Method for predicting suitable habitats of ginkgo fruit forests based on climate and soil factors

Abstract

The present invention discloses a method for predicting the suitable habitats of ginkgo fruit forests based on climate and soil factors, belonging to the technical field of species distribution prediction. First, a climate niche model and a soil niche model are established; the climate niche model and the soil niche model are respectively used to predict the habitat types of ginkgo fruit forests and classify the habitat types of ginkgo fruit forests; the models and prediction results are verified and tested through field experiments on seeds and fruits; then the suitable soil habitats are used to filter the suitable climate habitats; the changes in the suitable habitats of ginkgo fruit forests under future climate scenarios are predicted. By combining these two models, the present invention can predict the current and future under different climate change scenarios, verify the reliability of the model output through experimental data, the prediction process is more efficient, and the results are more accurate. The prediction of the habitat suitability classification model of the present invention conforms to the understanding of plant physiology and economics, and has important significance for guiding the production practice of ginkgo fruit forests.

Description

Technical Field

[0001] The present invention belongs to the technical field of ginkgo cultivation and protection, and particularly relates to a method for predicting suitable habitats for ginkgo fruit forests based on climate and soil factors. Background Art

[0002] Ginkgo biloba L. is a dioecious long-lived plant, and its male and female reproductive organs are distributed on different trees. Ginkgo fruits have high edible and medicinal values all over the world. Ginkgo is rich in vitamins and carotenoids, which have a certain effect on delaying aging and promoting skin metabolism. At the same time, it has a positive effect on preventing hypertension, heart disease, relieving cough and reducing phlegm. The global annual expenditure on botanical drugs exceeds $7 billion, and ginkgo ranks first among botanical drugs. However, due to the impact of climate change, the suitable habitats and fruit yields of this important fruit economic forest face great uncertainties. Therefore, it is necessary to evaluate the impact of climate change on its suitable habitats while widely using ginkgo fruit forests.

[0003] Future climate predictions show that extreme climates will increase, regional precipitation changes will be significant, the number of rainy days will increase, and the regional ecological environment will be unstable. Such frequent climate phenomena will lead to the maladaptation of trees to the environment, damage the productivity and ecological value of trees, and may lead to the loss of biodiversity and habitats. It is worth noting that China is a sensitive region to global climate change, and 90% of ginkgo resources are distributed in China. In addition, the complex phenological reproduction cycle of ginkgo is not conducive to the migration of ginkgo species. Therefore, scientific and sustainable methods should be selected to evaluate the potential suitability of ginkgo fruit forests.

[0004] Species distribution models (SDMs) have become tools for predicting the impact of future climate change on plant species and formulating effective conservation strategies. MaxEnt is an efficient and accurate species distribution modeling method. It can not only predict the occurrence probability of current species, but also predict the probabilities of past and future climates. Most ecological models only use climate variables as predictors, including the niche models constructed for ginkgo. However, soil is also considered an important factor in predicting the impact of climate change on species distribution. Since climate is a key factor in soil formation, if climate and soil variables are simply incorporated into the model, there will be a confounding effect between climate and soil variables, which will lead to the soil variables affecting the prediction accuracy of the model for future climate, or the climate variables dominating the soil effect. To avoid confusion, the climate-suitable habitats and soil-condition-suitable habitats are simulated separately. Then, the soil-suitable habitats are used to filter the climate-suitable habitats, and the soil effect is incorporated into the habitat prediction. This two-step method is used to consider the climate and soil effects described by Feng et al.

[0005] Previous inventions have shown that the size of fruits is related to abiotic variables (temperature, precipitation) and geographical origin. The ability of plants to survive and adapt to environmental changes results in certain differences in fruit traits. These differences are related to the adaptability of spatial habitats to temporal and climatic changes. There are also significant geographical differences in the traits of Ginkgo biloba seeds. In addition, the invention by Guo Ying et al. believes that the physiological changes of plants under different climatic conditions can indirectly support the reliability of model outputs. Therefore, the plasticity responses of Ginkgo biloba seed traits in different climate ecological regions will be used to verify the reliability of the niche model. Summary of the Invention

[0006] The technical problem solved by the present invention: Provide a method for predicting the suitable habitats of Ginkgo biloba fruit forests based on climate and soil factors, which combines two models, can predict the current and future under different climate change scenarios, and has a more efficient prediction process and more accurate results.

[0007] Technical solution: To solve the above technical problem, the technical solution adopted by the present invention is as follows:

[0008] A method for predicting the suitable habitats of Ginkgo biloba fruit forests based on climate and soil factors, comprising the following steps:

[0009] S1: Establish a climate niche model and a soil niche model;

[0010] S2: Respectively use the climate niche model and the soil niche model to predict the habitat types of Ginkgo biloba fruit forests and classify the habitat types of Ginkgo biloba fruit forests;

[0011] S3: Conduct verification tests on the models and prediction results through field seed experiments;

[0012] S4: Then use the soil suitable habitat to filter the climate suitable habitat and incorporate the soil effect into the habitat prediction; predict the changes in the suitable habitat of Ginkgo biloba fruit forests under future climate scenarios.

[0013] Preferably, in step S1, the steps of establishing the climate niche model and the soil niche model are as follows:

[0014] S11: Obtain the natural distribution point data of Ginkgo biloba fruit forests;

[0015] S12: Obtain the scale-free climate variables of the natural distribution points and 30 basic soil indicators obtained from the World Soil Database for model construction;

[0016] S13: Data processing: Use MaxEnt and ArcGIS to select the climate and soil factors that contribute the most to the distribution of Ginkgo biloba fruit forests: Use ArcGIS to calculate the Pearson correlation coefficient between climate variables and remove two paired variables greater than 0.8;

[0017] S14: Pre - construct a climate MaxEnt model using all climate variables and run it continuously for multiple times to eliminate climate variables unrelated to the model. Pre - construct a soil MaxEnt model using all soil variables and run it continuously for multiple times to eliminate soil variables unrelated to the model;

[0018] S15: Establish a climate niche model using the climate variables screened in step S14 and a soil niche model using the screened soil variables;

[0019] S16: Conduct model training and model validation.

[0020] Preferably, in step S16, the data is randomly divided into a 75% training set for model training and a 25% validation set for model validation. Through 10 - fold repeated cross - validation, 10 models are established to avoid the uncertainty in modeling.

[0021] Preferably, use the Receiver Operating Characteristic curve ROC of the built - in MaxEnt software to evaluate the robustness of the model. The model performance is divided into failing (AUC value is 0.5 - 0.6), poor (AUC value is 0.6 - 0.7), average (AUC value is 0.7 - 0.8), good (AUC value is 0.8 - 0.9), and excellent (AUC value is 0.9 - 1).

[0022] Preferably, in step S3, the predicted structural types of the forest habitats for ginkgo fruits include four habitat types: unsuitable (P < 0.2), low - suitable (0.2 ≤ P < 0.4), medium - suitable (0.4 ≤ P < 0.6), and high - suitable (0.6 ≤ P ≤ 1), where P is the threshold.

[0023] Preferably, in step S3, establish a binary response model using the following binary quadratic polynomial equation

[0024] Y = b 0 +b 1 X 1 +b 2 X 2 +b 3 X 1 2 +b 4 X 2 2 +b 5 X 1 X 2

[0025] In the formula: Y is the peel - out rate or single - grain weight,

[0026] X 1 and X 2 are climate variables,

[0027] X 1 X 2 is an interaction term

[0028] b0, b1, b2, b3, b4, and b5 are parameters to be determined

[0029] By comparing the combined regression models of various climate variables, the best combination of two climate variables is selected to establish the optimal bivariate climate response equation; the optimized bivariate climate response function is used to predict the potential suitable distribution area of Ginkgo biloba seeds in the future in China as a reference, and the raster climate data generated by ClimateAP is used to predict the next three periods. If the predicted value is 0 or negative, it is considered that the grid point is not conducive to the good growth of the outer seeds of Ginkgo biloba. If the predicted value is greater than 0, the habitat type is divided according to the predicted value: unsuitable (P < 0.2), low suitable (0.2 ≤ P < 0.4), medium suitable (0.4 ≤ P < 0.6), and high suitable (0.6 ≤ P ≤ 1), where P is the threshold

[0030] Preferably, the interaction term is excluded from the model. Through the fitting comparison of the models, it is determined that the combination of TD (continental temperature difference) and MAP (annual rainfall) among these two climate variables is the best bivariate climate response model for establishing the single-grain weight and shelling rate of Ginkgo biloba seeds

[0031] Preferably, four climate-suitable habitat types are filtered by the soil-suitable habitat to obtain a combination of three suitable habitat types. The filtered habitats are considered suitable for climate and soil conditions and are respectively called low-suitable habitats, medium-suitable habitats, and high-suitable habitats

[0032] Beneficial effects: Compared with the prior art, the present invention has the following advantages

[0033] The method for predicting the suitable habitat of Ginkgo biloba fruit forests based on climate and soil factors of the present invention first determines the climate niche of the species, establishes a climate niche model, then considers the constraints of soil conditions, establishes an ecological niche model of soil variables, and synthesizes the simulation results of the two models to simulate the climate-suitable habitat and the soil-condition-suitable habitat. Then, the climate-suitable habitat is filtered by the soil-suitable habitat, incorporating the soil effect into the habitat prediction. The addition of soil variables will affect the prediction of the size and change trend of the suitable area. Especially in future models, in order to eliminate the influence of the correlation between climate and soil variables on the model, the present invention combines these two models, which can predict the current and future under different climate change scenarios, verify the reliability of the model output through experimental data, the prediction process is more efficient, and the results are more accurate. The prediction of the habitat suitability classification model of the present invention conforms to the understanding of plant physiology and economics and is of great significance for guiding the production practice of Ginkgo biloba fruit forests Description of the Drawings

[0034] Figure 1 are the verification results of the variable climate response model;

[0035] Figure 2 are the response curves of three important climate variables (a, b, c) and three important soil variables (d, e, f) in the climate model and the soil model;

[0036] Figure 3 is the percentage of the distribution area of Ginkgo biloba fruit - bearing forests in China at present (1960 - 1990);

[0037] Figure 4 is the difference in Ginkgo biloba seed traits under different suitable habitat categories (one - way ANOVA);

[0038] Figure 5 is the area change of the suitable habitat of Ginkgo biloba fruit - bearing forests in China in the 2020s, 2050s, and 2080s under the RCP4.5 scenario;

[0039] Figure 6 is the area change of the suitable habitat of Ginkgo biloba fruit - bearing forests in China in the 2020s, 2050s, and 2080s under the RCP8.5 scenario. Detailed implementation manners

[0040] The following further clarifies the present invention in conjunction with specific embodiments. The embodiments are implemented on the premise of the technical solution of the present invention. It should be understood that these embodiments are only used to illustrate the present invention and not to limit the scope of the present invention.

[0041] A method for predicting the suitable habitat of Ginkgo biloba fruit - bearing forests based on climate and soil factors mainly includes the following steps:

[0042] Step S1: Establish a climate niche model and a soil niche model; the specific methods for establishing the climate niche model and the soil niche model are as follows:

[0043] S11: Obtain the natural distribution point data of Ginkgo biloba fruit - bearing forests;

[0044] The natural distribution point data of Ginkgo biloba forests (Ginkgo biloba fruit - bearing forests) that can bear seeds obtained from the Chinese Ginkgo germplasm resource bank (the germplasm resource data comes from county forestry bureaus, forestry stations, and universities in China). To avoid the influence of multiple recording points in the same grid and sampling bias, 358 coordinate data of Ginkgo biloba trees were selected for the model.

[0045] S12: Obtain the scale - free climate variables of the natural distribution points and 30 basic soil indicators obtained from the World Soil Database for model construction;

[0046] The environmental variables of the present invention include 16 climate variables (Table 1) and 30 soil variables (Table 2). The climate variables were obtained using ClimateAP software (http: / / ClimateAP.net). Taking 1961 - 1990 as the reference period, scale - free climate variables of 358 data points were generated for model construction. Grid climate data of 4×4 km for the reference period (1961 - 1990), 2020s (2011 - 2040), 2050s (2041 - 2070), and 2080s (2071 - 2100) were also generated to predict the geographical distribution of climate habitats in the current and future periods. Future climate data was taken from the General Circulation Models (GCMs) of the Coupled Model Intercomparison Project Phase 5 (CMIP5) in the Fifth Assessment Report of the Intergovernmental Panel on Climate Change (IPCC 2014). A set of 15 GCMs for two climate change scenarios RCP4.5 and RCP8.5 included in the climate action plan was used. The soil variables were from 30 basic soil indicators obtained from the Harmonized World Soil Database (HWSD, Table 2)

[0047] (http: / / www.iiasa.ac.at / web / home / research / researchPrograms / water / HWSD.html), which contains raster data layers of key soil properties with a spatial resolution of 30 arc - seconds.

[0048] Table 1: List of climate variables

[0049]

[0050] Table 2: Table of soil variables

[0051]

[0052] S13: Data processing: Use MaxEnt and ArcGIS to select the climate and soil factors that contribute the most to the distribution of ginkgo fruit - bearing forests. To avoid the influence of environmental multicollinearity, use ArcGIS to calculate the Pearson correlation coefficients between climate variables and remove two pairs of variables with a correlation coefficient greater than 0.8;

[0053] S14: Pre - construct a climate MaxEnt model using all climate variables (16 climate variables) and run it continuously 3 times to eliminate climate variables unrelated to the model. Pre - construct a soil MaxEnt model using all soil variables (30 soil variables) and run it multiple times to eliminate soil variables unrelated to the model. Through model operation, 6 climate variables and 14 soil variables were selected.

[0054] S15: Establish a climate niche model using the 6 climate variables screened in step S14, and establish a soil niche model using the 14 soil variables screened.

[0055] S16: Conduct model training and model validation. Randomly divide the data into a 75% training set for model training and a 25% validation set for model validation. Through 10 repeated runs of cross-validation, 10 models are established to avoid the uncertainty of modeling. Use the Receiver Operating Characteristic curve ROC of the built-in MaxEnt software to evaluate the robustness of the model. The model performance is divided into failing (AUC value of 0.5 - 0.6), poor (AUC value of 0.6 - 0.7), average (AUC value of 0.7 - 0.8), good (AUC value of 0.8 - 0.9), and excellent (AUC value of 0.9 - 1).

[0056] S2: Use the climate niche model and the soil niche model to predict the habitat types of ginkgo fruit forests respectively, and classify the habitat types of ginkgo fruit forests; classify the invention into four habitat types: unsuitable (P < 0.2), low suitable (0.2 ≤ P < 0.4), medium suitable (0.4 ≤ P < 0.6), and high suitable (0.6 ≤ P ≤ 1) for the climate model and the soil model.

[0057] S3: Conduct a validation test on the model and the prediction results through a seed field experiment;

[0058] Collect ginkgo seed materials from 60 experimental sites across the country for testing seed traits and validating the prediction results of the model. To exclude the effects of aspects such as plant age, slope, and genetic differences. The ginkgo fruit trees collected are about 20 years old (provided by the local forestry bureau), and the distance between trees is set to at least 50 meters. Collect seeds on the first branches in the east, west, north, and south directions of each fruit tree. Mix 100 seeds collected from each sampling point and bring them back to the laboratory. Then wipe off the impurities on the fruit peel with a clean paper towel and measure the seed trait indicators.

[0059] Establish a binary response model using the following binary quadratic polynomial equation

[0060] Y = b 0 + b 1 X 1 + b 2 X 2 + b 3 X 1 2 + b 4 X 2 2 + b 5 X 1 X 2

[0061] Where: Y is the hulling rate or single-grain weight,

[0062] X 1 and X 2 are climate variables,

[0063] X 1 X 2 is an interaction term,

[0064] b0, b1, b2, b3, b4, and b5 are parameters to be determined;

[0065] By comparing the combined regression models of various climate variables, the best combination of two climate variables was selected to establish the optimal bivariate climate response equation; a bivariate climate response model was established at the species level. Therefore, the genetic variation among all the surveyed samples was regarded as random error during the modeling process. The optimized bivariate climate response function used was to predict the potential suitable distribution area of Ginkgo biloba seeds in the future in China as a reference, and the raster climate data generated by ClimateAP was used to predict the next three periods. If the predicted value is 0 or negative, it is considered that the grid point is not conducive to the good growth of the outer seeds of Ginkgo biloba. If the predicted value is greater than 0, the habitat types are divided according to the predicted values: unsuitable (P < 0.2), low suitable (0.2 ≤ P < 0.4), medium suitable (0.4 ≤ P < 0.6), and high suitable (0.6 ≤ P ≤ 1), where P is the threshold.

[0066] Verification of the bivariate climate response model

[0067] Combined with 16 climate variables, 120 groups of bivariate climate response models for single-grain weight and 120 groups of bivariate climate response models for hulling rate were established respectively. The interaction term was excluded from the models. Through the fitting comparison of the models, it was determined that the combination of the two climate variables TD (continental temperature difference) and MAP (annual rainfall) was the best bivariate climate response model for establishing the single-grain weight and hulling rate of Ginkgo biloba seeds. The response surface diagrams of these two models ( Figure 1 ) explained 53% of the total variation of single-grain weight (R2 = 0.53, P < 0.0001) and 55% of the total variation of hulling rate (R2 = 0.55, P < 0.0001). This indicates that the models have high precision.

[0068] The response surfaces of single-grain weight and hulling rate reached their peaks under the climate conditions of a continental temperature difference of 20 - 25°C and an annual rainfall of 800 - 1500 mm. This result is consistent with the climate factors with the largest contribution rate screened by the species distribution model.

[0069] S4: Then, use the soil suitable habitat to filter the climate suitable habitat and incorporate the soil effect into the habitat prediction; predict the changes in the suitable habitat of Ginkgo biloba fruit forests under future climate scenarios. By filtering the four climate suitable habitat types through the soil suitable habitat, a combination of three suitable habitat types is obtained. The filtered habitats are considered suitable for both climate and soil conditions and are respectively called low-suitable habitats, medium-suitable habitats, and high-suitable habitats. The specific filtering method is as follows: regard the suitable area of the soil as 1, the high-suitable area of the climate as 3, the medium-suitable area as 2, the low-suitable area as 1, and the non-suitable area as 0. Multiply the values of the soil by the other values of the climate, and the obtained value is the division standard of the suitability degree. This process is called filtering.

[0070] The present invention evaluates the changes in the distribution of climate suitable habitats and the changes in the distribution of climate suitable habitats filtered by soil under six future climate scenarios.

[0071] Results and Analysis of This Example

[0072] Accuracy of the MaxEnt Model and Contribution Rates of Variables

[0073] The MaxEnt models for climate and soil provided satisfactory results. The AUC values of the model cross-validation training data and test data for the climate model were 0.943 and 0.944 respectively, and for the soil model were 0.900 and 0.844 respectively. The ROC curves of both models were higher than the random model (0.5).

[0074] For the climate model, the variable that contributed the most to the model was DD < 0 (63.3%), followed by MAP (14.1%) and MCMT (11.9%), which cumulatively explained 89.3% of the model (Table 4), and the contributions of the remaining variables to the model were less than 11%. For the soil model, S-CEC-SOIL (26.4%), T-BS (11.9%), and T-CEC-CLAY (11.9%) were the main contributing factors to the distribution model, with a cumulative contribution rate of 50.2% (Table 4).

[0075] The ranges suitable for Ginkgo biloba growth for the three most important climate variables (DD < 0, MAP, and MCMT) were 0 - 25 mm, 700 - 1600 mm, and -3 - 6 mm ( Figure 2 ). The suitable ranges for the three most significant soil variables (S-CEC-SOIL, T-BS, and T-CEC-CLAY) were 8 - 16%, 10 - 53%, and 2 - 38% ( Figure 2 ).

[0076] Table 3: Contributions of Environmental Variables to the Model

[0077]

[0078] Predicting current suitable habitats

[0079] In this example, the four climatic suitability distributions of the habitats of Ginkgo biloba fruit forests were studied across China. The highly suitable habitats account for 3.3% of the national land area. The moderately suitable habitats account for 11.4%, and the low-suitable habitats are distributed around the highly suitable habitats, accounting for 5.7%. The remaining 79.6% comes from unsuitable habitats.

[0080] In the suitable habitats of the soil in the Ginkgo biloba fruit forest area, the highly suitable habitats account for 5.9% of the total land area. The moderately suitable habitats account for 13.5%, and the low-suitable habitats account for 21.2%. From the perspective of soil conditions, other areas (59.4%) are not suitable for growing this plant. The distribution of suitable soil habitats is wider than that of suitable climate habitats.

[0081] After the soil habitat filters the climate habitat, the area of highly suitable habitats accounts for 2.9%, the area of moderately suitable habitats is 10.1%, the area of low-suitable habitats accounts for 4.6%, and the unsuitable habitats account for 82.4%. As expected, the climate suitable habitats filtered by the soil are significantly smaller than the unfiltered climate suitable habitats.

[0082] Bioclimatic model validation

[0083] The results of the analysis of variance ( Figure 4 ) showed that there were significant differences in the grain weight and husk removal rate among different habitat categories (p < 0.05). In addition, by ranking these two traits, the rationality of the SDM prediction results was verified. There were significant differences in the single grain weight and husk removal rate among the high, medium, low, and unsuitable habitats.

[0084] Changes in future suitable habitats

[0085] Compared with the current distribution. Under future climate conditions, the highly and moderately suitable habitats show a decreasing trend, while the low-suitable habitats show an increasing trend ( Figure 5 and 6 ). Under the RCP4.5 and RCP8.5 scenarios, by 2020, the areas of highly suitable habitats will decrease by 15.49% and 19.22% respectively, by the 2050s by 17.66% and 25.94%, and by the 2080s by 20% and 31.35% ( Figure 5 and 6 ). The decline in the area of moderately suitable habitats is greater than that of highly suitable habitats. However, under these two scenarios, by the 2020s, the areas of low-suitable habitats will increase by 33.83% and 36.51% respectively, by the 2050s by 36.47% and 44.25%, and by the 2080s will increase by 58.05% and 63.32%.

[0086] The change trends of the four climate suitable habitats after soil filtration are consistent with those of the unfiltered climate habitats. As Figure 5 andFigure 6 As shown, under the RCP4.5 and RCP8.5 scenarios, the area of highly suitable habitats will decrease by 13.56% and 22.74% respectively by 2020, by 18.95% and 27.26% in the 2050s, and by 23.76% and 31.41% in the 2080s. The reduction rate of moderately suitable habitats is similar to that of highly suitable habitats. Under these two scenarios, by 2020, the area of low-suitable habitats will increase by 39.84% and 47.77% respectively, by 48.07% and 64.59% by 2050, and by 70.35% and 76.17% respectively in the 2080s. The decline of filtered climate-suitable habitats (high and medium) is greater than that of unfiltered climate-suitable habitats (high and medium).

[0087] Rapid climate change affects the sexual reproductive cycle of Ginkgo biloba. Therefore, the distribution of suitable habitats for Ginkgo biloba will change. Without affecting the contribution of climate variables, this invention incorporates soil variables into future predictions, establishes a high-precision ecological model (AUC > 0.9), and determines important climate variables (DD < 0, MAP, MCMT) and soil variables (S-CEC-SOIL, T-BS, T-CEC-CLAY). The model prediction method is used for habitat suitability classification and verified through experiments. Future predictions show that the proportion of the area of suitable habitats reduced after soil habitat filtration is greater than that of climate-suitable habitats. Predicting suitable habitats using climate and soil variables

[0088] The addition of soil variables will affect the prediction of the size and change trend of suitable areas, especially in future models. This invention aims to eliminate the influence of the correlation between climate and soil variables on the model. First, the climate niche of the species is determined, and then the constraints of soil conditions are considered.

[0089] In terms of climatic conditions, two temperature-related variables (DD<0 and MCMT) and one precipitation variable (MAP) make important contributions to the climate niche model of ginkgo fruit forests. Among them, DD<0 (63.3%) and MCMT (11.9%) have relatively large contribution rates to the model, indicating that the suitable climate habitat for this species is mainly restricted by low temperature. This may be related to the delayed development mechanism of ginkgo seeds that are sensitive to temperature. Due to the low temperature delaying the germination time of ginkgo seeds, the seedlings accumulate less carbohydrates before entering the hibernation period. This also confirms the previous finding that it is currently not recommended to establish large-scale ginkgo fruit tree plantations in low-temperature areas. MAP is the second important climate variable. This may be because precipitation has a direct impact on the emergence and growth of plants. For soil variables, S-CEC-SOIL (26.4%), T-BS (11.9%), and T-CEC-CLAY (11.9%) are important factors for predicting distribution. The contents of S-CEC-SOIL and T-CEC-CLAY affect the water and fertilizer retention capacity of the soil and provide sufficient nutrients and water for the growth of ginkgo fruit forests. T-BS is related to the available nutrient content in the surface soil, so it is logical that it becomes the most important soil factor in the soil niche model.

[0090] Distribution of bioclimatic habitat classification

[0091] According to the model prediction, from high-habitat to unsuitable habitat, the single-seed weight and shelling rate under the four habitats show a downward trend. Seed weight is an important indicator for quickly screening excellent ginkgo varieties. Long-term ecological and geographical isolation has led to large variations in seed weight in different production areas. Therefore, the prediction of the habitat suitability classification model conforms to the understanding of plant physiology and economics and is of great significance for guiding the production practice of ginkgo fruit forests.

[0092] Impact of climate change

[0093] Under future climate scenarios, the potential climate suitable habitats for ginkgo fruit forests will decrease. Consistent with the predictions of other species in the same region, due to the impact of climate change, drought and severe high temperatures have led to a reduction in the range of suitable habitats.

[0094] After being filtered by the soil habitat, the overall area of suitable habitats still shows a downward trend, but is higher than that of the climate-suitable habitats. Under the RCP4.5 and RCP8.5 scenarios, by 2020, the area of suitable habitats (high and medium) will be reduced by half. This may be because, assuming in the time-scale model, the soil does not change as rapidly as the climate, so the current and future soil conditions are similar. In this case, the suitable species habitats determined by the soil are less affected by climate change than the suitable habitats determined by the climate. The present invention uses climate and soil variables to establish a niche-based model, which can map the distribution of the habitat types of ginkgo fruit forests. In addition, by combining these two models, the current and future under different climate change scenarios can be predicted. The reliability of the model output is verified by experimental data. Future predictions show that climate change will have a negative impact on the main production areas of ginkgo fruit forests.

[0095] The above are only the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. A method for predicting the suitable habitats of ginkgo fruit forests based on climate and soil factors, characterized in that, it includes the following steps: S1: Establish a climate niche model and a soil niche model; the steps are as follows: S11: Obtain the natural distribution point data of ginkgo fruit forests; S12: Obtain the scale-free climate variables of the natural distribution points, and 30 basic soil indicators obtained from the World Soil Database for model construction; S13: Data processing: Use MaxEnt and ArcGIS to select the climate and soil factors that contribute the most to the distribution of ginkgo fruit forests: Use ArcGIS to calculate the Pearson correlation coefficient between climate variables, and remove two paired variables greater than 0.8; S14: Pre-build a climate MaxEnt model with all climate variables and run it continuously multiple times to eliminate climate variables irrelevant to the model. Pre-build a soil MaxEnt model with all soil variables and run it continuously multiple times to eliminate soil variables irrelevant to the model; S15: Establish a climate niche model with the climate variables screened in step S14 and a soil niche model with the screened soil variables; S16: Conduct model training and model verification; randomly divide the data into a 75% training set for model training and a 25% validation set for model verification. Through 10 repeated runs of cross-validation, 10 models are established to avoid the uncertainty of modeling; Use the receiver operating characteristic curve ROC of the built-in MaxEnt software to evaluate the robustness of the model. The model performance is divided into the following categories: when the AUC value is 0.5 - 0.6, it is failing; when the AUC value is 0.6 - 0.7, it is poor; when the AUC value is 0.7 - 0.8, it is average; when the AUC value is 0.8 - 0.9, it is good; when the AUC value is 0.9 - 1, it is excellent; S2: Respectively use the climate niche model and the soil niche model to predict the habitat types of ginkgo fruit forests and classify the habitat types of ginkgo fruit forests; S3: Conduct a validation test on the model and the prediction results through field experiments; Establish a binary response model using the following binary quadratic polynomial equation Y = b 0 + b 1 X 1 + b 2 X 2 + b 3 X 1 2 + b 4 X 2 2 + b 5 X 1 X 2 ; where: Y is the shelling rate or single-grain weight, X 1 and X 2 are climate variables, X 1 X 2 is an interaction term b 0 ,b 1 ,b 2 ,b 3 ,b 4 and b 5 are parameters to be determined; By comparing the combined regression models of each climate variable, select the best combination of two climate variables to establish an optimal bivariate climate response equation; use the optimized bivariate climate response function to predict the potential suitable distribution area of future ginkgo seeds in China as a reference, and use the raster climate data generated by ClimateAP to predict the next three periods. If the predicted value is 0 or negative, it is considered that the grid point is not conducive to the better growth of ginkgo seeds; Combined with 16 climate variables, 120 groups of bivariate climate response models for single-seed weight and 120 groups of bivariate climate response models for shelling percentage were established respectively. Interaction terms were excluded from the models. Through the fitting comparison of the models, it was determined that the combination of the two climate variables, continental temperature difference TD and annual rainfall MAP, was the best bivariate climate response model for establishing the single-seed weight and shelling percentage of ginkgo seeds. The response surfaces of single-seed weight and shelling percentage reached their peaks under the climate conditions of a continental temperature difference of 20 - 25 °C and an annual rainfall of 800 - 1500 mm. Interaction terms were excluded from the models. Through the fitting comparison of the models, it was determined that the combination of the two climate variables, continental temperature difference and annual rainfall, was the best bivariate climate response model for establishing the single-seed weight and shelling percentage of ginkgo seeds. The predicted structural types of the habitats of ginkgo fruit forests include four habitat types: P < 0.2 is unsuitable, 0.2 ≤ P < 0.4 is low suitability, 0.4 ≤ P < 0.6 is medium suitability, 0.6 ≤ P ≤ 1 is high suitability, where P is the threshold. By filtering the four climate-suitable habitat types through the soil-suitable habitat, a combination of three suitable habitat types was obtained. The filtered habitats were considered suitable for climate and soil conditions and were respectively called low-suitable habitats, medium-suitable habitats, and high-suitable habitats. S4: Then, use the soil-suitable habitat to filter the climate-suitable habitat and incorporate the soil effect into the habitat prediction; predict the changes in the suitable habitats of ginkgo fruit forests under future climate scenarios.

Images