METHODS: We collected foliar samples (n = 1533) comprising 90 families, 198 genera and 495 plant species from ultramafic soils, further foliar samples (n = 177) comprising 45 families, 80 genera and 120 species from non-ultramafic soils and corresponding soil samples (n = 393 from ultramafic soils and n = 66 from non-ultramafic soils) from Kinabalu Park (Sabah, Malaysia). The data were geographically (Kinabalu Park) and edaphically (ultramafic soils) constrained. The inclusion of a relatively high proportion (approx. 14 %) of samples from hyperaccumulator species [with foliar concentrations of aluminium and nickel (Ni) >1000 μg g-1, cobalt, copper, chromium and zinc >300 μg g-1 or manganese (Mn) >10 mg g-1] allowed for hypothesis testing.
KEY RESULTS: Frequency distribution graphs for most elements [calcium (Ca), magnesium (Mg) and phosphorus (P)] were unimodal, although some were skewed left (Mg and Mn). The Ni frequency distribution was bimodal and the separation point for the two modes was between 250 and 850 μg g-1.
CONCLUSIONS: Accounting for statistical probability, the established empirical threshold value (>1000 μg g-1) remains appropriate. The two discrete modes for Ni indicate ecophysiologically distinct behaviour in plants growing in similar soils. This response is in contrast to Mn, which forms the tail of a continuous (approximately log-normal) distribution, suggestive of an extension of normal physiological processes.