Improved Forward Models of Leaf Wax Hydrogen Isotope Ratios from a Synthesis of 3,500+ Measurements in Modern Plants

The hydrogen isotopic composition of plant waxes stored in sediment archives is widely used as a proxy for past hydroclimate. Apparent fractionation - the sum of the offsets between the δ2H of meteoric waters and plant waxes - varies widely due to environmental conditions and plant traits. However, these variations are poorly constrained and complicate inferences of past climate. Here, we evaluate phylogenetic, adaptational, and environmental controls on hydrogen isotope apparent fractionation using a new synthesis of published leaf wax measurements from living plants (3,680 individuals belonging to 855 species from 102 sources). We find that environmental measures of aridity and phylogenetic factors best explain variations in apparent fractionation. However, more traditional measures such as growth form or photosynthetic pathway approximate phylogenetic effects and remain useful in paleoclimate studies.

Apparent fractionation - quantified as the ɛ or enrichment factor of n-alkane waxes to source water δ2H - has a grand average and standard deviation of - 125 ± 38 ‰. In agreement with previous efforts, the larger discrimination of the graminoid growth form's ɛ of - 163 ± 38 ‰ is amplified by the Bambusoideae-Oryzoideae-Pooideae clade's ɛ of - 175 ± 32 ‰. However, graminoids outside of that clade still show a ~ 20 ‰ more negative ɛ than the dataset average. While the C4 photosynthetic pathway has a stronger negative discrimination by ~ 25 ‰ than the C3 or CAM pathways, the C4 pathway is dominated by graminoids, and we find no significant difference between C3 and C4 graminoids. These phylogenetic effects appear to solely influence biosynthetic fractionation, as there are no similar patterns of isotopic enrichment of plant waters relative to environmental waters. When forward modelling leaf waxes, a fixed ɛwax-xylem yields model root-mean-square-error of 18 ‰ whereas addition of either phylogenetic- or modeled leaf water enrichment due to low relative humidity improves model error to 15 ‰ and addition of both further improves model error to 12 ‰.