Abstract
In the search for a better method of estimating the light-use efficiency (LUE) of evergreen tropical rainforests, we employed remotely sensed spectral vegetation indices (VIs) to monitor both CO2 flux and canopy spectral reflectance over 3years in a lowland dipterocarp forest in Peninsular Malaysia. We investigated the sensitivity of five VIs calculated from spectral reflectance: the photochemical reflectance index (PRI), the canopy chlorophyll index (CCI), the normalized difference vegetation index (NDVI), the enhanced vegetation index (EVI) and the water index (WI).
During the monitoring period, clear seasonal variations were not found in LUE, the observed VIs or the phenological timing (particularly new leaf flush) of dominant dipterocarp trees. Although leaf phenology tended to correlate with variations in the CCI, the highest correlation coefficient among the relationships between LUE and the VIs was observed in PRI (R=0.341, n=699). Among the relationships between LUE and meteorological factors, the strongest correlation was found between LUE and vapor pressure deficit (VPD; R=−0.580). These results suggest that unseasonal variation in LUE would be more affected by water conditions than leaf phenology or green leaf mass, and that the PRI has lower sensitivity for direct estimation of LUE compared to VPD in this evergreen tropical rainforest.
To improve the accuracy in estimating LUE, we examined the potential of combinational use of VIs and meteorological factors. Variable selection by stepwise multiple regression showed that the best variable combination for LUE estimation was the PRI and VPD (R=0.612). The relative root mean square error (rRMSE) in the simple regression models using PRI, VPD and PRI×VPD, and the multiple regression model using PRI and VPD, was 22.5%, 19.4%, 19.0% and 18.7%, respectively. Based on these results, we concluded that (1) the estimation method solely based on the PRI as in the case of other temperate deciduous forests is not suitable in the tropical evergreen rainforest, and (2) the combinational use of the PRI and VPD offers one of the better models for estimating LUE in tropical evergreen rainforests.
•The utility of VI for estimating LUE was studied in the evergreen tropical forest.•The PRI showed higher R2 in the relationship with LUE than leaf mass-related VIs.•But the explanation ability by the PRI is lower than VPD and Tair.•The combinational use of VPD and PRI can improve the estimation accuracy of LUE.•The PRI should be used as a complemental parameter in LUE model at tropical forest.