Aerial phenotyping tp identify superior sugarcane genotypes.

Abstract

Sugarcane is a globally important food and bioenergy crop which necessitates continual improvement through breeding to ensure its sustainable production under increasingly challenging environments. Compared to other major crops, yield gains in sugarcane have stagnated worldwide in recent years. This could be attributed to the resource-intensive and time-consuming nature of breeding a genetically complex crop with long growth cycles and large, diverse populations. The primary objective of sugarcane breeding is to develop superior genotypes with enhanced genetic gains, securing sustainable production for the future. Aerial phenotyping (AP) with high-throughput phenotyping sensor technologies and unmanned aerial vehicles (UAVs, commonly known as drones) could assist breeding by increasing selection efficiency and accuracy, uncovering genetic variation for yieldpromoting traits, and expediting large-scale trial screening. Key physiological traits governing canopy development and water use, namely green canopy cover (GCC) and stomatal conductance (gs), can be estimated from the aerially measured Normalized Difference Vegetation Index (NDVI) and canopy temperature (Tc), respectively. While promising, further research is required to evaluate the efficacy of AP in breeding. The study aimed to develop and test an AP procedure for identifying superior genotypes in sugarcane breeding. The specific objectives were to: (1) determine the impacts of ground (GCC and gs) and aerially measured traits (NDVI and Tc), on stalk dry mass yield (SDM) under well-watered (WW) and water deficit (WD) conditions; (2) develop an AP procedure for estimating gs, GCC and SDM from Tc and NDVI; (3) determine the genetic variation and broad-sense heritability of ground and aerially measured traits; (4) evaluate the feasibility and potential benefit of implementing AP to identify superior genotypes in breeding. These aims and objectives were addressed in three experimental phases. An unreplicated pilot trial with two genotypes grown under WW and WD conditions (~ 200 m2 in total) was used to establish preliminary relationships between ground and aerially measured traits under varying canopy and moisture conditions. Key findings were that FIPAR (fractional interception of photosynthetically active radiation - a surrogate measure of GCC) measured on the ground could be reliably estimated from NDVI, though the relationship required further investigation at partial canopy cover. Tc could be used to distinguish differences in measured gs between water treatments under moderate to severe

stress conditions only. Overall, the experiment was used to formulate a preliminary AP protocol, with recommendations for further improvement in the subsequent phase. A replicated field trial with 54 genotypes, grown under WW and WD conditions (~ 3 ha in total) in plant and first ratoon crops, was used to assess trait correlations, genetic variation and broad-sense heritability of traits, and to refine the AP procedure. In line with previous research, the study confirmed FIPAR and gs as influential traits for determining SDM. FIPAR, measured at 2-3 months after crop start, could be used to identify high- and lowyielding genotypes, and could be predicted well from NDVI, at partial canopy for wellwatered crops. Breeding programs for irrigated environments could benefit from the early identification of superior genotypes if traits with high heritability, like FIPAR, can be accurately and rapidly phenotyped. Furthermore, results suggested that high gs benefits wellwatered crops, while relatively low gs could be advantageous in dry environments, though this requires further validation. Phenotyping of gs from Tc was mostly unreliable, and its practical application in breeding programs requires further evaluation on a larger, genetically diverse population with improved measurement procedures. It was concluded that NDVI and Tc, which both showed significant genotypic variation and moderate to high heritability, could be used to identify high- and low-yielding genotypes when measured early in the crop cycle in young, partially canopied, well-watered crops planted in multi-row plots. Novel results also showed potential for screening of drought tolerance using water treatment differences in Tc and SDM, which has not been reported previously for sugarcane. Overall, this research was used to establish an AP procedure for subsequent use in breeding trials. Lastly, the AP procedure was implemented in two rainfed early-stage breeding trials, with 1770 to 2130 genotypes, planted in replicated single-row plots over ~3.5 – 6 ha. This validation phase was used to test the utility of AP for enhancing selection accuracy and efficiency and contribute to yield improvement. The limited number of flights in the first trial prevented adequate capture of temporal and genotypic variations in aerially measured traits, which are necessary for accurate yield prediction. In the second trial, early estimates of NDVI and Tc, measured approximately 1.5 to 3 months after crop start in partially canopied, well-watered crops, showed significant genotypic variation, moderate to high heritability, and significant correlation with yield. Tc was also significantly correlated with yield when measured shortly after canopy closure but before row overlap due to crop growth.

Despite these promising results, the AP procedure implemented in these early-stage breeding trials did not achieve the precision required for genotype selection. A comparison of direct (SDM-based) and indirect (based on aerially measured traits) selection approaches showed that the number of positive matches was mostly offset by a larger number of incorrectly identified genotypes using aerially measured traits. It was concluded that the effectiveness of AP in breeding is currently hindered by limitations in the precision of aerial measurements and challenges in breeding trial execution. The findings from this study highlight the potential and limitations of AP for physiological breeding. AP holds great promise for identifying genetic variation in yield-promoting traits, which could be leveraged in breeding for the identification of superior genotypes in irrigated environments, however further research is required to fully realize this potential. It is recommended to modify the design of early-stage trials by increasing plot length, number of rows, and row-spacing to facilitate accurate estimation of aerially measured traits using the developed AP procedure. Further efforts are also needed to overcome challenges inherent in breeding trial execution, such as lengthy planting periods introducing biases in early vigour, and variability in field soil composition, which directly and indirectly affect the quality of ground and aerially measured data. Should these recommendations be implemented, early screening of trials using AP could lead to shorter breeding cycles, the discovery of novel genetic variations, and improved selection efficiency, ultimately reducing the resourceintensive nature of traditional methods through early elimination of inferior genotypes from the program. In conclusion, this study demonstrates the potential of AP to enhance sugarcane breeding by facilitating the early detection of important yield-promoting traits, particularly in wellwatered crops. While AP shows promise to enhance sugarcane breeding, further work in refining its application is essential to fully realize its benefits. These research findings provide a strong foundation for future efforts to develop innovative breeding strategies and precision agriculture technologies.