Changes in Soil Phosphorus Fractions and Ecosystem Dynamics across a Long Chronosequence in Hawaii

Abstract

We tested the Walker and Syers (1976) conceptual model of soil development and its ecological implications by analyzing changes in soil P, vegetation, and other ecosystem properties on a soil chronosequence with six sites ranging in age from 300 yr to 4.1 x 106 yr. Climate, dominant vegetation, slope, and parent material of all of the sites were similar. As fractions of total P, the various pools of soil phosphorus behaved very much as predicted by Walker and Syers. HCI—extractable P (presumably primary mineral phosphates) comprised 82% of total P at the 300—yr—old site, and then decreased to 1% at the 20,000—yr—old site. Organic phosphorus increased from the youngest site to a maximum at the 150 000 yr site, and then declined to the 4.1 x 106 yr site. Occluded (residual) P increased steadily with soil age. In contrast to the Walker and Syers model, we found the highest total P at the 150 000—yr—old site, rather than at the onset of soil development, and we found that the non—occluded, inorganic P fraction persisted through to the oldest chronosequence site. Total soil N and C increased substantially from early to middle soil development in parallel with organic P, and then declined through to the oldest site. Readily available soil P, NH4+, and NO3— were measured using anion and cation exchange resin bags. P availability increased and decreased unimodally across the chronosequence. NH4+ and NO3— pools increased through early soil development, but did not systematically decline late in soil development. In situ decomposition rates of Metrosideros polymorpha litter were highest at two intermediate—aged sites with soil fertility (20 000 yr and 150 000 yr), and lowest at the less—fertile beginning (300 yr) and endpoint (4.1 x 106 yr) of the chronosequence. M. polymorpha leaves collected from these same four sites, and decomposed in a common site, suggested that leaves from intermediate—aged sites were inherently more decomposable than those from the youngest and oldest sites. Both litter tissue quality and the soil environment appeared to influence rates of decomposition directly. The highest mean soil N2O emissions (809 mg°m—2°d—1) were measured at the 20 000—yr—old site, which also had the highest soil nitrogen fertility status. Plant communities at all six chronosequence sites were dominated primarily by M. Polymorpha, and to a lesser extent by several other genera of trees and shrubs. There were, however, differences in overall vegetation community composition among the sites. Using a detrended correspondence analysis (DECORANA), we found that a high proportion of species variance among the sites (eigenvalue = 0.71) can be explained by site age and thus soil developmental stage. Overall, long—term soil development across the chronosequence largely coincides with the conceptual model of Walker and Syers (1976). How P is distributed among different organic and inorganic fractions at a given stage of soil development provides a useful context of evaluating contemporary cycling of P and other nutrients, and for determining how changes in P availability might affect diverse ecosystem processes.