<p>&#160;</p><p>Long-standing ecological theory establishes that the isotopic composition of the plant water reflects that of the root-accessed sources, at least in non-saline or non-xeric environments. However, a growing number of studies challenge this assumption by reporting plant-source offsets in water isotopic composition, for a wide range of ecosystems. We conducted a global meta-analysis to systematically quantify the magnitude of this plant-source offset in water isotopic composition and its potential explanatory factors. We compiled 108 studies reporting dual water isotopic composition (&#948;<sup>2</sup>H and &#948;<sup>18</sup>O) of plant and source water. From these studies, we extracted the &#948;<sup>2</sup>H and &#948;<sup>18</sup>O of both plant and source waters for 223 plant species from artic to tropical biomes. For each species and sampling campaign, within each study, we calculated the mean line conditioned excess (LC-excess), with the slope and intercept of the local meteoric water line, and the mean soil water line conditioned excess (SWL-excess), from the slope and intercept of the soil water evaporation line. For each study site and sampling campaign, we obtained land surface temperature and volumetric soil water from the ERA5 database. For each study species, we recorded the functional type, leaf habit and for those available wood density. We found, on average, a significantly negative SWL-excess: plant water was systematically more depleted in &#948;<sup>2</sup>H than soil water. In > 90% of the cases with significantly negative SWL-excess, we also found negative LC-excess values, meaning that access to sources alternative to soil water was unlikely to explain negative SWL-excess values.&#160;</p><p>Calculated SWL-excess was affected by temperature and humidity: there were larger mismatches between plant and source water in isotopic composition in colder and wetter sites. Angiosperms, broadleaved and deciduous species exhibited more negative SWL-excess values than gymnosperms, narrow-leaved and evergreen species. Our results suggest that when using the dual isotopic approach, potential biases in the adscription of plant water sources are more likely in broadleaved forests in humid, and cold regions. Potential underlying mechanism for these isotopic mismatches will be discussed.</p><p>&#160;</p>
Technical note: Evaporating water is different from bulk soil water in <i>δ</i><sup>2</sup>H and <i>δ</i><sup>18</sup>O and has implications for evaporation calculation
