Anthracnose diseases on strawberry are destructive and may be caused by three Colletotrichum species: C. acutatum J. H. Simmonds, C. fragariae, and C. gloeosporioides (Maas, 1998; Smith, 1998). C. acutatum is the primary causal agent of anthracnose fruit rot and irregular leaf spot. Both C. fragariae and C. gloeosporioides may infect any above-ground part of the plant and incite anthracnose crown rot, anthracnose fruit rot, and anthracnose leaf spot, also called black leaf spot (Howard and Albregts, 1983). Warm temperatures and high humidity allow these pathogens to produce spores rapidly, and these are easily dispersed throughout production fields by rain splash, people, animals, insects, and equipment.
Anthracnose disease control strategies include planting disease-free plants, good sanitation, and the use of cultural and chemical controls; however, overuse of fungicides has resulted in pathogen resistance and failure of some fungicides to control anthracnose epidemics (Forcelini and Peres, 2018; LaMondia, 1995; Smith and Black, 1993a, 1993b). Ideal anthracnose disease control relies on the development and planting of disease-resistant cultivars. Many years are required to develop anthracnose-resistant strawberry germplasm with desirable plant growth habit, fruit taste and flavor, yield, and resistance to insects and other diseases. To identify disease-resistant germplasm, thousands of seedlings must be produced and evaluated for disease response in the field using natural inoculum or in greenhouse trials relying on artificial inoculation. However, inoculation trials are time-consuming, and many plants may be killed by the disease. This presents a problem for breeders because the plants killed in inoculation trials might have possessed other desired horticultural traits that could be used in their breeding program.
Screening strawberry germplasm for disease resistance using detached leaf assays is an alternative to inoculating whole plants that allows plant disease response to be determined without destroying the plant, reduces the time between inoculation and disease assessment, and confines the pathogen to the laboratory, which allows breeders to test for pathogens or races of pathogens from other geographic areas without transferring the pathogen to the field or risking its introduction to the industry. Inoculating detached leaves with a conidial suspension of Colletotrichum species was shown to be an accurate nondestructive method of identifying anthracnose-resistant germplasm by Miller-Butler et al. (2018).
Disease severity refers to the amount of plant tissue that is diseased (necrotic) and may be expressed as the percentage of plant area destroyed by a pathogen. Percentage or numerical disease assessment scales, such as visual rating scales, are often used for quick assessments of disease severity of plant tissue. Percentage scales often are adapted from the Horsfall-Barratt scale (Horsfall and Barratt, 1945), which contains 12 gradations, from 1% to 100%, with the percent disease varying disproportionately. Visual bias can influence accuracy, and percentage scales may be difficult to use when evaluating plants that exhibit noticeably different amounts of infection (James, 1977; Sherwood et al., 1983; Slopek, 1989). When there is little disease, the rater’s visual focus is drawn to the small amount of necrotic or dark tissue compared with healthy or green tissue. When there is only a small amount of healthy tissue in a very diseased sample, the rater’s visual focus is drawn to the small amount of healthy (green) tissue (Sherwood et al., 1983). Small areas of disease or no disease can be seen and a percentage can be determined fairly accurately, but when the lesion area ranges from 10% to 90%, it is much more difficult to give an accurate percentage. Slopek (1989) compared five variations of a 1 to 5 visual rating scale for estimating the percent diseased leaf area of barley plants and found that two of the five visual rating scales worked well for estimating the leaf disease, were as precise as the Horsfall-Barratt scale, and decreased the time required for disease assessment. Other researchers have suggested the use of an equal interval scale over the Horsfall-Barratt scale for assessing disease severity (Nita et al., 2003) or the use of illustrated assessment keys with standard area diagrams of percentage scales that allows standardization in disease assessment methods for a variety of crops (James, 1977).
Disease symptoms on plants or leaves are often evaluated using numbered grade scales, sometimes referred to as arbitrary, nominal, or ordinal scales, which have some degree of subjective interpretation of the disease by the rater. An ordinal scale of 0 to 5 (0 = no disease, 1 = very slight, 2 = slight, 3 = moderate, 4 = severe, 5 = dead plant) is only interpretable in the arrangement of the order and can only provide qualitative data. Many disease assessment keys are ordinal and cannot quantitatively measure a difference between the values but were satisfactory when used by experienced observers for rating plants or plots in an order of increasing symptom severity (Russell, 1978).
When there is more than one visual rater, good agreement and association between raters are desirable. Understanding the interpretation of results from ordinal data can be explained by the concepts of accuracy (agreement) and precision (association). Accuracy is the raters’ ability to rate disease closest to a true value (such as percent disease measurements determined by image analysis) and precision is the repeatability of the scoring; however, accuracy and precision may not coincide. Statistical analysis helps determine whether raters are interpreting disease the same or very close to the same. The Kappa coefficient (k) (Cohen, 1960) is a measure of the difference between the raters’ agreement and the expected agreement. It determines a proportion of agreement (accuracy), correcting for chance agreement and is scaled to vary from –1 to +1. A negative Kappa coefficient indicates less than chance agreement, zero indicates exactly chance agreement, a positive value indicates better than chance agreement, and +1 indicates perfect agreement. The Kappa coefficient has been interpreted as <0 = less than chance agreement, 0.01 to 0.20 = slight agreement, 0.21 to 0.40 = fair agreement, 0.41 to 0.60 = moderate agreement, 0.61 to 0.80 = substantial agreement, and 0.81 to 0.99 = almost perfect agreement (Viera and Garrett, 2005). Agreement and disagreement are not mutually exclusive in an ordinal rating scale. If two raters see the disease on a plant as slight disease and moderate disease, they are not in complete agreement, but they are not in complete disagreement either because both raters have determined that there is disease on the plant. This problem was addressed with the weighted Kappa coefficient by assigning weights to different degrees of disagreement with lesser weight to agreement because categories are further apart (Cohen, 1960, 1968; Fleiss and Cohen, 1973). The degree of linear association (precision) between two sets of data, such as the two visual raters’ disease severity ratings, can be statistically measured with Pearson’s product moment correlation coefficient (rp). A positive Pearson correlation coefficient designates that both sets of data change in the same direction, and a negative Pearson correlation coefficient designates that both sets of data change in opposite directions.
Precise quantitative analysis can be performed on images of diseased plant tissues using computer software. Electronic images can be stored indefinitely, allowing the researcher to process the images as time permits. Digital imaging and analytical software were used by Wang et al. (2008) to develop a miniaturized strawberry leaf disk antifungal bioassay. Their goal was to determine percent disease caused by a Colletotrichum isolate used for inoculation of a leaf disk, as well as the percent phytotoxicity that might have been caused by the antifungal compounds using 15-mm excised strawberry leaf disks. The leaf disks were dipped in antifungal compounds and then inoculated with the same Colletotrichum species being used in this research. The analyzing software transformed the images to show healthy parts of the leaf as green, diseased parts as black, and parts exhibiting phytotoxicity as gray. In a similar study (Abril et al., 2009), photographs of detached strawberry leaves were used for visual assessment of disease severity in a study testing the efficacy of natural product-based fungicides, and the percent diseased leaf area was assessed with an arbitrary scale of 0 = no disease to 3 = most severe disease. When digital image analysis and visual assessment were both used by Kwack et al. (2005) to determine the percentage of diseased area caused by C. orbiculare on cucumber leaves, visual ratings were significantly higher than the image analysis ratings, and the authors reported processing the digital images took longer. When rating severity of foliar citrus canker symptoms, Bock et al. (2009) also reported image analysis was slower compared with visual assessment; however, image analysis gave more precise and accurate results. In contrast, Nutter et al. (1993) reported image analysis as a measure of disease severity was faster compared with visual assessments.
The objective of this study was to compare visual assessments with image analysis of anthracnose disease on inoculated detached strawberry leaves to determine the degree of agreement and association between the two methods. This research expands on preliminary trials to develop a reliable detached leaf assay for anthracnose resistance in strawberry plants (Miller-Butler et al., 2013).
AbrilM.CurryK.J.SmithB.J.DeluccaA.J.BoueS.WedgeD.E.2009Greenhouse and field evaluation of the natural saponin CAY-1 for control of several strawberry diseasesInt. J. Fruit Sci.9211220
BockC.H.ParkerP.E.CookA.Z.GottwaldT.T.2009Automated image analysis of the severity of foliar citrus canker symptomsPlant Dis.93660665
ChangP.SmithB.J.2007Response of ‘Treasure’ and 22 Florida strawberry genotypes to the anthracnose pathogens Colletotrichum acutatum C. fragariae and C. gloeosporioides p. 69–73. In: Proceedings of 2007 N. American Strawberry Symposium. F. Takeda D.T. Handley and E.B. Poling (eds.). North American Strawberry Growers Association Kemptville ON Canada
FleissJ.L.CohenJ.1973The equivalence of weighted kappa and the intraclass correlation coefficient as measures of reliabilityEduc. Psychol. Meas.33613619
ForceliniB.B.PeresN.A.2018Widespread resistance to QoI fungicides of Colletotrichum acutatum from strawberry nurseries and production fieldsPlant Health Prog.19338341
GallettaG.J.SmithB.J.GuptonC.L.1993Strawberry parent clones US70, US159, US292, and US438 resistant to anthracnose crown rotHortScience2810551056
GwetK.L.2012Handbook of inter-rater reliability. 3rd ed. Advanced Analytics LLC Gaithersburg MD
HowardC.M.AlbregtsE.E.1983Black leaf spot phase of strawberry anthracnose caused by Colletotrichum gloeosporioides (= C. fragariae)Plant Dis.6711441146
KwackM.S.KimE.N.LeeH.KimJ.ChunS.KimK.D.2005Digital image analysis to measure lesion area of cucumber anthracnose by Colletotrichum orbiculareJ. Gen. Plant Pathol.71418421
LaMondiaJ.A.1995Inhibition with benomyl to growth in vitro of Colletotrichum acutatum and C. fragariae and strawberry fruit infection by benomyl-resistant isolates of Colletotrichum acutatumAdv. Strawberry Res.142530
MaasJ.L.1998Compendium of strawberry diseases. APS press St. Paul MN
Miller-ButlerM.A.CurryK.J.KreiserB.R.SmithB.J.2013Visual and electronic evaluations of detached strawberry leaves inoculated with Colletotrichum speciesInt. J. Fruit Sci.13114125
Miller-ButlerM.A.SmithB.J.KreiserB.R.BlytheE.K.2019Comparison of anthracnose resistance with the presence of two SCAR markers associated with the Rca2 gene in strawberryHortScience54793798
Miller-ButlerM.A.SmithB.J.BabikerE.M.KreiserB.R.BlytheE.K.2018Comparison of whole plant and detached leaf screening techniques for identifying anthracnose resistance in strawberry plantsPlant Dis.10221122119
MunirM.AmsdenB.DixonE.VaillancourtL.GauthierN.W.2016Characterization of Colletotrichum species causing bitter rot of apple in Kentucky orchardsPlant Dis.10021942203
NitaM.EllisM.A.MaddenL.V.2003Reliability and accuracy of visual estimation of Phomopsis leaf blight of strawberryPhytopathology939951005
NutterF.W.JrGleasonM.L.JencoJ.H.ChristiansN.L.1993Accuracy, intra-rater repeatability, and inter-rater reliability of disease assessment systemsPhytopathology83806812
RussellG.E.1978Plant breeding for pest and disease resistance. Butterworths London UK
SherwoodR.T.BergC.C.HooverM.R.ZeidersK.E.1983Illusions in visual assessment of Stagonospora leaf spot of orchard grassPhytopathology73173177
SlopekS.W.1989An improved method of estimating percent leaf area diseased using a 1 to 5 disease assessment scaleCan. J. Plant Pathol.11381387
SmithB.J.1998Anthracnose crown rot p. 46–48. In: J.L. Mass (ed.). Compendium of strawberry disease. 2nd ed. American Phytopathological Society St. Paul MN
SmithB.J.BlackL.L.1990Morphological, cultural, and pathogenic variation among Colletotrichum species isolated from strawberryPlant Dis.746976
SmithB.J.BlackL.L.1993aIn vitro fungicide studies show the occurrence of benomyl resistant Colletotrichum spp. from strawberryAdv. Strawberry Res.124248
SmithB.J.GuptonC.L.GallettaG.J.MaasJ.L.EnnsJ.M.BallingtonJ.R.JrConstantinR.J.DiVittorioT.J.Himelrick. DD.1998‘Pelican’ StrawberryHortScience3310821084
WangX.WedgeD.E.NurhayatT.JohnsonR.D.CutlerS.J.PaceP.F.SmithB.J.ZhouL.2008Development of a miniaturized 24-well strawberry leaf disk bioassay for evaluating natural fungicidesNat. Prod. Commun.310791084