Phenotyping in the fields: dissecting the genetics of quantitative traits and digital farming.

Abstract

‘… development of novel approaches for studying traits like nutrient, water and radiation-use efficiency must be developed under field conditions …’ Crop production has to substantially increase within the next decades, following an already long history of changing plant traits to produce desired characteristics. Thus, plant performance has been substantially influenced during domestication and post-domestication selection, including modern breeding. In the group of Roberto Papa (University of Ancona, Italy) the complementing use of genomic analysis, molecular phenotyping, and high-throughput phenotyping of wild and domesticated common bean accessions revealed a tremendous impact of specific trait selection during domestication on genetic diversity and phenotypic expression (Bitocchi et al., 2013; Bellucci et al., 2014), which is known as ‘domestication syndrome’. Due to domestication and post-domestication processes the whole architecture of phenotypic expression showed a decrease of variation in plasticity in wheat indicating the need for the opportunity to use wild germplasm for further crop improvements. Particularly, in addition to shoot properties, optimizing root growth and architecture for enhanced resource use and the development of realistic whole plant models is important for quantitative phenotyping in the laboratory and field (Postma et al., 2014). According to Christian S. Jensen (DLF-TRIFOLIUM, Store Heddinge, Denmark), specific agronomic traits such as seed yield, disease resistance, stem length, and heading are nevertheless continuously in the focus, because these are of outmost importance for farmers. The large payoff of these measurements justifies enduring effort for improving these traits. To widen the spectrum of relevant traits under field conditions, recent advances in the development and application of novel noninvasive sensors (Fiorani et al., 2012; Araus & Cairns, 2014) could significantly contribute by providing higher precision in screening or by decreasing screening and selection efforts. However, these promising methodologies still need considerable improvement to become valuable tools to support breeding programmes, by identifying novel and relevant traits, establishing robust sensors, close monitoring of the environment linked to the development of predictive models. High-throughput methods of genomics and molecular breeding often focus on selection based on genotypic information but these approaches require phenotypic data as pointed out by Xavier Sirault (CSIRO, Canberra, ACT, Australia). Complex interactions between the crop genotype, environmental factors in combination with plant population dynamics and crop management greatly affect plant phenotypes in field experiments (Porter & Christensen, 2013). Hence, development of novel approaches for studying traits like nutrient, water and radiation-use efficiency must be developed under field conditions. Novel techniques should be kept cost-effective and robust under varying field conditions and should allow for the monitoring of various and complex traits (Andrade-Sanchez et al., 2014). Furthermore, interdisciplinary efforts that include for example genomics and deep plant physiology should be used as the basis for validation of proximal (noninvasive) field phenotyping approaches and to reliably capture traits (Brown et al., 2014). To facilitate phenotypic evaluations high-throughput phenotyping methods and infrastructure is crucial for current and future field-based applications, which needs to be complemented by deep phenotyping at the individual plant level under controlled conditions. Additionally, effective data acquisition and translation of data into results useful for agriculture have to be provided. As one important approach for high-throughput field phenotyping, Pablo J. Zarco-Tejada (IAS-CSIC, Córdoba, Spain) introduced the use of unmanned aerial vehicles as platforms that can be equipped with various imaging devices such as red-green-blue digital cameras, hyperspectral or thermal cameras. The use of hyperspectral sensors facilitates the combined acquisition and evaluation of narrow-bands which also allows for the quantification of physiological parameters such as the emitted fluorescence (Damm et al., 2014) or spectral indices related to biomass or pigment composition (Zarco-Tejada et al., 2013). Monitoring the canopy temperature allows the calculation of indicators of water stress (Bellvert et al., 2014), while high resolution images were used for three-dimensional (3D) reconstruction and height determination of trees (Zarco-Tejada et al., 2014). Such devices have been successfully deployed in phenotyping campaigns to various crops ranging from trees to cereals to relate to the physiological status of the crops under stress conditions. Increasing world population and global climatic change are major challenges for agriculture and breeding in future (Tester & Langridge, 2010; Fiorani & Schurr, 2013). Additionally, due to legal constraints such as for the use of classic pesticides, alternative strategies are needed to maintain food safety and security. Lénaïc Grignard (GEOSYS, Toulouse, France) presented a vision of digital farming as a novel technological approach for precision agriculture: real-time data acquisition in the field where data evaluation as decision support can feed into optimized crop management. Field phenotyping approaches could significantly contribute to facilitate these parameters of digital farming by adding many spatiotemporal layers of dynamic information on plant phenotypic behaviour in agro-ecosystems. Thomas Roitsch (University of Copenhagen, Denmark) pointed out that recent developments of genetic and genomic techniques, as well as complementary phenotyping approaches, provide access to functional genomics which is useful for basic and applied plant sciences. However, to develop robust, noninvasive predictors for crop yield and quality, high-throughput genotyping and phenotyping of whole plants has to be correlated with the underlying physiological processes, introduced as physiological phenotyping which includes, for example, phytohormone profiling (Großkinsky et al., 2014) or metabolic fingerprinting (Bertram et al., 2010). By this integrative, multidimensional phenomics approach basic understanding of genotypes and their interactions with the environment can be addressed, which may support breeders and crop management. The correlation of noninvasive field phenotyping to the physiological status of the crops is one key step to achieve the envisioned digital farming in real time combined with high precision. Current high-throughput field phenotyping methods give promising results which can be used as a basis to develop and improve techniques to achieve reliable time- and cost-efficient phenotyping platforms useful for precision agriculture and to assist breeding programmes by monitoring important known traits or identifying novel traits. Phenotypic data including the physiological status of a crop must be associated to the genetic background. Furthermore, the generated phenotypic data have to be acquired and processed in a quantitative, robust and fast way and processed into a format that can be directly interpreted by end-users. These goals can best be achieved by applying general standards to phenotyping campaigns and by the intense exchange between stakeholders (academia, agro-industry) to identify their needs, and the needs of suppliers, to illustrate technical possibilities and limitations. From a breeding perspective field phenotyping will have to facilitate a wide range of applications based on individual breeding strategies and purposes. Which traits to phenotype, how to do it cost-effectively, when, and how to study specific stress to investigate functional genomics and in what way to phenotype both aboveground and belowground are some of the many aspects that will challenge the phenotyping community and overall development of novel cost-efficient field phenotyping methodologies.