Scaling-up productivity (NPP) using light or water use efficiencies (LUE, WUE) from a two-layer tropical plantation

Net primary productivity (NPP) is a key driver of ecosystem C balance. Scaling NPP up to larger areas requires indirect methods: (a) for examble epsilon models based on light use efficiency (LUE = NPP/APAR, where APAR is the absorbed photosynthetically active radiation by green elements of canopy, or else models based on water-use-efficiency (WUE = NPP/E, where E = evapo-transpiration); (b) remote sensing tools to estimate the fraction of APAR (fAPAR) from vegetation indexes, or to estimate E. However, LUE and WUE are suspected to vary in space (edapho-climatic conditions, planting density) and time (seasonality, age), which needs to be documented before scaling up. Moreover, the application of this scaling approach to agroforestry systems with a stratified canopy may be difficult, since each layer contributes to the overall ecosystem light- and water-use efficiencies. The seasonal and inter-annual variabilities of LUE and WUE was assessed in a very simple bi-layer tropical coconut grove displaying minimum climatic and LAI variations, distinguishing the upper layer of coconuts, the herbaceous under-storey and the whole stand (subscripts C, H and S, respectively). We monitored NPP biometrically during 3 years above and below ground, together with microclimate and ES above the canopy (eddy-covariance), transpiration (T C) by sapflow, and fAPARC by LAI-2000 combined with canopy light absorption models. The partitioning of APAR, NPP and E was very close to the rule-of-thumb of canopy coverage by upper-layer (75%). Also the mean annual value of LUES (1.7 gDM MJ PARi −1 ) or mean WUES (3.7 gDM \( {\rm kg}_{{{\text{H}}_{2} {\text{O}}}}^{ - 1} \)) were mainly driven by the upper-layer of coconuts. However, the under-storey experienced around twice as much seasonal variations of NPP, E, LUE and WUE than the upper-storey. Given that NPPS varied by only 23% over the year, the high seasonal variations of WUES (240%) and LUES (250%) were mostly driven by the variations of APARS (230%) and were adjusted successfully using climate, age and density data, as a first step to estimate NPP on larger scales using climate, GIS and remote-sensing.