A multiscale approach to investigate fluorescence and NDVI imaging as proxy of photosynthetic traits in wheat

With the development of the digital phenotyping, repeated measurements of agronomic traits over time are easily accessible, notably for morphological and phenological traits. However high throughput methods for estimating physiological traits such as photosynthesis are lacking. This study demonstrates the links of fluorescence and reflectance imaging with photosynthetic traits. Two wheat cultivars were grown in pots in a controlled environment. Photosynthesis was characterised by gas-exchange and biochemical analysis at five time points, from booting to 21 days post anthesis. On the same days imaging was performed on the same pots, at leaf and plant scale, using indoor and outdoor phenotyping platforms, respectively. Five image variables (Fv/Fm and NDVI at the whole plant level and Fv/Fm, Φ(II)532 and Φ(NPQ)1077 at the leaf scale) were compared to variables from A-Ci and A-Par curves, biochemical analysis, and fluorescence instruments. The results suggested that the image variables are robust estimators of photosynthetic traits, as long as senescence is driving the variability. Despite contrasting cultivar behaviour, linear regression models which account for the cultivar and the interaction effects, further improved the modelling of photosynthesis indicators. Finally, the results highlight the challenge of discriminating functional to cosmetic stay green genotypes using digital imaging.