control) and years were highly significant (treatment: df = 1; χ 2 = 22.18, P < 0.0001; year: df = 2; χ 2 = 824.73, P < 0.0001), while the interaction (treatment × year) was not significantly different (df = 2, χ 2 = 4.03, P = 0.1332).
GLM analyses of the amounts of glycosylated isoflavones and malonyl glycosylated isoflavones showed that treatment (df = 1, χ 2 = 11.26, P = 0.0008) and malonylation (glycoside or malonyl glycoside) (df = 1, χ 2 = 81.15, P < 0.0001) were both significantly different and that the interaction (treatment × malonylation: df = 1, χ 2 = 3.73, P = 0.0534) was marginally significant.