Hyperspectral reflectance as a tool to measure biochemical and physiological traits in wheat

Abstract

Leaf hyperspectral reflectance can be used by the wheat physiology and breeding communities to rapidly estimate Rubisco activity, electron transport rate, leaf nitrogen, leaf dry mass per area, and relative chlorophyll content.

Predicting dark respiration rates of wheat leaves from hyperspectral reflectance

“…Our ability to predict R dark might be an indirect reflection of photosynthesis. Considering that the light saturated ambient rate of photosynthesis and the two major determinants of photosynthetic performance— V c,max and J max —can also be predicted from leaf reflectance (Ainsworth et al, ; Barnes et al, ; Dechant et al, ; Doughty et al, ; Heckmann, Schlüter, & Weber, ; Serbin et al, ; Silva‐Pérez et al, ; Yendrek et al, ), one possibility is that variations in R dark are coupled to variations in V c,max and/or J max and that the ability to predict R dark from leaf reflectance is, in part, due to spectral signatures of key photosynthetic components. Dechant et al () reported that the prediction of V c,max 25 from leaf reflectance is a secondary one, driven primarily by the prediction of leaf N. However, because the prediction of R dark here for wheat using N area , LMA, or their combination was poor (for R dark_LA , highest r 2 = 0.12) compared with the PLSR model (see Table S6 for multiple regression results for R dark_LA ), our success in predicting R dark indicates that there is additional information contained within the reflectance spectra associated with R dark.…”

“…Poor model performance across experiments is not uncommon. Silva‐Pérez et al () reported that models derived from field‐grown aspen leaves ( Populus tremuloides Michx. ; Serbin et al, ) gave poor predictions when applied to wheat leaves.…”

“…The device was fitted with an ASD fibre optic cable and leaf clip. A mask attached to the leaf clip reduced the width of the aperture through which leaf reflectance was recorded to 11.5 mm, enabling easier measurement of leaf widths down to 12 mm (Silva‐Pérez et al, ). Leaf spectral reflectance was captured between 1000 and 1400 hr from the adaxial surface and close to the midpoint of the leaf.…”

“…Prior to data analysis, a multiplicative correction module (ASD Spectral Analysis and Management System [SAMS®] version 3.2) was applied to the reflectance data at 1000 and 1800 nm to correct for “jumps” observed in apparent reflectance at the intersections between different detector ranges. As did Silva‐Pérez et al (), reflectance spectra with values greater than 0.7 between 800 and 1000 nm were treated as an outlier and removed.…”

“…Variation in R dark 25 among genotypes and environments is predictable from other leaf traits such as N concentration or content, LMA, and the carboxylation capacity of Rubisco at 25°C ( V c,max 25 ; Atkin et al, ; Reich, Walters, Ellsworth, et al, ; Reich, Walters, Tjoelker, Vanderklein, & Buschena, ; Ryan, ). Both N and LMA can be predicted from hyperspectral reflectance data (Ecarnot et al, ; Serbin et al, ; Silva‐Pérez et al, ). It is also possible to predict V c,max 25 , but with lower accuracy and precision (Ainsworth et al, ; Dechant, Cuntz, Vohland, Schulz, & Doktor, ; Doughty, Asner, & Martin, ; Serbin et al, ; Silva‐Pérez et al, ).…”