irrigation conditions when cultivating tomatoes and the expression level of genes related to the rate of damage occurrence, and a second relationship between the expression level of genes related to the rate of damage occurrence in tomatoes and the rate of damaged fruit occurrence in tomatoes (S54). The simulation model generation unit 32 also generates simulation model (A) (a model showing the relationship between cultivation conditions and the sugar content of tomatoes cultivated under those cultivation conditions), simulation model (B) (a model that calculates the maximum weight Wm based on plant height and stem diameter or plant height), and simulation model (C) (a model that calculates the daily threshold for relative weight Rw). The decision tree model generation unit 34 generates a decision tree model by performing simulations a predetermined number of times (for example, 10,000 times) using the simulation model generated by the simulation model generation unit 32. The controller 107 of the automatic irrigation system 100 uses a decision tree model to estimate irrigation conditions (a threshold for relative weight Rw) such that the tomatoes have a desired rate of defective fruit and a desired amount of nutrients (sugar content). In this embodiment, appropriate irrigation conditions can be estimated in a shorter time compared to trial and error with cultivation conditions. Furthermore, because the controller 107 uses a decision tree model, the computational load on the controller 107 can be reduced compared to when estimation is performed each time using a simulation model, thereby reducing the processing load.

[0066] Furthermore, according to the automatic irrigation system 100 of this embodiment, since automatic irrigation is performed based on appropriate irrigation conditions, it is possible to automatically control the system so that the rate of damaged fruit occurrence in high-sugar content tomatoes is reduced.

[0067] In the above embodiment, the case in which the controller 107 of the automatic liquid supply system 100 determines the liquid supply conditions (thresholds) using a decision tree model was described, but this is not the only case. For example, the server 10 may determine the liquid supply conditions (thresholds) using a decision tree model and notify the controller 107.

[0068] In the above embodiment, the case in which the irrigation conditions (thresholds) set using a decision tree model are used in the automatic irrigation system 100 has been described, but this is not the only case. For example, the irrigation conditions estimated using the decision tree model may be output (notified) to a terminal used by a grower who cultivates manually.

[0069] In the above embodiment, a decision tree model was generated using a simulation model, and cultivation conditions (thresholds) were determined from the generated decision tree model. However, the invention is not limited to this. For example, the cultivation conditions may be estimated using the simulation model itself. It is also possible to input cultivation conditions into the simulation model. For example, if cultivation conditions are input into a simulation model (first model) that shows the relationship between cultivation conditions and the rate of defective fruit occurrence, the rate of defective fruit occurrence when tomatoes are grown under the input cultivation conditions will be output. Therefore, growers can check from the output information whether the input cultivation conditions are appropriate.

[0070] In the above embodiment, the case where the crop is a tomato was described, but it is not limited to this. Examples of high-sugar crops include melons and mangoes. Also, in the above embodiment, the case where the amount of component of the crop used in the simulation model is sugar content was described, but it is not limited to this, and the amount of component of the crop used in the simulation model may be other component amounts (such as glutamic acid content or acidity).

[0071] The above processing functions can be implemented by a computer. In this case, a program describing the processing content of the functions that the processing unit should have is provided. By executing this program on a computer, the above processing functions are implemented on the computer. The program describing the processing content can be recorded on a storage medium that can be read by a computer (except for carrier waves).

[0072] When distributing a program, it may be sold in the form of a portable storage medium such as a DVD (Digital Versatile Disc) or CD-ROM (Compact Disc Read Only Memory) on which the program is recorded. Alternatively, the program can be stored in the storage device of a server computer and transferred from the server computer to other computers via a network.

[0073] A computer executing a program stores programs, for example, those recorded on a portable storage medium or transferred from a server computer, in its own memory. The computer then reads the program from its memory and executes the processing according to the program. Alternatively, the computer can directly read the program from the portable storage medium and execute the processing according to that program. Furthermore, the computer can sequentially execute the processing according to the programs received as they are transferred from the server computer.

[0074] The embodiments described above are preferred examples of the present invention. However, the invention is not limited thereto, and various modifications are possible without departing from the spirit of the invention. [Explanation of symbols]

[0075] 10 servers 32 Simulation Model Generation Unit 34 Decision Tree Model Generation Unit 36 Output section 50 Cultivation conditions DB 52 Gene Expression Database 54 Crop Trait Database 100 Automatic Fluid Dispensing Systems

Claims

1. A first step is to generate a first model showing the relationship between the water supply conditions and the rate of damaged fruit in the crop, using a first relationship between the water supply conditions when cultivating the crop and the expression level of genes related to the incidence of damage in the crop, and a second relationship between the expression level of genes related to the incidence of damage in the crop and the rate of damaged fruit in the crop. A second step involves using the first model to estimate the irrigation conditions necessary for cultivating the crop so that it achieves a desired rate of defective fruit occurrence, A method for estimating liquid supply conditions, including the following.

2. The method for estimating irrigation conditions according to claim 1, characterized in that, in the second step, a second model showing the relationship between the cultivation environment of the crop and the amount of components contained in the crop is used to estimate the irrigation conditions for cultivating the crop so that it contains a desired amount of components.

3. The method for estimating irrigation conditions according to claim 2, characterized in that the second model is a model generated using a third relationship between the cultivation environment of a crop and the expression level of a gene associated with the inclusion of a specific component in the crop, and a fourth relationship between the expression level of a gene associated with the inclusion of a specific component in the crop and the amount of the component contained in the crop.

4. In the second step described above, Using the first and second models described above, simulations are performed a predetermined number of times to determine the irrigation conditions necessary for cultivating the crop so that it achieves a desired rate of defective fruit and a desired amount of components. Based on the results of the predetermined number of simulations, a decision tree model is created to determine the irrigation conditions. The method for estimating irrigation conditions according to claim 2, characterized in that the decision tree model is used to estimate the irrigation conditions necessary for cultivating the crop so that it has a desired rate of defective fruit occurrence and a desired amount of components.

5. The method for estimating liquid supply conditions according to claim 1, characterized in that the gene related to the incidence of injury includes an ethylene-jasmonic acid-responsive chitinase gene.

6. The method for estimating liquid supply conditions according to claim 2, characterized in that the amount of the aforementioned component is sugar content.

7. The method for estimating irrigation conditions according to claim 1, characterized in that the aforementioned crop is one of tomatoes, melons, or mangoes.

8. A method for controlling the cultivation environment of crops, characterized by controlling the cultivation environment of crops with the water supply conditions estimated by the water supply condition estimation method described in any one of claims 1 to 7.

9. A first step is to generate a model showing the relationship between the water supply conditions and the rate of damaged fruit in the crop, using a first relationship between the water supply conditions when cultivating the crop and the expression level of genes related to the incidence of damage in the crop, and a second relationship between the expression level of genes related to the incidence of damage in the crop and the rate of damaged fruit in the crop. A second step involves using the aforementioned model to estimate the irrigation conditions necessary for cultivating the crop so that it achieves a desired rate of defective fruit occurrence, A liquid supply condition estimation program characterized by having a computer execute the following.

Images