Diagrammatic scale to assess downy mildew severity in melon

Escala diagramática para avaliação da severidade do míldio do meloeiro

COMUNICAÇÃO CIENTÍFICA SCIENTIFIC COMMUNICATION

Diagrammatic scale to assess downy mildew severity in melon

Escala diagramática para avaliação da severidade do míldio do meloeiro

Sami J Michereff^I; Marissônia A Noronha^II; Gaus SA Lima^III; Ígor CL Albert^I; Edilaine A Melo^III; Luciana O Gusmão^III

^IUFRPE-Dep^to.Agronomia, Fitossanidade, Av. Dom Manoel de Medeiros, s/n, 52171-900 Recife-PE

^IIEmbrapa Meio Norte, C. Postal 01, 64006-220 Teresina-PI

^IIIUFAL-CCA, BR 104 Norte, km 85, 57100-000 Rio Largo-AL; [e-mail]

ABSTRACT

The downy mildew, caused by Pseudoperonospora cubensis, is an important melon disease in Northeast Brazil. Considering the lack of standard methods for its assessment, a diagrammatic scale was developed with 2, 4, 8, 16, 32, 64, 82, and 96% of affected leaf area. The scale was then checked for its accuracy, precision, and reproducibility in estimating downy mildew severity. The diagrammatic scale was validated by eight disease raters; using 50 leaves with different severity levels, previously measured using the software Assess^®. Two evaluations were performed on the same set of leaves, but in a different sequence order, by the same raters, within a 15-day interval. The accuracy and precision of each rater was determined by simple linear regression between the actual and the estimated severity. The scale provided good levels of accuracy (means of 87.5%) and excellent levels of precision (means of 94%), with absolute errors concentrated around 10%. Raters showed great repeatability (means of 94%) and reproducibility (>90% in 90.3% of cases) of estimates. Therefore, we could conclude that the diagrammatic scale presented here was suitable for evaluating downy mildew severity in melon.

Keywords:Cucumis melo, Pseudoperonospora cubensis, foliar disease, epidemiology, phytopatometry.

RESUMO

O míldio, causado por Pseudoperonospora cubensis, é uma importante doença do meloeiro no nordeste brasileiro. Devido à inexistência de métodos padronizados para quantificação desta doença, foi elaborada uma escala diagramática com os níveis 2; 4; 8; 16; 32; 64; 82 e 96% de área foliar lesionada, avaliando-se a acurácia, a precisão e a reproducibilidade das estimativas de severidade da doença. A escala diagramática foi validada por oito avaliadores que utilizaram 50 folhas com sintomas da doença em diferentes níveis de severidade, mensuradas previamente pelo programa Assess^®. Foram realizadas duas avaliações em um intervalo de 15 dias, em que seqüências diferentes das mesmas folhas foram estimadas visualmente pelos mesmos avaliadores. A acurácia e a precisão de cada avaliador foram determinadas por regressão linear simples entre a severidade real e a estimada. A escala diagramática proporcionou bons níveis de acurácia (média de 87,5%) e excelentes níveis de precisão (média de 94%), com os erros absolutos concentrando-se na faixa de 10%. Os avaliadores apresentaram elevada repetibilidade (média de 94%) e reproducibilidade (>90% em 90,3% dos casos) das estimativas. Portanto, a escala proposta mostrou-se adequada para avaliação da severidade do míldio do meloeiro.

Palavras-chave:Cucumis melo, Pseudoperonospora cubensis, doença foliar, epidemiologia, fitopatometria.

The Northeast region of Brazil responds for nearly 95% of the national melon (Cucumis melo L.) production. Located at this region, States of Rio Grande do Norte and Ceará supply respectively 55 and 28% of the total melon harvested in Brazil (FNP, 2007). The expansion of the area used for melon production in this region, in addition to an intensive year-round growing system, contribute to rise the intensity of several diseases, amongst them downy mildew, caused by Pseudoperonospora cubensis (Berk. & Curt.) Rost. (Santos et al., 2000). Downy mildew is one of the most important melon diseases in Northeast Brazil. It causes up to 60% reduction in fruit production (Cardoso et al., 2002a) and 49% in the content of soluble solids (Cardoso et al., 2002b).

Downey mildew symptoms start showing up in older leaves, as angular light-yellow spots, limited by the veins. Following, lesions may coalesce and rot the tissue, which assumes a bronze to brown hue. Olive-green to purple fungi frutifications, the pathogen sporangiophores and spores, can be seen in the adaxial leaf surface. Severe infection results in early leaf dropping, raquitism, and malformed and underdeveloped fruits (Rego & Carrijo, 2000; Tavares, 2002).

For any program of integrated disease management to succeed, it is crucial to assess disease intensity with accuracy and precision. Accuracy refers to how faithful an estimate is from the actual amount of the evaluated disease, while precision corresponds to the confidence and/or repeatability associated to such estimate (Nutter Jr. & Schultz, 1995). Considering that the melon downy mildew is a leaf spot, its intensity is more appropriately expressed by assessing severity, i.e., the percent or proportion of affected leaf area. Most of the times, disease severity is assessed visually. Therefore, it is prone to large subjectivity, which can give rise to significant errors in accuracy and precision (Nutter Jr. et al., 2006).

Downy mildew severity in melon has been assessed mainly using descriptive score scales (Pan & More, 1996; Cardoso et al., 2002a; Santos et al., 2004). Considering that disease assessment must be easy and quick to perform in a broad range of conditions, and even then produce accurate, precise, and reproducible estimates (Campbell & Madden, 1990), the use of diagrammatic scales reduces subjectivity in severity estimates and is a good aid to the disease rater, once these scales are very functional.

Due to the lack of downy mildew quantification methods in melon that went through rigorous validation procedures, this work aimed at both developing a diagrammatic scale for assessing disease severity, and also analyzing the quality of the estimates it brought about.

MATERIAL AND METHODS

Development of the diagrammatic scale

To develop the diagrammatic scale, 100 melon leaves (cultivar AF-682), with distinct levels of mildew severity, were collected in an experimental field in Rio Largo County, State of Alagoas. Leaf images (200 dpi) were produced using the software Assess^â (The American Phytopathological Society, St. Paul, MN, USA). The percent of disease affected area was determined in each image and used as the disease severity level. A diagrammatic scale was then developed, with eight disease severity levels, based on the Weber-Fechner visual accuracy law (Horsfall & Cowling, 1978) and on the highest severity level observed in the collected leaves. In the scale, we tried to reproduce patterns, shapes, and the lesion scattering standard, as observed in the sampled leaves.

Validation of the diagrammatic scale

To validate the diagrammatic scale, we used colored photocopies of 50 melon leaves, with different levels of mildew severity. Eight agronomists (A to H), without experience on assessing diseases, were asked to estimate severity using the scale. To evaluate the reproducibility of the estimates, 15 days after the first evaluation, a new sequence of the same leaves was assessed by the same group, again using the scale.

Accuracy and precision of each rater were determine through simple linear regression, with the actual disease severity electronically obtained out of the images as the independent variable and the estimates of severity of each rater as the dependent variable. The accuracy of each rater and of the group of raters was determined applying the t test to the intercept of the linear regression (a), to prove the hypothesis Ho: a = 0; and to the line angular coefficient (b), to prove the hypothesis Ho: b = 1, both at 5% probability. The precision of the estimates was expressed by the regression determination coefficient (R²), by the variance of absolute errors (difference between the estimate and actual severity), and by the repeatability of the estimates, which, in its turn, was determined by the regression between the second and first disease evaluation over the same sample. Reproducibility was determined based on the R² values of linear regressions involving severity estimates from different raters, sorted in pairs, over the same sample (Nutter Jr. & Schultz, 1995). Regression analyses were carried out using the software Microsoft^â Office Excel 2003 (Microsoft Corporation, 2003).

RESULTS AND DISCUSSION

The highest severity value observed in the 100 harvested melon leaves was 96%. Therefore, the diagrammatic scale developed to assess disease severity consisted of the levels 2; 4; 8; 16; 32; 64; 82, and 96% of leaf affected area (Figure 1). For the severity levels from 32% above, symptom representation included rotted tissue and coalesced spots. Spot coalescence is characteristic of the melon downy mildew, and may progress to extensive areas of leaf rotted tissue (Rego & Carrijo, 2000; Tavares, 2002).

In the diagrammatic scale validation, the intercept of 37.5 and 50% of the raters in respectively the first and second evaluations significantly (p<0.01) differed from zero for the regression lines between the actual and the estimated severity. Most of these raters underestimated the actual severity (Table 1). The values of the line angular coefficient for 87.5% of the raters, in both evaluations, did not differed significantly from 1 (p<0.01) (Table 1), with no systematic deviations. Each 1% increment in the melon downy mildew severity assessed electronically corresponded to an increment of 0.85 and 0.87% in the severity estimated by raters, respectively in the first and second evaluation, attesting the trend for undervaluing. The tendency demonstrated by raters to underestimate severity levels of the melon downy mildew resembles what was observed on the validation of diagrammatic scales of the cassava brown leaf spot (Michereff et al., 1998), yam leaf blight (Michereff et al., 2000), Cercospora leaf spot in lettuce (Gomes et al., 2004), and grapevine bacterial canker (Nascimento et al., 2005).

In the precision analysis, the visual severity estimates in the first evaluation explained from 88 to 98% of the variation (R²) observed in the electronically disease assessment, scoring an average of 94%. In the second evaluation, visual estimates explained from 91 to 98% of the electronically assessed variation, with an average of 95% (Table 1). Most of the raters improved estimate precision from the first to second evaluation, which might indicate that as raters get better acquainted with the disease or are trained for disease assessment, estimate accuracy and precision are expected to improve (Michereff et al., 2000; Tovar-Soto et al., 2002). The precision of the visual estimates of downy mildew severity in melon with the aid of the scale was similar to those observed in recent studies for validation of diagrammatic scales (Barbosa et al., 2006; Godoy et al., 2006), getting fairly close to an ideal agreement, which denotes high precision (Kranz, 1988; Nutter Jr. & Schultz, 1995). Most of the raters reached excellent estimate repeatability when using the scale: in average, 94% of the variation in the first evaluation was explained by the second evaluation.

The absolute errors (residue) in both evaluations, when the diagrammatic scale was used, were concentrated around 10%, which, in addition to be fairly acceptable in evaluations of diagrammatic scales, can be reduced by training the raters (Nutter Jr. & Schultz, 1995; Nutter Jr. et al., 2006). The reproducibility of estimates of disease severity was higher than 90% in 89.3% of the cases in the first evaluation, and in 91.1% of the cases in the second evaluation (Table 2). Such a high reproducibility level is good evidence that the diagrammatic scale can be placed into practice to compare different experiments, carried out by different raters.

The difference among raters in the assessment of the melon downy mildew severity confirms the observations by Nutter Jr. & Schultz (1995) about the human distinct ability to discriminate disease levels. The quality of the disease estimate is influenced not only by psychological stimulus and response, but also by factors such as the complexity of the sample, lesion size and shape, color and number of lesions in a sample (Kranz, 1988), and rater fatigue and difficulty to keep the focus on the task (Shokes et al., 1987).

To put forward a proposal of a standard system to guide severity evaluation for a given disease is a considerable responsibility. If the system is defective, the costs related to its use may be heavier than the benefits it granted (Nutter Jr. & Schultz, 1995; Leite & Amorim, 2002). Nevertheless, standardization should be pursued. The use of a standard system for disease evaluation is the most effective way to allow the confrontation of results from different groups, institutions, and places (Bergamin Filho & Amorim, 1996).

The diagrammatic scale presented here to assess downy mildew severity in melon was very straightforward. At the same time, it was effective in providing a quick disease estimate, with reasonable accuracy and excellent precision and reproducibility. Therefore, we consider the diagrammatic scale to assess downy mildew severity in melon presented here to be a valuable tool for field surveys, epidemiological studies, and comparison among disease control methods.

ACKNOWLEDGEMENTS

Authors thank the raters who took part in the scale validation. Sami J Michereff holds a Research Grant from CNPq (The National Council for Scientific and Technological Development).

(Recebido para publicação em 18 de abril de 2008; aceito em 27 de fevereiro de 2009)

(Received in April 18, 2008; accepted in February 27, 2009)

• BARBOSA MAG; MICHEREFF SJ; MORA-AGUILERA G. 2006. Elaboração e validação de escala diagramática para avaliação da severidade da ferrugem branca do crisântemo. Summa Phytopathologica 32: 57-62.
• BERGAMIN FILHO A; AMORIM L. 1996. Doenças de plantas tropicais: epidemiologia e controle econômico São Paulo: Agronômica Ceres. 299p.
• CAMPBELL CL; MADDEN LV. 1990. Introduction to plant disease epidemiology. New York: John Wiley & Sons. 532p.
• CARDOSO JE; SANTOS AA; VIDAL JC. 2002a. Perdas na produção do meloeiro devido ao míldio. Summa Phytopathologica 28: 187-191.
• CARDOSO JE; SANTOS AA; VIDAL JC. 2002b. Efeito do míldio na concentração de sólidos solúveis totais em frutos do meloeiro. Fitopatologia Brasileira 27: 378-383.
• FNP. 2007. Agrianual 2007 - anuário da agricultura brasileira São Paulo: Instituto FNP. 532p.
• GODOY CV; KOGA LJ; CANTERI MG. 2006. Diagrammatic scale for assessment of soybean rust severity. Fitopatologia Brasileira 31: 63-68.
• GOMES AMA; MICHEREFF SJ; MARIANO RLR. 2004. Elaboração e validação de escala diagramática para cercosporiose da alface. Summa Phytopathologica 30: 38-42.
• HORSFALL JG; COWLING EB. 1978. Pathometry: the measurement of plant disease. In: HORSFALL JG; COWLING EB (eds). Plant disease: an advanced treatise - how disease develops in populations New York: Academic Press, v. 2, p. 119-136.
• KRANZ J. 1988. Measuring plant disease. In: KRANZ J; ROTEM J (eds). Experimental techniques in plant disease epidemiology Heidelberg: Springer-Verlag. p. 35-50.
• LEITE RMVBC; AMORIM L. 2002. Elaboração e validação de escala diagramática para mancha de Alternaria em girassol. Summa Phytopathologica 28: 14-19.
• MICHEREFF SJ; MAFFIA LA; NORONHA MA. 2000. Escala diagramática para avaliação da severidade da queima das folhas do inhame. Fitopatologia Brasileira 25: 612-619.
• MICHEREFF SJ; PEDROSA RA; NORONHA MA; MARTINS RB; SILVA FV. 1998, Escala diagramática e tamanho de amostras para avaliação da severidade da mancha parda da mandioca (Cercosporidium henningsii). Agrotrópica 3: 143-148.
• NASCIMENTO ARP; MICHEREFF SJ; MARIANO RLR; GOMES AMA. 2005. Elaboração e validação de escala diagramática para cancro bacteriano da videira. Summa Phytopathologica 31: 59-64.
• NUTTER JR. FW; SCHULTZ PM. 1995. Improving the accuracy and precision of disease assessments: selection of methods and use of computer-aided training programs. Canadian Journal of Plant Pathology 17: 174-184.
• NUTTER JR. FW; ESKER PD; COELHO NETTO RA. 2006. Disease assessment concepts and the advancements made in improving the accuracy and precision of plant disease data. European Journal of Plant Pathology 115: 95-103.
• PAN RS; MORE TA. 1996. Screening of melon (Cucumis melo L.) germplasm for multiple disease resistance. Euphytica 88: 125-128.
• REGO AM; CARRIJO IV. 2000. Doenças das cucurbitáceas. In: ZAMBOLIM L; VALE FXR; COSTA H (eds). Controle de doenças de plantas: hortaliças Viçosa: Universidade Federal de Viçosa. v. 2, p. 535-620.
• SANTOS AA; CRISÓSTOMO JR; CARDOSO JE. 2004. Avaliação de híbridos de melão quanto às principais doenças nos Estados do Ceará e Rio Grande do Norte Fortaleza: Embrapa Agroindústria Tropical. 14p.
• SANTOS AA; FREIRE FCO; LIMA JAA; CARDOSO JE. 2000. Doenças do meloeiro em áreas irrigadas no Estado do Ceará Fortaleza: Embrapa Agroindústria Tropical. 11p.
• SHOKES FM; BERGER RD; SMITH DH; RASP, JM. 1987. Reliability of disease assessment procedures: a case study with late leafspot of peanut. Oléagineux 42: 245-251.
• TAVARES SCCH. 2002. Doenças fúngicas - tecnologia no manejo de controle. In: TAVARES SCCH (ed). Melão - fitossanidade: aspectos  técnicos Brasilia: Embrapa Informação Tecnológica. p. 11-19.
• TOVAR-SOTO A; HERNANDEZ-MARTÍNEZ M; CRISTÓBAL-ALEJO J; ROMERO-HIJO R; MORA-AGUILERA G. 2002. Escala logarítimica diagramática de severidad de la mancha negra (Colletotrichum gloeosporioides Penz.) en Chirimoya (Annona cherimola Mill). Revista Mexicana de Fitopatología 20: 103-109.

Publication Dates

• Publication in this collection
  22 June 2009
• Date of issue
  Mar 2009

History

• Received
  18 Apr 2008
• Accepted
  27 Feb 2009