Sindhuja Sankaran 2016-05-03 02:02:17
An increase in agricultural production to feed a projected nine billion people by 2050 is a food security challenge that is being addressed by researchers worldwide. Crop improvement through plant breeding has potential to sustain and even increase global food production, especially considering the abiotic and biotic plant stress conditions that come with escalating climate change. Food security and phenotyping Plant phenotyping, which is the assessment of expressed traits (as influenced by genetic make-up and changes in the environment), is a vital process in crop improvement programs. Phenotyping gives us a better understanding of the inheritance of traits and the selection of genetic markers with desirable traits. Examples of such traits include drought tolerance, disease resistance, higher yield potential, and superior fruit quality. Current advances in DNA sequencing techniques, such as marker-assisted selection and direct introduction of desired alleles, have improved the genomic efficiency and decreased the associated costs. However, the phenotyping process remains slow, expensive, and sometimes destructive. In the last few years, high-throughput phenotyping platforms and sensing technologies have been developed for accurate, rapid, and efficient evaluation of phenotypes, especially in field conditions. In the Pacific Northwest, researchers at Washington State University and the USDA-ARS, in collaboration with local commodity producers, have established more than 15 distinct improvement programs for crops grown under rainfed and irrigated conditions. These programs include grains and legumes (wheat, barley, chickpea, pea, and lentil) along with several specialty crops (apple, cherry, pear, and hop). Phenotyping is an important part of all these programs. Our collaborative efforts are directed toward high-throughput phenotyping in controlled environment and field conditions through the development of integrated sensor platforms to quantify desired crop traits, as well as data mining procedures to support crop improvement decisions. Sensor platforms and automation Sensor platforms can range from commercial units (such as those offered by LemnaTec, Phenospex, etc.) to customdesigned robotic platforms for controlled environment and open-field applications. These platforms need precision positioning, a high level of automation, and ruggedness to acquire high-quality data in a rapid and continuous manner. Current ground-based phenotyping efforts in the U.S. lean toward customizing tractors or boom sprayers with commercial sensors. These platforms allow plot-level data capture with GPSenabled navigation for geospatial analysis. However, there is no single, low-cost, rugged, “fit-for-all” ground platform for phenotyping diverse crops that can be adjusted to different row and plant spacings, plot sizes, and plant heights. This adaptability is especially important in Washington State, which has very diverse agriculture, ranging from row and field crops to tree fruits. Aerial platforms, such as small UAS, are evolving rapidly and may be a solution for high-throughput sensing in diverse cropping systems. Small UAS can be a “fit-for-all” sensing platform that allows low-altitude, high-resolution (1 to 2 cm) data acquisition. Regular monitoring of field plots throughout the growing season with GPS-guided UAS can provide consistent high-quality data. Small UAS are easy to transport from one location to another, such as when the same breeding lines are being tested under different environmental conditions. Small UAS also provide operational flexibility in situations where ground-based phenotyping is not practical, such as for evaluating crop response immediately after irrigation or rainfall. Currently, small UAS have a limited sensor payload (size and weight), operating altitude (regulatory issues), and flight time (battery life). However, the smaller plot sizes used in plant breeding programs avoid these limitations, making small UAS attractive for crop phenotyping. In some of our ongoing projects, our team is using both ground-based and UAS-based platforms to evaluate water use efficiency in dry bean, potato, and grapevine production, as well as emergence in winter wheat and potato, canopy closure and senescence in potato, and grain and legume yield potential using multispectral and thermal imaging. The data processing challenge The types of sensors that can be used with ground and aerial platforms to evaluate different phenotypes include multispectral, hyperspectral, thermal, sonar, lidar, and fluorescence sensors, among others. The sensing capabilities, including spatial and spectral resolution, as well as the frequency, precision, and accuracy of data acquisition, are improving every day, enhancing our ability to acquire timely, high-quality data. However, converting this “big data” into meaningful phenotypic information remains a challenge. In addition, many commercial sensors require careful calibration and independent field tests for use in crop phenotyping. This data-processing bottleneck can be addressed through the development of algorithms and data management protocols to extract useful phenotypic information from the data. In addition, research on the relationships between remotely sensed data and ground reference measurements is vital for wider adoption of rapid, non-contact sensing in crop phenotyping. Our research at WSU focuses on all these aspects, and our goal is to contribute solutions to the global food challenge. ASABE member Sindhuja Sankaran, Assistant Professor, Department of Biological Systems Engineering, Washington State University, Pullman, USA, firstname.lastname@example.org.
Published by ASABE. View All Articles.