Plant-soil feedbacks on free-living nitrogen fixation over geological time.

Free-living heterotrophic nitrogen fixation (FNF) is a widespread nitrogen input pathway in terrestrial ecosystems. However, questions remain over the relative influence of co-occurring controls on patterns of heterotrophic FNF activity, especially across generalized stages of primary succession, from biomass accumulation to retrogressive phases. Here, we experimentally test two alternative hypotheses regarding FNF rates during ecosystem development: (H1) site (i.e., changes in soil fertility during succession) is the primary driver of leaf-litter FNF rates, vs. (H2) leaf-litter chemistry is the primary determinant of FNF activity across a broad range of ecosystem conditions. We evaluated these hypotheses across a well-studied soil chronosequence in California (i.e., the Ecological Staircase), which spans ~1 million years of ecosystem development and displays extreme ranges in plant-soil nutrient conditions, culminating in the nutrient depleted and stunted Pygmy forest. Across this successional gradient, we implemented a reciprocal leaf-litter transplant and a common garden litter bag decomposition experiment with senesced needles of Pinus muricata. Our results support H1: rates of FNF were similar for all leaf-litter types decomposed at the same site regardless of initial leaf-litter C and nutrient contents. FNF rates sharply declined from the maximal to retrogressive stage of succession. Trends in P dynamics during decomposition suggest an important role of P in regulating FNF. For example, P. muricata litter collected from the infertile Pygmy site displayed substantially higher FNF rates when decomposed at the fertile site, in part by immobilizing significant quantities of P from the soil at the fertile site. Conversely, P. muricata litter collected from the fertile site decomposed more slowly at the Pygmy site, with concomitant declines in FNF rates that matched those of Pygmy site litter decomposed in situ. These results are consistent with the idea that, over millennia, long-term declines in P availability feedback to constrain FNF rates, in part explaining the emergence of extremely nutrient-poor and retrograded ecosystems.