An ecosystem approach to understanding soil C in humid tropical forests: influences of minerals, microbes, and roots

Background/Question/Methods

Hans Jenny posed the “organic-matter problem of the tropics” as a question of explaining how mineral soils retain carbon (C) in spite of potentially rapid decomposition rates in humid tropical ecosystems. Recent work emphasized the importance of poorly-crystalline minerals in determining soil C storage, but their relative importance in comparison with other ecosystem variables has seldom been tested. Here, we examined the relative impact of mineralogical, physical, and biological drivers of soil C concentrations and stocks across a gradient of montane tropical forest ecosystems differing in precipitation and soil oxygen availability in the Luquillo Experimental Forest, Puerto Rico. We collected 150 soil samples across this gradient and analyzed soil C concentrations and predictor variables: hydrolytic enzymes, nitrogen (N) stable isotopes, iron and aluminum species, a redox indicator (reduced iron concentrations), live and dead fine root biomass, root and soil C/N ratios, and texture. We analyzed relationships among soil C and predictor variables using mixed effects and path analysis models.

Results/Conclusions

Activity of the extracellular enzyme N-acetyl glucosaminidase (NAGase), an index of microbial N acquisition from chitin and peptidoglycan, provided the best correlate of soil C concentrations. The activity of NAGase normalized by soil C displayed a strong power law relationship that explained 80% of the variance in soil C concentrations among samples, which varied between 2 – 20 % C. Relative declines in NAGase activity (normalized by soil C) were associated with increased C concentrations, although NAGase displayed an overall positive relationship with C. Soil d¹⁵N declined from approximately 6 ‰ to 1 ‰ across the O₂ gradient, indicative of lower rates of N cycling. Whether relative declines in NAGase and d¹⁵N and increased soil C/N reflect a cause or a consequence of C accumulation is uncertain. Our optimal model of soil C concentrations included NAGase, soil C/N, poorly-crystalline aluminum, a redox indicator, live (but not dead) fine root biomass, root C/N, and crystalline and poorly-crystalline iron (Fe). Notably, Fe correlated negatively with C, suggesting that Fe oxides may not enhance C protection in these fluctuating redox soils. Together, our data provide an ecosystem context for understanding impacts of physical, biological, and geochemical factors on soil C in humid tropical forests.