HairNet2: deep learning to quantify cotton leaf hairiness, a complex genetic and environmental trait

BACKGROUND

Cotton accounts for 80% of the global natural fibre production. Its leaf hairiness affects insect resistance, fibre yield, and economic value. However, this phenotype is still qualitatively assessed by visually attributing a Genotype Hairiness Score (GHS) to a leaf/plant, or by using the HairNet deep-learning model which also outputs a GHS. Here, we introduce HairNet2, a quantitative deep-learning model which detects leaf hairs (trichomes) from images and outputs a segmentation mask and a Leaf Trichome Score (LTS).

RESULTS

Trichomes of 1250 images were annotated (AnnCoT) and a combination of six Feature Extractor modules and five Segmentation modules were tested alongside a range of loss functions and data augmentation techniques. HairNet2 was further validated on the dataset used to build HairNet (CotLeaf-1), a similar dataset collected in two subsequent seasons (CotLeaf-2), and a dataset collected on two genetically diverse populations (CotLeaf-X). The main findings of this study are that (1) leaf number, environment and image position did not significantly affect results, (2) although GHS and LTS mostly correlated for individual GHS classes, results at the genotype level revealed a strong LTS heterogeneity within a given GHS class, (3) LTS correlated strongly with expert scoring of individual images.

CONCLUSIONS

HairNet2 is the first quantitative and scalable deep-learning model able to measure leaf hairiness. Results obtained with HairNet2 concur with the qualitative values used by breeders at both extremes of the scale (GHS 1-2, and 5-5+), but interestingly suggest a reordering of genotypes with intermediate values (GHS 3-4+). Finely ranking mild phenotypes is a difficult task for humans. In addition to providing assistance with this task, HairNet2 opens the door to selecting plants with specific leaf hairiness characteristics which may be associated with other beneficial traits to deliver better varieties.

The Past, Present, and Future of Host Plant Resistance in Cotton: An Australian Perspective

Cotton is a key global fiber crop. However, yield potential is limited by the presence of endemic and introduced pests and diseases. The introduction of host plant resistance (HPR), defined as the purposeful use of resistant crop cultivars to reduce the impact of pests and diseases, has been a key breeding target for the Commonwealth Scientific and Industrial Research Organisation (CSIRO) cotton breeding program. The program has seen success in releasing cultivars resistant to Bacterial blight, Verticillium wilt, Fusarium wilt, and Cotton bunchy top. However, emerging biotic threats such as Black root rot and secondary pests, are becoming more frequent in Australian cotton production systems. The uptake of tools and breeding methods, such as genomic selection, high throughput phenomics, gene editing, and landscape genomics, paired with the continued utilization of sources of resistance from Gossypium germplasm, will be critical for the future of cotton breeding. This review celebrates the success of HPR breeding activities in the CSIRO cotton breeding program and maps a pathway for the future in developing resistant cultivars.

Cotton Breeding in Australia: Meeting the Challenges of the 21st Century

The Commonwealth Scientific and Industrial Research Organisation (CSIRO) cotton breeding program is the sole breeding effort for cotton in Australia, developing high performing cultivars for the local industry which is worth∼AU$3 billion per annum. The program is supported by Cotton Breeding Australia, a Joint Venture between CSIRO and the program's commercial partner, Cotton Seed Distributors Ltd. (CSD). While the Australian industry is the focus, CSIRO cultivars have global impact in North America, South America, and Europe. The program is unique compared with many other public and commercial breeding programs because it focuses on diverse and integrated research with commercial outcomes. It represents the full research pipeline, supporting extensive long-term fundamental molecular research; native and genetically modified (GM) trait development; germplasm enhancement focused on yield and fiber quality improvements; integration of third-party GM traits; all culminating in the release of new commercial cultivars. This review presents evidence of past breeding successes and outlines current breeding efforts, in the areas of yield and fiber quality improvement, as well as the development of germplasm that is resistant to pests, diseases and abiotic stressors. The success of the program is based on the development of superior germplasm largely through field phenotyping, together with strong commercial partnerships with CSD and Bayer CropScience. These relationships assist in having a shared focus and ensuring commercial impact is maintained, while also providing access to markets, traits, and technology. The historical successes, current foci and future requirements of the CSIRO cotton breeding program have been used to develop a framework designed to augment our breeding system for the future. This will focus on utilizing emerging technologies from the genome to phenome, as well as a panomics approach with data management and integration to develop, test and incorporate new technologies into a breeding program. In addition to streamlining the breeding pipeline for increased genetic gain, this technology will increase the speed of trait and marker identification for use in genome editing, genomic selection and molecular assisted breeding, ultimately producing novel germplasm that will meet the coming challenges of the 21st Century.

HairNet: a deep learning model to score leaf hairiness, a key phenotype for cotton fibre yield, value and insect resistance

BACKGROUND

Leaf hairiness (pubescence) is an important plant phenotype which regulates leaf transpiration, affects sunlight penetration, and provides increased resistance or susceptibility against certain insects. Cotton accounts for 80% of global natural fibre production, and in this crop leaf hairiness also affects fibre yield and value. Currently, this key phenotype is measured visually which is slow, laborious and operator-biased. Here, we propose a simple, high-throughput and low-cost imaging method combined with a deep-learning model, HairNet, to classify leaf images with great accuracy.

RESULTS

A dataset of [Formula: see text] 13,600 leaf images from 27 genotypes of Cotton was generated. Images were collected from leaves at two different positions in the canopy (leaf 3 & leaf 4), from genotypes grown in two consecutive years and in two growth environments (glasshouse & field). This dataset was used to build a 4-part deep learning model called HairNet. On the whole dataset, HairNet achieved accuracies of 89% per image and 95% per leaf. The impact of leaf selection, year and environment on HairNet accuracy was then investigated using subsets of the whole dataset. It was found that as long as examples of the year and environment tested were present in the training population, HairNet achieved very high accuracy per image (86-96%) and per leaf (90-99%). Leaf selection had no effect on HairNet accuracy, making it a robust model.

CONCLUSIONS

HairNet classifies images of cotton leaves according to their hairiness with very high accuracy. The simple imaging methodology presented in this study and the high accuracy on a single image per leaf achieved by HairNet demonstrates that it is implementable at scale. We propose that HairNet replaces the current visual scoring of this trait. The HairNet code and dataset can be used as a baseline to measure this trait in other species or to score other microscopic but important phenotypes.

Correction to: Improved lint yield under field conditions in cotton over-expressing transcription factors regulating fibre initiation

Due to an unfortunate misunderstanding, an extra middle initial erroneously appeared in the original publication and the full name of the first author should read Shi Ming Liu.

Enhancing Integrated Pest Management in GM Cotton Systems Using Host Plant Resistance

Cotton has lost many ancestral defensive traits against key invertebrate pests. This is suggested by the levels of resistance to some pests found in wild cotton genotypes as well as in cultivated landraces and is a result of domestication and a long history of targeted breeding for yield and fiber quality, along with the capacity to control pests with pesticides. Genetic modification (GM) allowed integration of toxins from a bacteria into cotton to control key Lepidopteran pests. Since the mid-1990s, use of GM cotton cultivars has greatly reduced the amount of pesticides used in many cotton systems. However, pests not controlled by the GM traits have usually emerged as problems, especially the sucking bug complex. Control of this complex with pesticides often causes a reduction in beneficial invertebrate populations, allowing other secondary pests to increase rapidly and require control. Control of both sucking bug complex and secondary pests is problematic due to the cost of pesticides and/or high risk of selecting for pesticide resistance. Deployment of host plant resistance (HPR) provides an opportunity to manage these issues in GM cotton systems. Cotton cultivars resistant to the sucking bug complex and/or secondary pests would require fewer pesticide applications, reducing costs and risks to beneficial invertebrate populations and pesticide resistance. Incorporation of HPR traits into elite cotton cultivars with high yield and fiber quality offers the potential to further reduce pesticide use and increase the durability of pest management in GM cotton systems. We review the challenges that the identification and use of HPR against invertebrate pests brings to cotton breeding. We explore sources of resistance to the sucking bug complex and secondary pests, the mechanisms that control them and the approaches to incorporate these defense traits to commercial cultivars.

Lack of adaptation to a new host in a generalist herbivore: implications for host plant resistance to twospotted spider mites in cotton

BACKGROUND

The twospotted spider mite (Tetranychus urticae Koch) is an important pest of cotton. This pest has a broad host range, but when changing between hosts an initial decline in fitness often occurs. This is usually followed by an increase in fitness after rapid adaptation to the new host, usually within five generations.

RESULTS

The generality of this adaptive response was tested by assessing elements of fitness when mites were reared on a host to which they were adapted (Gossypium hirsutum L. cv. Sicot 71) or on a new host, Gossypium arboreum L. (accession BM13H). In a first experiment, mites reared on the new host for ten generations showed declining immature survival compared with those reared on the adapted host. In a second experiment, the intrinsic capacity for increase of mites cultured on the new host for six generations was significantly lower than that of mites cultured on the adapted host for six generations and then transferred to the new host. Hence, exposure to the new host for six or ten generations resulted in declining fitness.

CONCLUSION

This 'negative adaptation' indicates robust antibiosis traits in G. arboreum accession BM13H, which therefore have value in developing mite-resistant G. hirsutum cultivars.

Jasmonic acid is associated with resistance to twospotted spider mites in diploid cotton (Gossypium arboreum)

The twospotted spider mite (Tetranychus urticae Koch) is capable of dramatically reducing the yield of cotton crops and is often difficult and expensive to control. This study investigated and compared two important plant hormones, jasmonic acid (JA) and salicylic acid (SA), as constitutive and/or induced defence response components in a mite susceptible commercial cotton cultivar, Sicot 71 (Gossypium hirsutum L.) and a resistant diploid cotton BM13H (Gossypium arboreum L.). Foliar application of JA and methyl jasmonate (MeJA) reduced the mite population and leaf damage but application of other potential elicitors, SA and methyl salicylate (MeSA) did not. The concentrations of JA and SA in leaf tissues of induced and non-induced Sicot 71 and BM13H were quantified by liquid chromatography coupled to electrospray ionisation tandem mass spectrometry (LC-ESI-MS/MS). The JA content was constitutively higher in BM13H than Sicot 71 and also highly induced by mite infestation in BM13H but not in Sicot 71. However, SA was not significantly induced in either BM13H or Sicot 71. The expression levels of JA related genes, LOX, AOS and OPR were measured by quantitative PCR and elevated expression levels of JA related genes were detected in mite-infested BM13H. Therefore, JA and MeJA were implicated as key biochemical components in both the constitutive and induced defence responses of BM13H to spider mites.

Fitness of twospotted spider mites is more affected by constitutive than induced resistance traits in cotton (Gossypium spp.)

BACKGROUND

Life history parameters are useful tools for comparing the fitness of pests on different host plants. This study compared life history parameters of twospotted spider mites (Tetranychus urticae Koch) on two resistant cotton Gossypium genotypes (BM13H and Sipima 280) and one susceptible genotype (Sicot 71). The effects of both constitutive and induced defences were assessed.

RESULTS

Mites reared on the resistant genotypes had longer immature development times, lower immature survival and reduced adult fecundity. Mites reared on BM13H that had been induced by prior exposure to mites had a small additional decrease in adult fecundity. The contribution to mite resistance of constitutive resistance mechanisms was much greater than induced responses. The effect of morphological constitutive defences was minor, implicating biochemical defences as the major mite-resistance mechanism. Sensitivity analysis and a population development study using life history parameters of mites showed that a lower immature survival rate on resistant genotypes had the greatest effect on mite fitness and population development.

CONCLUSION

Use of life history parameters provided valuable insight into the mite-resistance mechanisms of these Gossypium genotypes. Further, the results largely explained mite population development on these genotypes in the field.