Study On The Influence Of Ecological Factors On The Quality Of Flue-cured Tobacco In Yuxi Area

In order to meet the demand for precise procurement of flue-cured tobacco raw materials by my country's current cigarette industry enterprises,improve the refinement of flue-cured tobacco zoning;understand the quantitative relationship between the chemical components of flue-cured tobacco and ecological factors,and improve the intelligence of flue-cured tobacco quality evaluation.This study analyseded the meteorological data,soil data,and flue-cured tobacco quality data from 2009 to 2018,including appearance quality,physical properties,chemical composition,sensory quality,155 flue-cured tobacco distribution data and 23environmental variable data,using a small grid to calculate the model in Yuxi area.The MaxEnt model analyzes the climatic suitability of Yuxi flue-cured tobacco,and selects the most accurate climatic suitability zoning results;The quality of Yuxi flue-cured tobacco is divided by index sum method,AHP analytic hierarchy method,and two-dimensional graph theory clustering method,and four quality type areas are obtained;ecological factor influence comprehensive index,BP neural network optimized by the gray wolf algorithm,partial minimum The two-fold method constructs the ecological prediction model of flue-cured tobacco chemical composition based on mechanism algorithm,intelligent algorithm,and partial least square algorithm.Finally,it will provide a scientific basis for optimizing the planting layout of flue-cured tobacco and improving the quality of flue-cured tobacco in Yuxi area,which is of great significance for the sustainable development of the flue-cured tobacco industry in Yuxi and the improvement of flue-cured tobacco meteorological services.The main conclusions of this study are as follows:(1)Comparative analysis of the zoning results of flue-cured tobacco climate suitability of different methods(small grid estimation model and MaxEnt model)shows that the general zoning results are basically consistent.The most suitable areas are located in Jiangchuan,Huaning,and Tonghai,while suitable and sub-suitable areas are mainly distributed in parts of Eshan,Xinping,Yimen and Yuanjiang.The unsuitable areas are relatively concentrated,located in the narrow and long area from Yuanjiang to Xinping,where the altitude is below1000m and is not suitable for the growth and development of flue-cured tobacco.The simulated climate suitability zoning results of the MaxEnt model are better than the results of the small grid calculation model,and are more suitable for Yuxi flue-cured tobacco climate suitability analysis.Compared with the traditional four regionalization index methods,the six regionalization indexes selected by MaxEnt model and statistical method can effectively improve the accuracy of MaxEnt model simulation results.(2)Four quality type areas are obtained through two-dimensional graph theory clustering.The proportions of these four types of areas are 29.09%,25.45%,23.64%,and 21.82%respectively.The four types of areas are evenly distributed on the whole,indicating that two-dimensional graph theory clustering has a good effect.Analysis of variance was used to study the differences of various quality types.The quality zoning indicators all showed significant,among which altitude,climate suitability index CFI,flue-cured tobacco quality index FQI,sweet aroma,roasted sweet aroma,and burnt sweet aroma reached extremely significant(P<0.01),the soil suitability index SFI reached significant(p<0.05).Therefore,the specific differences can be described by comparing the average values of the quality zoning indicators of different types of areas.The first type area is the most suitable area for flue-cured tobacco planting climate and soil,and its style is characterized by a typical fresh-sweetness scent;the second type area is a flue-cured tobacco planting area with sub-suitable climate and the most suitable soil,and its style is characterized by fresh-sweetness scent;the third type area is the most suitable area for flue-cured tobacco planting climate and soil,and its style is characterized by typical fresh-sweetness scent;the fourth type area is the suitable area for flue-cured tobacco planting climate and soil,and its style is weak fresh-sweetness scent.(3)The ecological prediction model of the chemical composition of flue-cured tobacco with intelligent algorithm has the best simulation effect and can provide accurate results in practical applications.The absolute error of the intelligent chemical composition prediction model optimized by the gray wolf algorithm is significantly smaller than before optimization,indicating that the simulation effect of the optimized model is better.Its R~2 is greater than0.95,and the RMSE is less than 0.1.The R~2 of other prediction models except total nitrogen is as high as 0.99.In actual production,different prediction models can be selected based on different purposes.Among them,the mechanism prediction model parameters have clear physical meanings,and the partial least squares algorithm prediction model can show the importance of each ecological factor to the chemical composition.The prediction model of mechanism algorithm can judge the influence degree of ecological factors on the chemical components of flue-cured tobacco and whether the influence is promoting effect or adverse effect by the size,positive and negative of the comprehensive index of ecological factors.The average R~2 of mechanism algorithm ecological prediction model is 0.29,and the average RMSE is 0.13.Partial least squares algorithm prediction model can find out the key ecological factors affecting each chemical component by the value of VIP value of projection importance coefficient of variables.For example,the key ecological factor of nicotine is the sunshine hours in flourishing period,and its VIP value is 1.89.Except for total sugar and reducing sugar,R~2 of total nitrogen,potassium and chlorine prediction model of partial least squares algorithm prediction model is greater than 0.8,and RMSE is less than 0.1.