Fluorescence-based versus reflectance proximal sensing of nitrogen content in Paspalum vaginatum and Zoysia matrella turfgrasses

Newly developed non-destructive fluorescence-based indices were used to evaluate nitrogen (N) fertilization rates and leaf nitrogen content in Paspalum vaginatum and Zoysia matrella turfgrasses. They were defined by combinations of the red (RF) and far-red (FRF) chlorophyll fluorescence signals excited under ultraviolet (UV), green (G) or red (R) radiation, as follow: Flavonol index, FLAV = log(FRFR/FRFUV); Chlorophyll indices, CHL = FRFR/RFR and CHL1 = FRFG/FRFR; Nitrogen Balance Indices, NBI = FRFUV/RFR and . Measurements were performed in situ by using a portable optical sensor able to scan 1 m2 plots, with a 0.2 m resolution, under 6 different nitrogen rates, from 0 to 250 kg ha−1, with four replicates each. From the same plots, reflectance spectra were recorded and several reflectance-based indices calculated. Most of them, as well as the fluorescence-based indices of chlorophyll, CHL and CHL1, had a quadratic response to N rate with a flattening above 150 kg ha−1 and 100 kg ha−1 for P. vaginatum and Z. matrella, respectively. The fluorescence-based NBI1 index was the only one able to discriminate all the 6 N levels applied to both P. vaginatum and Z. matrella plots. This result is due to the character of NBI1 as a ratio between an index of chlorophyll and an index of flavonols that present opposite responses to N rates. The spatial heterogeneity within and between plots treated with different levels of N was well represented by the map of the NBI indices. When the NBI1 and NBI were regressed against leaf N content linear fits were obtained with high regression coefficients in both P. vaginatum (R2 = 0.85–0.87, RMSE = 0.23–0.24% N) and Z. matrella (R2 = 0.75–0.78, RMSE = 0.20–0.22% N). The best relationships between leaf N content and reflectance-based indices, found for R730/R1000 (R2 = 0.71, RMSE = 0.43% N) and MCARI (R2 = 0.80, RMSE = 0.22% N) for P. vaginatum and Z. matrella, respectively, were curvilinear and, therefore, less effective than NBI indices in the estimation of N. Nevertheless, a reflectance vegetation index suitable as proxy of leaf N common to both turf species was not found. Our results indicate the high potential of the fluorescence-based method and sensors for the in situ proximal sensing of N status in the management of N fertilization in turfgrass.