Digital seedling phenotyping & quality assessment

Abstract

Two major challenges in the field of agriculture are the need for increases in food production due to the ever-increasing world population and simultaneously climate change caused by global warming. The increasing world population requires more food, and the food needs to be of a higher standard. Currently, the increase and high quality food production by farming are becoming difficult due to more inconsistency of the climate. Plant breed companies use old-fashioned science in their attempts to overcome the agriculture challenges. For plant breeding, multiple plants are combined to develop a plant crop with desired characteristics that can withstand to the environmental challenges, produce high quality crops and provide the production volumes according to the required specifications. If such a crop is developed using breeding, a batch of plant seeds will be produced to sell the plant seeds all over the world. Two main problems do come from breeding and production of these newly developed crops, firstly breeders need information about the plants for efficient plant breeding, and secondly, the quality of the produced plant seeds has to be ensured. The quality of the plant seeds is an important factor to sell the produced plant seeds over the world to growers and farmers. To ensure the quality of the seeds, phenotype experts obtain the information and perform quality assessments of the seeds and grown plants. To obtain information on plants, the experts use plant phenotyping in which the characteristics of the plant are described and evaluated. The evaluation of the plant characteristics provides a quality assessment of the batch of produced plant seeds. The quality assessment is done by growing the plant seed to seedlings and compare the grown seedlings with each other. The visual comparison between the seedlings is used to give a quality label to the produced batch of seeds. At the moment this visualization and subjective comparison of seedlings is the state of the art. Experts did not believe it is possible to assess the seedlings otherwise. To move away from the subjective approach, Rijk Zwaan BV is looking into more objective measurement systems to obtain plant information. The objective plant information of the seedlings provides solutions of objectively assessing quality and inform plant breeders with extra information making the plant breeding process more efficient. In this study, the feasibility of a digital phenotyping system able to give plant information and a quality assessment of the seedlings was examined. To show the possibilities of the system a model of a digital seedling quality system was developed. By applying the literature of plant phenotyping systems, a phenotype and quality assessment data pipeline were established. The data pipeline was developed to be understandable and transparent for the experts to ensure acceptance of a digital system. The plant phenotype pipeline consists out of multiple components. Each component of the plant phenotype data pipeline has been developed to form the seedling quality system model. The developed model is called SeeQS (Seedling Quality System) and is able to seek out the quality of the seedlings and provide phenotype information of the seedlings. SeeQS was evaluated using multiple samples of pepper seedlings to show the performance and capabilities of a digital phenotyping-based quality system. The system performance shows to be very capable and accurate to assess the quality of seedlings and give plant measurements. The model shows that it is possible to assess the quality of seedlings using a digital phenotyping-based system. Due to the transparency of the developed model, the experts understand the versatility of the digital system. Now, the phenotype experts accept the possibility of a digital system to obtain plant information and give a quality assessment. Due to these positive results, the company wants to continue with the development and implementation of a digital seedling quality and phenotyping system.