Objectives Soil fertility in coffee plantations of Yunnan was surveyed through sampling analysis to understand its effects on quality of green coffee beans and provide reference for soil management in the area.

Methods A total of 49 soil samples (0–20 cm depth) and 38 green coffee bean samples were collected in main coffee growing-regions of Yunnan Province. The soil organic matter, pH, available N, P and K contents were measured. The ash content, caffeine, total sugar, reducing sugar and fat contents of green coffee bean samples were analyzed. The membership function types and their thresholds for each fertility index were determined. The principal component analysis was employed to calculate the weight of each fertility item, and the additive and multiplicative theorem was used to estimate integrated fertility index (IFI) of each soil sample. The IFI values were clustered by using the Euclidean distance clustering method, and then each class was defined according to the IFI values. The correlations between quality of green coffee beans and soil nutrients were analyzed by using the canonical correlation analysis (CCA).

Results The results of the comprehensive evaluation indicated that there was significant variation (P < 0.05) in soil fertility in Yunnan coffee growing regions. The IFI values of the soil samples mostly ranged from 0.43 to 0.67 with a mean value of 0.53. Four well-defined clusters of IFI values were formed roughly at the same dissimilarity levels. The class Ⅰ was defined as suitable soil fertility (IFI 0.55–0.67), the class Ⅱ was defined as moderate soil fertility (IFI 0.43–0.53), the class Ⅲ was defined as poor soil fertility (IFI 0.35–0.39), and the class Ⅳ was defined as very poor soil fertility (0.24–0.29). The sum of class Ⅰ and the Ⅱ area accounted for 98.8% of the total coffee growing-regions, among them the class Ⅰ accounted for 54.2%, mainly distributed in Puer and Dehong districts. The class Ⅱ (44.6%) was mainly distributed in Lincang and Baoshan districts. The order of IFI values in different coffee growing-regions was as the following: Dehong (0.64) > Puer (0.58) > Lingcang (0.46) > Baoshan (0.43). The quality of green coffee bean was significantly correlated with soil nutrients(P < 0.05). The canonical correlation analysis indicated that the contents of caffeine and total sugar decreased with the increase of soil readily available potassium content and the fat content decreased with the increase of soil pH value and alkali-hydrolytic nitrogen content.

Conclusions Soil characteristics of Yunnan coffee growing regions are suitable for coffee growing. Soil readily available potassium, alkali-hydrolyzed nitrogen contents and pH have important effects on the quality of green coffee bean. Hence, an excessive or a defficient level may decrease quality of green coffee bean.