• co-limitation;
  • ecosystem decline;
  • ecosystem progression;
  • ecosystem retrogression;
  • pedogenesis;
  • plant–soil (below-ground) interactions;
  • productivity;
  • soil age;
  • stoichiometry;
  • succession


1. The classical model of long-term ecosystem development suggests that primary productivity is limited by nitrogen (N) on young substrates and phosphorus (P) on older substrates. Measurements of foliar and soil nutrients along soil chronosequences support this model, but direct tests through nutrient-addition experiments are rare.

2. We conducted a nutrient-limitation bioassay using phytometer species grown in soils from five stages of a >2-million-year dune chronosequence in south-western Australia. This long-term chronosequence is located within a region of exceptionally high plant species diversity and has not been previously studied in the context of ecosystem development.

3. Growth of unfertilized phytometers, a proxy for primary productivity, peaked on young soils (hundreds to a few thousand years) and then declined steadily on older soils. This decline was linked to P limitation, and its rapid appearance (<7000 years) compared to other sequences reflects the low P concentration in the parent material. As predicted, growth of canola was N-limited on the youngest soil (stage 1), co-limited by multiple nutrients in stage 2 and increasingly P-limited thereafter.

4. Growth of wheat was P-limited from stage 2 onwards, yet on the youngest soil it was co-limited by potassium (K) and micronutrients – most likely iron (Fe). Nitrogen addition also decreased the root:shoot ratio of wheat such that shoot growth was higher than in the control. We attribute these responses to a parent material that is very low in K and N and strongly alkaline (pH [H2O] > 9), being of a marine origin (i.e. carbonate dunes). Fe is poorly soluble at high pH and K likely plays a role in the secretion of Fe-mobilizing exudates from wheat roots.

5.Synthesis. Our results provide strong support for the long-term ecosystem-development model, particularly with regard to the appearance of P limitation and associated declines in productivity. However, our study also shows that N cannot be assumed to invariably be the most important limiting nutrient in young soils, and it is unlikely to be the only limiting nutrient in calcareous soils. This south-western Australian long-term chronosequence provides an excellent opportunity to explore edaphic controls over plant species diversity.