Long-lived plants resorb many of the nutrients in their leaves before they senesce, which often leaves the resulting litter relatively nutrient-poor. The nutrient content that remains influences how quickly it decomposes. Plant species vary in their efficiency of nutrient resorption, which affects both their own fitness and the biogeochemical fluxes driven by the decomposition of their litter. Measuring litter chemistry requires lab procedures that, for large numbers of samples, can be laborious and expensive. Reflectance spectroscopy holds the promise of allowing researchers to measure many chemical traits on leaf tissue quickly and with low marginal cost. Spectroscopic models of litter chemistry could make it easier to study plants’ influence on nutrient cycling.
Here, we test whether we can use reflectance spectroscopy to estimate five traits of leaf litter: leaf mass per area (LMA), nitrogen content (%N), carbon content (%C), neutral detergent fiber (NDF), and acid detergent fiber (ADF). The latter two traits describe the amount of tissue recalcitrant to a neutral (NDF) or a harsher, acidic (ADF) detergent, which correlates with litter recalcitrance to decomposition. We measured these traits and reflectance spectra of intact and ground leaf litter from more than 300 samples collected from broadleaf and coniferous trees in the Forests and Biodiversity (FAB) experiment at Cedar Creek Ecosystem Science Reserve (East Bethel, MN). We trained and validated partial least-squares regression (PLSR) models to predict each trait from the reflectance spectra.
Results/Conclusions
For all traits but LMA, models built using reflectance spectra of ground tissue were more accurate than those using intact tissue. The ground tissue models achieved high accuracy in predicting %N (observed vs. predicted calibration R2 > 0.95), ADF (R2 > 0.85), NDF (R2 > 0.8), and %C (R2 > 0.75). The intact tissue model produced accurate estimates (R2 > 0.8) of LMA and chemical traits on a per-area basis. Reflectance near the red edge and in the short-wave infrared range was especially important for predicting traits, especially elemental composition. We are developing models to estimate decomposition rates directly from reflectance spectra.
Leaves from different species of trees each senesce in their own way, producing and breaking down various pigments in rapid succession. Nevertheless, we show that reflectance spectroscopy can provide accurate estimates of leaf litter chemistry nearly in real time. This advance could simplify measurements of nutrient resorption efficiency and predictions of litter decomposition, which would contribute to our understanding of nutrient cycling in forests.