High Throughput and Precision Phenotyping
The efficient exploration of genetic resources -for crop improvement and to identify genetic and mechanistic basesrequires a combination of precision and high throughput phenotyping approaches. Non-intrusive high spatial resolution spectral imagery can be applied to the monitoring of physiological characteristics such as canopy temperature, hydration status, and pigment composition, as well as permitting estimates of agronomic traits such as biomass and yield. Spaceborne remote sensing platforms have proven efficient at measuring some of these characteristics at a field scale, however their spatial resolution proves too low for accurate data retrieval at plot level in a plant breeding context. While ground-based remote sensing is used for predicting physiological and agronomic traits at a plot scale, temporal variations of environmental variables, such as air temperature, can introduce confounding factors, especially when applied to large trials. Low-level airborne remote sensing platforms overcome these restrictions, allowing for fast, non-destructive screening of plant physiological properties over large areas, with enough resolution to obtain information at plot level while being able to measure thousands genetic resources in the field with one take.
Rapid advances in remote sensing technologies, data processing and availability of instruments has made it easier to implement remote sensing techniques to research numerous plant properties (e.g., Leinonen and Jones 2004; Jones et al. 2007; Berni et al. 2009). The increase in demand for large scale vegetation monitoring means that there is a move to such remote sensing applications in which simultaneous measurements of greater target areas can readily be made. In a recent study, Tattaris et al. (2013) applied a low level airborne remote sensing platform to derive indices relating to plant properties such as canopy temperature, water status, and pigment composition. Sampling was performed with a multispectral camera and thermal camera mounted on an eight rotor unmanned aerial vehicle (UAV) and helium filled tethered blimp. Airborne indices were validated by equivalent indices collected at ground level. These ground-based measurements have already been proven to be linked with yield and biomass (e.g., Reynolds et al. 1994; Aparicio et al. 2000). This strong agreement between methods acts to validate the use of the airborne indices. In addition, significant genetic correlations were found between the airborne indices derived using imagery and yield/biomass, larger than the equivalent correlations between the ground-based measurements and yield/biomass. This is probably because airborne approaches reduce error in two ways. Firstly, simultaneous measurement permitted through airborne platforms avoid the confounding effects associated with environmental drift when measuring plots one at a time with ground-based instruments. Secondly, aerial imaging techniques permit data smoothing through elimination of outlier pixels. These factors give additional confidence to the use of such methodologies in screening genetic resources and breeding progeny at a large scale where thousands of plots are involved.