A comparison of methods for detecting association between plant species

The type and significance of association between twenty-three pairs of species were detected in nine plant associations in arid and semiarid rangelands of Yazd, Iran. Six methods including L12 function, Pielou's, Dixon's and Goodall's tests, Chi square test using presence-absence data from circular quadrat and circular density-sized quadrat and four measures including Ochiai, Dice, Jaccard and Cole indices were compared to determine the type and significance of association. If a significant positive or negative association is likely to occur between two species with approximately similar density, or a significant negative association may occur between an infrequent and frequent species, the Pielou's and the Dixon's tests can provide efficient and approximately similar results and can be used with L12 function to detect the scales at which the significant association may occur. If a significant positive association is likely to occur between two species and the density of one of the species is much lower than the other, i.e., most of the individuals of an infrequent species occur in close association with a frequent species, Chi-square test and Ochiai and Dice indices using presence- absence data can be used to detect the type and significance of association. If a non-significant association exists between two species, the results of a combination of the tests may well complement each other. •The association between arid plant species was determined using a variety of methods.•Estimates of the methods were influenced by plant density and degree of association.•Pielou's and Dixon's tests were the best for the species with similar densities.•Ochiai and Dice indices were the best for species with significantly different density.•For non-significant associations, a combination of tests may well complement each other.