The oxygen isotopic composition of precipitation (δ¹⁸O_prec) is well known to be a valuable (paleo-)climate proxy. Paleosols and sediments and hemicelluloses therein have the potential to serve as archives recording the isotopic composition of paleoprecipitation. In a companion paper (Zech et al., 2014) we investigated δ¹⁸O_{hemicellulose} values of plants grown under different climatic conditions in a climate chamber experiment. Here we present results of compound-specific δ¹⁸O analyses of arabinose, fucose and xylose extracted from modern topsoils (n=56) along a large humid-arid climate transect in Argentina in order to answer the question whether hemicellulose biomarkers in soils reflect δ¹⁸O_prec. The results from the field replications indicate that the homogeneity of topsoils with regard to δ¹⁸O_{hemicellulose} is very high for most of the 20 sampling sites. Standard deviations for the field replications are 1.5‰, 2.2‰ and 1.7‰, for arabinose, fucose and xylose, respectively. Furthermore, all three hemicellulose biomarkers reveal systematic and similar trends along the climate gradient. However, the δ¹⁸O_{hemicellulose} values (mean of the three sugars) do not correlate positively with δ¹⁸O_prec (r=-0.54, p<0.014, n=20). By using a Péclet-modified Craig-Gordon (PMCG) model it can be shown that the δ¹⁸O_{hemicellulose} values correlate highly significantly with modeled δ¹⁸O_{leaf water} values (r=0.81, p<0.001, n=20). This finding suggests that hemicellulose biomarkers in (paleo-)soils do not simply reflect δ¹⁸O_prec but rather δ¹⁸O_prec altered by evaporative ¹⁸O enrichment of leaf water due to evapotranspiration. According to the modeling results, evaporative ¹⁸O enrichment of leaf water is relatively low (~10‰) in the humid northern part of the Argentinian transect and much higher (up to 19‰) in the arid middle and southern part of the transect. Model sensitivity tests corroborate that changes in relative air humidity exert a dominant control on evaporative ¹⁸O enrichment of leaf water and thus δ¹⁸O_{hemicellulose}, whereas the effect of temperature changes is of minor importance. While oxygen exchange and degradation effects seem to be negligible, further factors needing consideration when interpreting δ¹⁸O_{hemicellulose} values obtained from (paleo-)soils are evaporative ¹⁸O enrichment of soil water, seasonality effects, wind effects and in case of abundant stem/root-derived organic matter input a partial loss of the evaporative ¹⁸O enrichment of leaf water. Overall, our results prove that compound-specific δ¹⁸O analyses of hemicellulose biomarkers in soils and sediments are a promising tool for paleoclimate research. However, disentangling the two major factors influencing δ¹⁸O_{hemicellulose}, namely δ¹⁸O_prec and relative air humidity controlled evaporative ¹⁸O enrichment of leaf water, is challenging based on δ¹⁸O analyses alone.