REPLYING TO S. R. Saleska et al. Nature 531, 10.1038/nature16457 (2016)

Multiple mechanisms could lead to upregulation of dry-season photosynthesis in Amazon forests, including canopy phenology and illumination geometry. We specifically tested two mechanisms for phenology-driven changes in Amazon forests during dry-season months, and the combined evidence from passive optical and lidar satellite data1 was incompatible with large net changes in canopy leaf area or leaf reflectance suggested by previous studies2,3,4,5. We therefore hypothesized1 that seasonal changes in the fraction of sunlit and shaded canopies, one aspect of bidirectional reflectance effects in Moderate Resolution Imaging Spectroradiometer (MODIS) data, could alter light availability for dry-season photosynthesis and the photosynthetic capacity of Amazon forests without large net changes in canopy composition. Subsequent work supports the hypothesis that seasonal changes in illumination geometry and diffuse light regulate light saturation in Amazon forests6,7. These studies clarify the physical mechanisms that govern light availability in Amazon forests from seasonal variability in direct and diffuse illumination. Previously, in the debate over light limitation of Amazon forest productivity, seasonal changes in the distribution of light within complex Amazon forest canopies were confounded with dry-season increases in total incoming photosynthetically active radiation2,3,8. In the accompanying Comment9, Saleska et al. do not fully account for this confounding effect of forest structure on photosynthetic capacity.

Saleska et al.9 investigated one of the three lines of evidence in our paper to argue that near-zero seasonal changes in corrected MODIS enhanced vegetation index (EVI) are actually non-zero (figure 1 in ref. 9; 0.071 to 0.016, a 77% reduction). Following this logic, our data also show a small but statistically significant decrease in normalized difference vegetation index (NDVI; extended data figure 4 in ref. 1), a pattern that we attributed to residual artefacts from changes in sun-sensor geometry, as no leaf-level mechanism for increased forest productivity generates opposing responses in these vegetation indices (see supplementary discussion in ref. 1). Indeed, the comparison between NDVI and EVI responses is a useful diagnostic tool1 that could have been used to investigate residual bidirectional reflectance effects in multiangle implementation of atmospheric correction (MAIAC) data (figure 2 in ref. 9).

In isolation, MODIS data provide limited insight into the mechanisms for seasonal changes in Amazon forests9. MODIS EVI is primarily sensitive to changes in near-infrared reflectance1,4,10, not photosynthetically active radiation absorption that drives forest productivity. Saleska et al. misrepresent data from extended data figure 7 of ref. 1 as fully corrected in their figure 2 (ref. 9), and further confound seasonal changes through spatial averaging of 1 km2 data over large regions (25–121 km2). A previous study using 1 km2 data for these same tower sites shows little or no seasonality in MAIAC EVI11 (see supplementary figure 5 in ref. 1).

One of the key messages from our study was the need for careful attention to uncertainty in satellite-based measurements of forest seasonality. The presentation of in situ and satellite data by Saleska et al. (figure 2a in ref. 9), and the MAIAC product in general, could be improved with quantitative estimates of uncertainty to support assertions of forest seasonality.

Subtle variability in canopy structure and reflectance properties of Amazon forests remains a key area for further study, particularly with large-scale field studies12, to better understand the spatial and temporal heterogeneity of leaf phenology strategies in Amazonia13. Other mechanisms for seasonal changes in photosynthetic capacity also merit further investigation, including how diurnal and seasonal variability in illumination alter the distribution of photosynthetically active radiation at the leaf level1,6,7,14. NASA satellite data remain an important foundation for future research on tropical forest dynamics, within the limits of calibration, measurement, and model uncertainty that can be realistically achieved with space-based sensors.