Materials and Methods

Two trials were conducted in a greenhouse in Mayagüez, PR, during February to April of 2012. These months were generally sunny and greenhouse temperatures varied from a low of ≈23 °C to a high of ≈35 °C. In trial 1, watermelon accessions (Plant Genetic Resources Conservation Unit, USDA-ARS, Griffin, GA) listed in Table 1 were seeded in 10-cm plastic pots filled with Promix PGX (Premier Tech Horticulture, Quakertown, PA) and arranged in a completely randomized design (CRD). The experimental unit was a pot with a single seedling. Each genotype was replicated six times. In trial 2, the same accessions were tested along with susceptible watermelon cultivars Crimson Sweet and Sugar Baby (Willhite Seed Company, Poolville, TX). Genotypes were again arranged in a CRD but replicated 10 times. PI 485580 did not germinate adequately in trial 1; only data from trial 2 are presented for this PI. For both trials, 14-d-old seedlings were mechanically inoculated with a Puerto Rico isolate of ZYMV (ZYMV-PR, GenBank accession number MN422959, Rodrigues Virology Collection No.161). In each trial, buffer-inoculated plants were included as controls. ZYMV-PR was originally collected from a single leaf of a single plant of Cucurbita moschata Duchesne in a 2008 survey in Puerto Rico and produced severe disease symptoms when inoculated to various cucurbit species. ZYMV-PR was multiplied on seedlings of C. moschata ‘Waltham’ [True Leaf Market (previously Mountain Valley Seed Company), Salt Lake City, UT] and the infected tissue was lyophilized and preserved for further use. As needed for this and other studies, ‘Waltham’ seedlings were inoculated with the original lyophilized material and fresh tissue from those plants were used as inoculum. Reverse-transcriptase polymerase chain reaction and sequencing of a coat protein gene fragment were conducted to monitor the stability of the isolate (GenBank accession MN422959). No meaningful variations were observed. Mechanical inoculation was done as follows: The first three true leaves of watermelon genotypes were inoculated 14 d after seeding; in a few plants, the third leaf was not sufficiently expanded to allow for inoculation, in which case only the first two leaves were inoculated. Leaves were lightly dusted with 350 mesh carborundum (Thermo Fisher Scientific, Waltham, MA). Fresh leaves of ZYMV-infected ‘Waltham’ were macerated in a cold mortar and pestle with 0.02 M phosphate buffer at pH 7.0. A tissue-to-buffer ratio of 1:10 (wt/vol) was used. At 20 d post inoculation, the first, third, fifth, and seventh leaves were visually evaluated for symptom severity using a 0 to 3 scale where 0 = no symptoms, 1 = mild chlorosis or mottling, 2 = strong chlorosis or mottling, and 3 = severe virus symptoms including blistering and leaf deformity. Ratings of 0 or 1 would be considered resistant. At 28 d post inoculation, a 0.1-g sample of tissue from the fourth leaf of each plant was sampled and tested for ZYMV using a commercial ELISA kit (Agdia, Elkhart, IN). The kit uses a double antibody sandwich assay and alkaline phosphatase label. The ZYMV ELISA from Agdia “detects isolates… from all over the world, including CT, USDA, SJBCA, CA, IT, NY, FL, Z18” (https://orders.agdia.com/pathoscreen-zymv-alkphos-psa-77700). Positive, negative, and extraction buffer controls were included on each ELISA plate. All controls were supplied by Agdia. Within a plate, controls and individual plant samples were replicated two times (two subsamples). Absorbance at 405 nm was measured with a Dias Microplate Reader (Dynex Technologies, Chantilly, VA). Individual absorbance values were adjusted by subtracting the average reading of the negative controls within a plate. Negative adjusted readings were changed to zero.

Table 1.

Watermelon accessions tested for resistance to a Puerto Rico isolate of Zucchini yellow mosaic virus (ZYMV). Accessions in the table were previously reported as resistant to ZYMV as cited in the last column of the table.

View Table

The absorbance reading of the fourth leaf (mean of two subsamples per plant); the symptom severity of first, third, fifth, and seventh leaves; the average severity of the first, third, and fifth leaves; and the average severity of the first, third, fifth, and seventh leaves were analyzed using InfoStat (Di Rienzo et al., 2019). Data from the buffer-inoculated control plants and, in the case of ELISA, from the positive, negative, and buffer controls were not included in the analysis of variance (ANOVA) to comply with the assumption of a common variance. Because there was a significant trial × genotype interaction, trials were analyzed separately using a one-way ANOVA. Means were separated with a Fisher’s protected test of least significant difference using α = 0.05. In addition, for each trial a Pearson’s correlation coefficient was calculated for each pair of variables used to assess resistance. Correlations were calculated using individual plant data (the mean of two subsamples per plant for ELISA readings, and a single observation per leaf for symptom severity at each leaf position).

Results

Although there was a significant trial × genotype interaction for the various measures of symptom severity, PI 537277 and PI 595201 were almost always among the accessions with the lowest severity, no matter which leaf was scored (Tables 2 and 3). Two other accessions, PI 595200 and PI 595203, were also often among the genotypes with the lowest severity, although less consistently so than the previously mentioned accessions.

Table 2.

Trial 1: Means of enzyme-linked immunosorbent assay (ELISA) absorbance readings (405 nm) in the fourth leaf and symptom severity of leaves at various positions in watermelon accessions mechanically inoculated (first three leaves, 14 d after seeding) with a Puerto Rico isolate of Zucchini yellow mosaic virus (ZYMV).

View Table

Table 3.

Trial 2: Means of enzyme-linked immunosorbent assay (ELISA) absorbance readings (405 nm) in the fourth leaf and symptom severity of leaves at various positions in watermelon accessions mechanically inoculated (first three leaves, 14 d after seeding) with a Puerto Rico isolate of Zucchini yellow mosaic virus (ZYMV).

View Table

Trial × genotype interaction was even greater for absorbance readings (ELISA) (Tables 2 and 3). The relative ranking of ELISA readings for genotypes varied between the two trials, except for PI 537277 and PI 595203. These two accessions were among the genotypes with the lowest ELISA readings in both trials. In trial 1, PI 595200 was among the genotypes with the lowest ELISA reading, but this same accession was among the genotypes with the highest ELISA reading in trial 2.

Correlations between assessment methods (symptom severity and ELISA) were always positive and generally significant, but some correlations were noticeably stronger than others (Table 4). Among measurements of symptom severity, correlations were much stronger between the fifth vs. the seventh leaf (r = 0.886 in trial 1 and r = 0.867 in trial 2) compared with the first vs. the seventh leaf (r = 0.442 in trial 1 and r = 0.607 in trial 2). When the severity scores of either the first to the fifth leaf or the first to the seventh leaf were averaged, this value was reasonably well correlated even with scores from the first leaf (r ≥ 0.542). Correlations between symptom severity measured in different leaves and absorbance in the ZYMV assay using the fourth leaf were much smaller and sometimes not significant. The strongest correlation was only r = 0.452.

Table 4.

Correlation coefficients between enzyme-linked immunosorbent assay (ELISA) absorbance readings (405 nm) in the fourth leaf and symptom severity of leaves at various positions in watermelon plants mechanically inoculated (first three leaves, 14 d after seeding) with a Puerto Rico isolate of Zucchini yellow mosaic virus (ZYMV).

View Table

To determine if the weak correlation between symptom severity and ELISA was a general phenomenon or specific to certain genotypes, we graphed absorbance against symptom severity for each genotype across both trials. Because results were similar for each leaf scored, and some genotypes had not expanded to the seventh leaf, we present only the graph for ELISA (sampled in the fourth leaf) vs. symptom severity in the fifth leaf (Fig. 1). Most watermelon genotypes showed a range of absorbance readings, from low (clearly negative for ZYMV) to high (strongly positive for ZYMV). The only genotype with all plants testing negative (absorbance < 0.200) was PI 537277. In addition, all plants of this genotype had a severity score of 0 in the fifth leaf (Fig. 1) and in all other leaves as well (data not shown). Most plants of PI 595201 and PI 595203 had both low ELISA readings and low symptom severity. However, there were exceptions. Some plants of these accessions had ELISA readings strongly positive for ZYMV, but almost all these cases were of plants with mild disease symptoms (severity of 0 or 1). Most plants of PI 595200 had a score of 0 or 1 for symptom severity, but some of these plants were strongly positive for ZYMV. The results for PI 494529 were similar to PI 595200, but more plants had severity scores of 2 or 3. The remaining accessions, as well as ‘Crimson Sweet’ and ‘Sugar Baby’, had highly variable ELISA readings and severity ratings, with little association between the two assessment methods. There were multiple cases of genotypes with plants with high absorbance readings (positive for ZYMV) but low severity (0 or 1). The reverse was also observed: 15 plants over the two trials had severity ratings of 2 or 3, but tested negative for ZYMV in the ELISA test.

Enzyme-linked immunosorbent assay (ELISA) absorbance readings (405 nm) in the fourth leaf and symptom severity in the fifth leaf of individual plants of watermelon accessions inoculated with a Puerto Rico isolate of Zucchini yellow mosaic virus (ZYMV). Genotype numbers refer to PI numbers of the U.S. Department of Agriculture Plant Genetic Resources System. CS = Crimson Sweet, SB = Sugar Baby. ELISA readings below the lower horizontal line are considered negative for ZYMV; readings above the upper horizontal line are considered positive for ZYMV. Some points on the graph cover other points with a similar or equal absorbance reading.

Citation: HortScience 55, 9; 10.21273/HORTSCI15166-20

• Download Figure

View Full Size

Enzyme-linked immunosorbent assay (ELISA) absorbance readings (405 nm) in the fourth leaf and symptom severity in the fifth leaf of individual plants of watermelon accessions inoculated with a Puerto Rico isolate of Zucchini yellow mosaic virus (ZYMV). Genotype numbers refer to PI numbers of the U.S. Department of Agriculture Plant Genetic Resources System. CS = Crimson Sweet, SB = Sugar Baby. ELISA readings below the lower horizontal line are considered negative for ZYMV; readings above the upper horizontal line are considered positive for ZYMV. Some points on the graph cover other points with a similar or equal absorbance reading.

Citation: HortScience 55, 9; 10.21273/HORTSCI15166-20

• Download Figure

Enzyme-linked immunosorbent assay (ELISA) absorbance readings (405 nm) in the fourth leaf and symptom severity in the fifth leaf of individual plants of watermelon accessions inoculated with a Puerto Rico isolate of Zucchini yellow mosaic virus (ZYMV). Genotype numbers refer to PI numbers of the U.S. Department of Agriculture Plant Genetic Resources System. CS = Crimson Sweet, SB = Sugar Baby. ELISA readings below the lower horizontal line are considered negative for ZYMV; readings above the upper horizontal line are considered positive for ZYMV. Some points on the graph cover other points with a similar or equal absorbance reading.

Citation: HortScience 55, 9; 10.21273/HORTSCI15166-20

• Download Figure

Discussion

In agreement with Guner (2004) and Guner et al. (2019), we found PI 535277 to be highly resistant to ZYMV (Tables 2 and 3; Fig. 1). This accession is classified as Citrullus colocynthis (L.) Schrad. a perennial species known as the desert watermelon (Levi et al., 2012). Among the tested plants of this accession, none exhibited disease symptoms. ZYMV was not detected in serological tests, suggesting the failure of virus infection and/or virus replication. In phylogenetic studies, both Levi et al. (2001) and Chomicki and Renner (2015) grouped accessions of C. colocynthis as quite genetically distinct from the other phylogenetic groups tested in our study. C. colocynthis can be crossed with watermelon only with difficulty (Levi et al., 2002). Therefore, deploying PI 533277 in a watermelon breeding program for ZYMV resistance will present challenges.

All other PIs showing good levels of ZYMV resistance in our screening belonged to either C. lanatus (PI595200, PI 595201) or the sister species [according to Chomicki and Renner (2015)] Citrullus mucososperma (PI 595203). PI 595203 should be of interest to watermelon breeders because it has been shown to also be resistant to both ZYMV-CH (Xu et al., 2004) and ZYMV-FL (Guner 2004; Guner et al., 2019; Ling et al., 2006). Guner (2004) found this PI to be the most resistant of 1644 cultigens tested. PI 595203 has been shown to have a single nucleotide polymorphism mutation in the eIF4E genomic region of watermelon closely associated with resistance to ZYMV (Harris et al., 2009; Ling et al., 2009). USVL-370, a breeding line with PI 595203 as the source of resistance to ZYMV-FL, was recently released (Levi et al., 2016).

The remaining PIs tested showed strong disease symptoms and/or high ELISA readings even though previous studies had shown them to be resistant. Provvidenti (1991) found a single recessive gene for resistance to ZYMV-FL in PI 483361, and Boyhan et al. (1992) confirmed its resistance. But like Guner and Wehner (2008), our study did not confirm this resistance. In fact, we found this accession to be among the most susceptible (Tables 2 and 3; Fig. 1). Likewise, we found PI 482276, PI 485580, PI 494528, PI 560016, and PI 5956662 to be susceptible to ZYMV-PR, whereas previous studies had found these accessions to be resistant (Boyhan et al., 1992; Guner, 2004; Guner and Wehner, 2008; Guner et al., 2019). Possible explanations include a differential response to ZYMV isolates, with these PIs being resistant to ZYMV-FL but not to ZYMV-PR. Another possible explanation is that, for a specific accession, the genetic composition of plants tested in our study differs from those in previous studies. Watermelon is assumed to be a highly cross-pollinated crop (Kumar et al., 2013). Thus, many accessions in the USDA collection are likely to be a composite of multiple genotypes.

How ZYMV-PR compares with ZYMV isolates used in previous screening studies is difficult to determine because tracing back previous isolates is problematic. ZYMV-PR produces severe symptoms in susceptible cucurbits. Guner (2004) described the isolate ZYMV-FL he used for screening as one producing severe symptoms and deriving from a subculture of isolate 2088 received from E. Hiebert at the University of Florida. Therefore, ZYMV-FL used in Guner (2004), Guner and Wehner (2008), and Guner et al. (2019) was likely the same as the isolate designated “SV” (for “severe”) in the University of Florida study by Wisler et al. (1995). The screening studies of Provvidenti (1986, 1991) do not specify where he obtained his isolate of ZYMV-CT and ZYMV-FL, but it is likely these isolates were collected in Connecticut in 1982 and Florida in 1983, respectively, and reported in Provvidenti et al. (1984). The screening by Boyhan et al. (1992) used ZYMV-FL obtained from R. Provvidenti. The inheritance study by Xu et al. (2004) used an isolate from R. Provvidenti who confirmed it to be ZYMV-CH. Confirmation was likely done using serology, line differentials, and/or biological reaction, not sequencing. Wisler et al. (1995) described the P1 protein of SV (=ZYMV-FL) and deposited the sequence as GenBank: L35589.1 (“Florida severe”). No genomic comparison can be made between ZYMV-FL and ZYMV-PR because the latter was characterized using the coat protein (CP) and not the P1 protein. The CP region of the genome is the most widely used for Potyvirus (Adams et al., 2005; Shukla and Ward, 1988). The ZYMV-PR CP gene fragment showed a high percentage of identity to ZYMV isolates recently reported in squash in Iran (98.17%), Turkey, Venezuela, and Egypt, and in bees in Australia (97.72%) (Roberts et al., 2018). Our work appears to be the only reported watermelon screening study in which the ZYMV isolate used has had its CP sequence clearly documented. Considering the advances in sequencing and the importance of this virus to cucurbits in general, it would be useful to obtain the full genomes of ZYMV-FL, ZYMV-CT, and ZYMV-PR to better compare these isolates.

The response of plants exposed to virus can be expressed in terms of the behavior of the virus within the plant and in terms of the disease response of the plant (Bos and Parlevliet, 1995; Cooper and Jones, 1983). The ideal assessment method from the plant breeder’s perspective is one that is easy to carry out, reliably indicates the virus’s ability to infect and/or replicate within the plant, consistently measures the disease response of the plant, and can be obtained at the earliest moment possible in the life cycle of the plant. In this study, symptom severity assessed on either the first or third leaf of a seedling was often not indicative of degree of severity on later leaves. This was especially the case when the first few leaves showed few or no symptoms. However, once a leaf exhibited strong disease symptoms in any leaf, the following leaves generally expressed the same or greater symptom severity (data not shown).

ELISA readings were not reported in previous studies assessing resistance among watermelon accessions (Boyhan et al., 1992; Guner, 2004; Guner and Wehner, 2008; Provvidenti, 1986, 1991), although they were used in studies of inheritance of ZYMV resistance in PI 595203 (Ling et al., 2006; Xu et al., 2004) and for assessing presence or absence of virus in individual accessions in Guner et al. (2019). We believe that ELISA absorbance readings offer an additional tool for assessing virus resistance in watermelon accessions. The simultaneous use of more than one technique to assess a plant’s response to a virus should be encouraged among plant breeders. In some cases, we observed plants with severity ratings of 0 or 1, but strongly positive absorbance readings. This suggests that virus replication was occurring within these plants without the usual development of symptoms and damage to plant tissue. The reverse situation was also observed in those plants with strong symptoms but with ELISA readings negative for ZYMV. To rule out the possibility of accidental contamination with Papaya ringspot virus (PRSV), these plants were tested by ELISA for PRSV. Together with ZYMV, PRSV is the other most common cucurbit virus in Puerto Rico (Paz-Carrasco and Wessel-Beaver, 2002). All readings for PRSV were negative in these symptomatic plants. Astier et al. (2007) points out that virus titer within a plant varies over time and in different parts of the plant. Serological tests sometimes fail to detect the virus in highly symptomatic leaves. Nevertheless, serological assays potentially offer the breeder an easy means of quantifying the degree to which a virus has replicated in a plant. The ideal ZYMV resistant genotype would be one that limits the replication of a virus within a plant and minimizes the intensity of disease symptoms when subjected to a variety of strains of the virus. Genotypes showing resistance to multiple virus strains, such as PI 595203, should rank high as sources of resistance.