Investigation of ecological factors controlling quality of flue-cured tobacco (Nicotiana tabacum L.) using classification methods

• Ecological factors affecting tobacco quality were explored using k-means and CART.
• Geographical attributes and weather factors showed high effects on tobacco quality.
• For weather indices, temperature-related ones had strong impacts on tobacco quality.
• High contents of soil nutrients could not ensure high quality of tobacco.

The growth and quality of tobacco are associated with ecological conditions, such as soil, climate or weather, and geographical attributes. Tobacco, especially flue-cured tobacco, is an important cash crop and widely cultivated in southwestern China. However, knowledge about critical ecological indicators affecting quality of flue-cured tobacco is limited in this region. In the current study, two well-known clustering algorithms, i.e., k-means and classification and regression trees (CART), were applied to investigate the critical ecological indicators controlling quality of flue-cured tobacco. On the basis of six quality indices and Davis–Bouldin index, a total of 142 flue-cured tobacco leaf samples were classified into three groups with low, medium, and high quality using k-means algorithm. The results obtained by CART model showed that geographical attributes (altitude, latitude, and longitude) and weather indicators had high effects on the quality of flue-cured tobacco followed by soil properties and varieties. Flue-cured tobacco plants with high quality preferred to be grown in areas with low values of altitude, rainfall and relative humidity, high values of latitude, longitude, sunshine hours, and temperature-related indices (mean, maximum and minimum temperatures and their difference), and low concentrations of soil nutrients in this study area. Nevertheless, further study should be conducted to understand the interaction among the ecological variables.