Gap formation and carbon cycling in the Brazilian Amazon: measurement using high-resolution optical remote sensing and studies in large forest plots

Background: The dynamics of gaps plays a role in the regimes of tree mortality, production of coarse woody debris (CWD) and the variability of light in the forest understory. Aims: To quantify the area affected by, and the carbon fluxes associated with, natural gap-phase disturbances in a tropical lowland evergreen rain forest by use of ground measurements and high-resolution satellite images. Methods: We surveyed two large forest inventory plots of 114 and 53 ha of the Tapajos National Forest (TNF) in the Brazilian Amazon during 2008 and 2009, respectively. We mapped all gaps and collected data on light availability, CWD stocks and tree mortality in the field. Gap location, canopy openness (CO) and leaf area index (LAI) estimated in the field were compared with two IKONOS-2 high-resolution satellite images acquired at approximately the time of the field measurements. Results: In the two large plots (167 ha total area) we found 96 gaps. The gaps represented 1.42% of the total area and gaps <1-year-old accounted for 0.81% of the plot area. In TNF, the production of CWD in recent gaps was 0.76 Mg C ha(-1) year(-1) and the mean tree mortality was 2.38 stems ha(-1) year(-1). The area of gaps estimated using thresholds of light intensity measured by remote sensing optical instruments was twice as large as the gap areas measured on the ground. We found no significant correlation between spectral remote sensing images and CO or LAI, probably due to the high degree of shadow in the high-resolution satellite images. Conclusions: We present the first statistics of CWD production based on gap size in the tropical forest literature. Tree mortality and CWD flux and the forest floor light environment were closely related to gap area. However, less than 30% of the annual tree mortality and CWD flux was associated with gaps, and gaps were difficult to detect using remote sensing methods because of the high proportion of shadow in the images. These results highlight the need for permanent plots in long-term carbon studies.