Data for S. caseolaris were not included in ANOVA tests, since this would have
led to non-factorial design given the absence of medium saline results. However, there were
non-overlapping 95% confidence intervals for final mean shoot length between S. caseolaris
and the low salinity means of B. gymnorrhiza, B. sexangula and R. mucronata suggesting
significant differences for these contrasts (Fig. 2).