In this study, the purpose was to comprehensively evaluate the cold tolerance of cotton varieties (lines) at seedling emergence stage, establish a reliable evaluation model, screen and identify indicators, and provide a simple and effective evaluation method for the selection and identification of cold-tolerant varieties in cotton.

200 upland cotton varieties (lines) were used to test hypocotyl length, root length and 100-grain weight, etc. under three treatments of constant chilling (CC), diurnal variation of chilling (DVC) and normal conditions. A combination of integrated cold tolerance coefficient difference analysis, frequency analysis, drop analysis, principal component analysis, cluster analysis, and multiple regression analysis were used to classify their cold tolerance types, establish cold tolerance prediction models, and screen evaluation parameters.

The variation of each parameters at normal conditions were minor fluctuations ranging from 3.12% to 18.89%. The seedling emergence rate was above 85.00%, which had high viability and could be used for subsequent cold tolerance analysis. The variability of each parameter within the accessions increased under chilling stress, ranging from 7.14%-108.33%, and the most variable parameter were root length under CC condition and germination index under DVC condition. Principal component analysis converted the 14 parameters under chilling stress and 100-grain weight measured into six mutually independent composite indicators, representing 74.98% of the total data information. The comprehensive cold tolerance evaluation value (D) was calculated by the affiliation function method and then clustering analysis was performed. 200 cotton varieties (lines) were divided into five categories according to their cold tolerance, with 2 of the group Ⅰ being strongly cold tolerant, 42 of the group Ⅱ being cold tolerant, 69 of the group Ⅲ being medium cold tolerant, 83 of the group Ⅳ being more sensitive, and 4 of the group Ⅴ being sensitive, of which Xinluzhong 16 was the most cold-tolerant material. A multiple regression analysis was applied to establish a prediction model for cold tolerance of cotton at seedling emergence as Y=-4.10+0.58X4+0.40X14+0.32X1+0.22X5 (R²=0.92), and four parameters for cold resistance evaluation were confirmed, namely total length of seedling, emergence rate, and dry weight under CC stress, germination rate under DVC stress. The cold-tolerant varieties (lines) had higher seedling emergence rates of early sowing experiment in the field, which were basically consistent with the results of the indoor results.

It is feasible to use CC and DVC stress combined with multivariate statistical analysis to evaluate the cold tolerance of cotton at seeding stage, and total length of seedling, emergence rate, and dry weight under CC stress, germination rate under DVC stress can be used as evaluation parameters.