Determining Plant Ratings without Field Trials— Prediction of Clonal Disease Ratings for Fiji Leaf Gall Using Nir Spectroscopy and Chemometric Techniques

FIJI leaf gall is one of Australia’s most serious diseases of sugarcane and represents a significant problem in almost half of the total area under sugarcane production. Rating sugarcane clones for resistance/susceptibility is difficult and expensive, due to the nature of field-based methods and variable infection levels of trials. Despite the recent development of glasshouse methods to evaluate clones, there are still issues with throughput, quarantine and the timescale to yield final results. We investigated NIR spectroscopy as an alternative means to rate clones without requiring specific field trials and can successfully predict resistance ratings for a set of traditional standard clones from leaf NIR spectra. Further research intends to develop this into a tool for use within the plant breeding and selection program to screen at an earlier stage(s) for resistance, thereby providing cost savings, productivity benefits and increased numbers of resistant clones to later selection stages. The primary advantages of this method are the timescale to produce ratings (potentially hours as opposed to months) and that requirements of specific field trials and quarantine will be minimal. Other prospects exist to develop improved understanding of the basis of resistance to Fiji leaf gall and similar screening technologies for other pest/disease resistance issues that are either difficult or expensive under traditional techniques. Introduction Fiji leaf gall is one of the Australian sugar industry’s most serious diseases, and is a result of plant infection by the Fiji disease virus (FDV) which is transmitted by the sugarcane planthopper Perkinsiella saccharicida Kirkaldy (Smith, 2000). Past epidemics have threatened industry viability in sugarcane growing regions south of Proserpine, which produce around 2.3 million tonnes of sugar (43% of the total Australian crop). Current Purcell, D.E. et al. Proc. Aust. Soc. Sugar Cane Technol., Vol. 27: 2005 __________________________________________________________________________ 335 control methods include the planting of resistant varieties in regions where the disease is prevalent, and use of disease-free planting material. The disease can cause total crop loss, and varietal resistance is now a prerequisite for the release of a variety in areas south of Proserpine. Recently, new variety release has been compromised, as Fiji leaf gall ratings have been difficult to determine due to the lack of infection of appropriate field trials. Traditionally, new varieties have been screened for resistance by planting small plots of the varieties between rows of cane infected with Fiji leaf gall. Natural populations of the planthopper spread the disease to the test varieties. Over the last 28 years, 13 of the annual trials failed due to insufficient or excessive infection (Croft et al., 2004). Earlier attempts to develop a glasshouse method to screen varieties were unsuccessful, as the ratings obtained did not always correlate well with field reaction (Ledger and Ryan, 1977). A new glasshouse/field method recently reported has produced encouraging results (Croft et al., 2004), but this and traditional methods are relatively labour intensive, thereby limiting the number of varieties that can be screened per year. A few varieties have been released before a Fiji leaf gall resistance rating was available, placing the industry at some risk. If the varieties were actually susceptible, the disease, which is currently under reasonable control, may have flared again. Fortunately, these varieties have now been rated and have acceptable resistance characteristics. During the selection process, approximately 15–30% of prospective new varieties are discarded due to Fiji leaf gall susceptibility, and as such, the overall costs of the plant breeding and selection program would be greatly reduced by being able to discriminate between clones and discard susceptible ones at an earlier stage. In addition, there are other intangible costs incurred including the reduction of overall genetic gain by restricting crosses to resistant parents. Background—NIR spectroscopy and chemometrics Modern spectroscopic instrumention coupled with the application of chemometrics has allowed the development of fast, reliable, nondestructive routine analyses in many fields, including petrochemical, beverage, food and pharmaceutical industries. Generally speaking, these types of analyses have considerable advantages over traditional methodologies including cost, throughput, non-destructive sample preparation and analysis, and reducing risk where potentially dangerous chemicals and/or procedures are involved. Within the Australian sugar industry, there are well-documented applications of NIR spectroscopy using both at-line methodology for plant-breeding purposes (Berding et al., 1989; Brotherton and Berding, 1995, 1998), and on-line cane analysis systems for fibre analyses (Staunton et al., 1999), payment purposes (Staunton et al., 2004), cane-quality schemes (Pope et al., 2004), and factory control (Jones et al., 2002). These analyses were primarily performed on shredded or fibrated cane, although there are further reports providing data on analyses of other substrates derived from sugarcane and various factory streams (Brotherton and Berding, 1995; Schäffler et al., 1993). Based on the success of the above reports in developing inferential methods to predict the outcomes of laboratory analyses conducted on sugarcane, we hypothesised that the same generic methods could be used to predict the behaviour of sugarcane Purcell, D.E. et al. Proc. Aust. Soc. Sugar Cane Technol., Vol. 27: 2005 __________________________________________________________________________ 336 clones with respect to pest and/or disease resistance. Chemical and inferential measurement methods have been used to correlate plant resistance to pests and diseases, both in sugarcane and other crops. Analysis of sugarcane surface waxes allows identification of clones with resistance to the borer Eldana saccharina (Rutherford et al., 1993). This work used gas chromatography (GC) to separate and identify the volatile wax components while, subsequently, Rutherford et al. (1994) and Rutherford and Van Staden (1996) developed rapid non-destructive NIR methods of wax analysis from which a stepwise linear multiple regression model was built and used to predict borer resistance. Based on that work, we have also examined sugarcane surface waxes with GC and spectroscopic techniques, and after chemometric data treatment, discriminated between sugarcane plant properties (Purcell et al., 2003, 2004). More recent research using NIR spectroscopy allowed the identification of pest resistant eucalypt species, and maintained that plant resistance to pests can be based on physical, chemical or ecological traits (Floyd and Foley, 2001). Physical traits such as tough leaves and hairs can render feeding difficult, chemical traits are often based on components within plant tissues which may act as toxins, antifeedants or digestibility reducers, while some insects use olfactory response and surface chemistry to identify their preferred host plant. The study on eucalypts implied that chemical traits conferred resistance to both vertebrate and invertebrate herbivores and that it was directly attributable to particular compounds, thereby suggesting that these resistance traits were more likely to be influenced by genetics rather than by environmental conditions. This paper discusses promising research utilising near infrared (NIR) spectroscopy and chemometric data treatment to predict ratings for Fiji leaf gall disease for sugarcane plants. Further development of this research to a field-based application would offer an effective screening tool which could be applied in close to real-time to significantly larger numbers of plants than can be tested under the current system. Such a tool may be employed earlier in the plant selection process and would complement existing screening methods, which would most likely still be employed for prospective varieties close to commercial release. Materials and methods Plant sampling A set of 10 standard clones is used in conventional Fiji leaf gall field testing with their ratings established by extensive field experience during disease epidemics and their repeated testing in Fiji resistance trials. The ratings are based on the International Society of Sugar Cane Technologists’ recommended rating system, with highly resistant clones given a rating of 1, and the most susceptible clones rated at 9. The ratings of these standard clones cover the full breadth of the rating scale (Table 1). The standard clones were used throughout the current study to allow the construction of a meaningful correlation on which to base future predictive classifications. Purcell, D.E. et al. Proc. Aust. Soc. Sugar Cane Technol., Vol. 27: 2005 __________________________________________________________________________ 337 Table 1—Standard clones used in this study and their Fiji leaf gall rating. Clone Fiji leaf gall rating