Abstract

Precipitation is an important control of the nitrogen cycle and known to affect the isotopic nitrogen signal (δ15N) of total soil N and leaves. In general, soil and foliar δ15N decreases with increasing mean annual rainfall across diverse ecosystems. Such changes in the N cycle would be important to detect in paleoecological records, especially in arid regions where rainfall is a limiting factor for biological activity. Here, we present a modern analog study of the δ15N signal along an environmental gradient in the Atacama Desert of northern Chile. We performed vegetation and soils surveys along a 2000 m altitudinal gradient in the western Andean slope adjacent to the Atacama Desert. Rainfall increases by an order of magnitude along this gradient. We sampled 22 sites, starting at 4500 m (mean annual precipitation (MAP) = 160 mm/yr) and ending at 2500 m (MAP < 10 mm/yr). We analyzed the N content and δ15N of 28 soil samples and foliar δ15N of 50 specimens from widespread AndeanAtacama species (Baccharis tola, Jarava frigida, Opuntia camachoi, Parastrephia quadrangularis and Atriplex imbricata). We also compared the signal of two different seasons (after and before the rainfall season in summer). Our results show no relation between total soil N (mg/ kg) and MAP (R2 = 0.03, p>0.05). In contrast, there is a significant negative correlation between mean soil δ15N and MAP (R2 = 0.77, p < 0.01). This soil δ15N signal is persistent across the wet and dry seasons. Although foliar δ15N for individual species shows no correlation with MAP, mean foliar δ15N per site shows a significant positive correlation (R2 = 0.52, p < 0.01). Thus despite the elevated intra-site variation observed for the foliar δ15N values, a large part of this variation can be explained by which species and season were analyzed. Before we consider the paleoclimatic potential of foliar δ15N as a precipitation proxy in the Atacama, two aspects need to be addressed. First, mean soil δ15N is correlated with MAP as expected, but other processes could change the isotopic soil signal (e.g. age and depth) and should be considered before using it as a precipitation proxy. Second, mean foliar δ15N per site shows an important correlation with precipitation but individual species tend to be much more variable. Finally, the challenge remains of finding suitable integrators of foliar δ15N in the paleoecological record. Fossil herbivore dung and urine δ15N preserved in caves and in rodent middens hold much promise in this regard.