The Merlot and Chardonnay blocks were in the Oak Knoll District and the Cabernet Sauvignon block in Calistoga. Given the arrangement of three cultivars at two sites, cultivar and site effects are confounded in this design. Due to the close location of the two sites, these effects are analyzed as cultivar effects with the assumption that the site effects are not large. The Chardonnay and Merlot blocks were planted in 2008, on 101-14 rootstock, and trained to a bilateral system. The Cabernet Sauvignon block was planted in 2006, also on 101-14 rootstock and trained to a quadrilateral system. All blocks were cane-pruned .Following inoculation, the vines were monitored at 14-day intervals for the remainder of the 2017 growing season. Subsequently, in the 2018 and 2019 growing seasons, the vines were monitored from bud break through leaf fall. On each monitoring date, the vines were visually assessed for common symptoms of PD: stunted growth, chlorotic leaves, shriveled berries, leaf scorch, incomplete lignification, and matchstick petioles. Each symptom was recorded as present or absent on each of 5 areas of the vine: west and east, corresponding to the terminal ends of each cane; mid-west and mid-east, corresponding to the middle of each cane; and center corresponding to the head of the vine, plastic flower pots where the canes originate. Shriveled berries could not be evaluated after the harvest date for each block. For Cabernet Sauvignon, the harvest date for 2017 was October 27th, 2017 and for 2018 – 2019 the harvest was after the completion of our data collection.
For Merlot, the harvest dates were September 13th, 2017, September 25th, 2018, and October 8th, 2019. For Chardonnay, the harvest dates were September 13th, 2017, September 25th, 2018, and September 9, 2019. Three times annually , petioles were collected from all vines. The basal petiole from the terminal shoots as well as from the shoots adjacent to the trunk were collected at each time point . An additional sampling point was also measured in only July 2017 for finer resolution on the first year’s movement: shoot 2 was measured, the next shoot over from the inoculation site. For the quadrilaterally trained vines, eight samples were taken per plant instead of four, with additional designation as originating from either the south or north side of the vine.Petiole samples were analyzed via DNA extraction and quantitative polymerase chain reaction to quantify X. fastidiosa population size per gram of petiole tissue and population distribution across the vines. DNA was extracted from petioles using the DNeasy plant mini kit . All extractions were 5-fold diluted with RNAse free water before being quantified by qPCR wherein they were run using a primer pair targeting the gene encoding RecF, RecF1_F+R ; for qPCR protocol see Sicard et al. 2020. Following analyses with LinRegPCR , bacterial populations were interpolated using a standard curve. To produce the standard curve, we grew Napa1 cells on PWG media in a series of 10x dilutions and plated D0-D8 while also keeping the same volumes for qPCR. CFU/g were counted per streak of 20 μl and compared to the Ct values from qPCR to generate a slope to calculate CFU/g based on Ct values. qPCR was run in duplicate with positive and negative controls on each plate; any plates that had duplicates of more than 1 cycle apart were re-run.
All samples with a Ct value greater than 36 are marked as undetected.All analyses were performed using R statistical software version 4.1.2 . We used generalized linear models with binomial error to build several models. First we tested grapevine cultivar as a factor predictor for inoculation success . The only predictor value used was cultivar which included three levels . We also tested the effect of pathogen presence on symptom detection with a GLMM for each symptom with a random effect of plant identity. We then computed significance using an analysis of deviance with a chi-squared test statistic. Additionally, we built GLMMs with binomial error to test if the CFU/g in a dataset of only positive plants was predictive of each symptom’s presence or absence again with a random effect of plant identity. Most symptoms were tested across all three seasons except for stunting and leaf chlorosis, early season symptoms for which the 2017 year was not included due to spring inoculation precluding year 1 symptoms. For both sets of symptom models , the response variable was clustered by symptom monitoring dates around the three annual qPCR detection dates. The May qPCRs were paired with symptoms from spring , July qPCRs were paired with summer symptoms , and September qPCRs were paired with fall symptoms . If symptoms were not detected during these windows, then the response variable was 0. If symptoms were detected at any point, the response variable was 1. Winter recovery was analyzed as a binomial GLM, testing the effect of cultivar on plant recovery between 2017 – 2018. The response, plant recovery, was a binary variable in which plants that had been positive during the 2017 year were marked as either recovered or still infected. We then built a generalized linear mixed effect model with binomial error to test the effects of cultivar and year on binary infection status, with plant identity included as a random effect. Finally, we built a set of annual vine position models to test the effect of sampling location and cultivar on detection of the pathogen, with plant identity as a random effect.
For all GLM models we used the glm function in the stats package in R, and for all GLMM models we used the glmmTMB package . A survival analysis for symptom onset in 2017 was conducted using the Surv function in the survival package in R wherein symptom onset times were compared .In the Merlot and Chardonnay blocks, symptoms were first noted on July 21, 2017 , and included uneven lignification, leaf scorch and shriveled berries . Of these, the shriveled berries were the most consistent symptom across vines and within single vines. In the Cabernet Sauvignon block, the first symptoms were also noted on July 21, 2017. The only symptom noted on that date, and the following evaluation date, was shriveled berries. Leaf scorch and uneven lignification were not noted until August 28, 2017, for the Cabernet block. Stunted growth was noted on the first evaluation dates of 2018 and 2019 , coinciding with bud break and early shoot growth on Apr 10, 2018, and Apr 23, 2019 and Apr 25, 2018, and May 6, 2019 . In 2017, shriveled berries were the most common symptom; however, all other symptoms were also observable in the first year. The only symptom that was detectable in the control inoculated vines was stunting, which, in 2018 and 2019 was still much less common in the SCP controls compared to the pathogen-inoculated plants . These symptoms were analyzed with survival curves based on their times of emergence in 2017 and the cox proportional hazards models did not converge to compare symptoms to each other. A comparison of symptom emergence in controls and inoculated plants can be seen in the supplemental materials .The three cultivars showed different trends of symptom development over the three years of the study. Overall, Chardonnay exhibited the highest counts of symptoms including leaf chlorosis, leaf scorch, matchstick petioles, stunting, and shriveled berries. In Cabernet Sauvignon year 1, plastic garden container the prominent symptoms were shriveled berries and uneven lignification in the fall. In years 2-3, the few continuously infected plants exhibited all measured symptoms starting in July or August through the end of the growing season . The controls and experimental vines exhibited stunting in Cabernet Sauvignon in May and June . In Chardonnay, year 1 began with shriveled berries in July that peaked in September and exhibited subsequent leaf scorch and matchstick petioles in the fall. The following years showed high stunting, scorch, and chlorosis, all beginning in May when symptoms were first measured. There was uneven lignification detected in the summer and fall. Matchstick petioles increased throughout the fall in both years. For Merlot, July of year 1 showed shriveled berries and leaf scorch/chlorosis, followed by fall matchstick petioles. The following years showed stunting beginning in May similar to Chardonnay along with some leaf scorch, matchstick petioles, uneven lignification, and shriveled berries in the fall . With the data from three cultivars combined, annual trends emerged. Shriveled berries were the most informative symptom during the late season of the inoculation year. Although less abundant than shriveled berries, other symptoms in the first year included leaf scorch, matchstick petioles, uneven lignification, and stunting. In the second year, there was widespread stunting immediately observable after bud break, as well as leaf scorch which increased in August and September. Although many plants in this dataset had recovered by the third year, during 2019 there was predominantly early-season stunting along with some leaf chlorosis, followed by late season leaf scorch and matchstick petioles.
The established paradigm for PD of low recovery and leaf scorch in Napa Valley was challenged with this study. Our results indicate shriveled grape berries as the predominant early symptom, rapid spread of the bacterial pathogen through the grapevine, and high overwinter curing. Our findings suggest that previously overlooked initial symptoms are significantly associated with bacterial presence at the inoculation site. Leaf scorch is widely considered the most prominent and first visible symptom of PD. However, we showed that the most prominent symptoms in year 1 were shriveled berries and uneven lignification. While the symptom of shriveled berries has been previously observed , it has been largely ignored since then. These symptoms became visible in the year the vines were inoculated, near the point of veraison , as vines were starting to ripen fruit. In addition, shriveled berries were the only symptoms that occurred in all three cultivars at that time. Given the large amount of variability between cultivars, having a symptom that is consistent across cultivars may be essential for management. On the other hand, leaf scorch was observable in Chardonnay and Merlot, but was much less frequent in Cabernet Sauvignon. Uneven lignification was a prominent early symptom in Chardonnay, but not in the two other cultivars. Shriveled berries appeared more reliably and sooner after inoculation than foliar symptoms in all cultivars. Foliar symptoms tended to appear more consistently in grapevines that had been infected for over a year and may already be serving as sources of inoculum. Given the consistency of shriveled berries across the three wine grape cultivars, this appears to be the key symptom to visually detect early infections of X. fastidiosa, whereas other symptoms such as stunting, leaf scorch and uneven lignification arise later in the infection cycle. Symptom and pathogen movement through the plants were tracked throughout the project. In the first year, neither symptoms nor detectable bacterial populations extended past the trunk to the non-inoculated side of the plant . However, in years 2-3, symptoms and populations were evenly distributed on both sides of the trunk, even concentrated near the trunk, supporting that the infection favors the side of initial inoculation only in the first year. Although not explored in the current study, symptom distribution and pathogen movement may be affected by the manner of pruning. These plants were annually subjected to cane pruning, whereby the fruiting wood is retained as 8 to 15 node fruiting units . Annual growth is from buds on these canes, and the permanent vine structure consists of trunk and arms. In contrast, spur pruning retains shorter fruiting units distributed along the length of a cordon, another above ground permanent structure. Differences in the way in which fruiting wood is selected and retained between cane and spur pruned vines, as well as the overall proportion of permanent to annual vine structure may impact recovery rates and should be investigated further. Overwinter curing varied substantially between the three wine grape cultivars, but overall, it was greater than previously hypothesized, with two of the cultivars experiencing nearly 75% recovery of plants. Based on early-season inoculation and high bacterial population levels, these plants were expected to become chronically infected. The high recovery rates suggest that there might be substantially more plants exposed to X. fastidiosa infections than has been previously presumed. If many more plants are infected, but avoid chronic infection, we may be underestimating the amount of vector spread of this pathogen.