Introduction

Potato virus Y (PVY; family Potyviridae, genus Potyvirus) is a widespread agricultural virus causing substantial losses of potato crops around the world. PVY exists as a complex of at least nine distinct strain groups, including the parental non-recombinant strains PVYO, PVYC and PVYN, and several recombinant strains which co-occur in varying proportions in all major potato growing regions of the world (Chikh-Ali et al. 2013; Kehoe and Jones 2016). Recombinant strains of PVY were only discovered in recent decades, and first reported in North America from Canada in 1991 (Singh 1992). Recently, these recombinant strains, particularly PVYNTN and PVYN:O, have quickly risen in the population to overcome the traditional PVYO strain around the world (Schubert et al. 2007; Chikh-Ali et al. 2010; Djilani-Khouadja et al. 2010; Gray et al. 2010; Karasev and Gray 2013; Davie et al. 2017). These new strains are not only spreading quickly, but they also display different symtomology than PVYO, including cryptic foliage symptoms hampering identification, and tuber necroses in many susceptible varieties when infected with PVYNTN. In ongoing surveys of PVY strains in New Brunswick, Canada from 2009 to 2016, PVYO has been declining rapidly relative to a rise in PVYNTN and more recent rise in PVYN:O(Nanayakkara et al. 2012, this study).

In the field, PVY is transmitted between potato plants and several other crop and weed species by aphid vectors. PVY can be spread by a wide range of aphid species (Pelletier et al. 2012), but in a non-persistent manner, meaning that after aquisition of PVY from an infected plant, the aphid only remains viruliferous for a brief time and only capable of infecting a few other plants (Fereres and Moreno 2009; Wróbel 2009).

The recent rise of recombinant PVY strains in many potato growing regions of the world has lead to an interest in whether these strains have some intrinsicly greaterability to spread between plants. Several recent reports have begun to show potentially important mechanisms which may be associated with differential infectivity of various strains (Verbeek et al. 2010, Mello et al. 2011, Mondal et al. 2016, Mondal and Gray 2017). Possibile mechanisms accounting for greater spread of these PVY strains may include: strain-specifc hypersensitive response (HR) in some potato varieties which preferentially resists PVYO transmission (Gray et al. 2010; Funke et al. 2017), more efficient transmission of recombinants between plants by aphids, greater rates of tubers carrying PVY following primary infection, greater likelihood of harvest, storage and planting in next season’s crop, typically more cryptic foliar symptoms in these strains compared to PVYO which may reduce roguing success, along with several other potential factors affecting the strains differently. In the 2014 through 2016 crop seasons, our research group conducted a number of PVY trials across the potato growing regions of New Brunswick and Manitoba, Canada with Goldrush and Russet Burbank potato varieties. These varieties either show strong HR including localized necrosis and leaf drop in respose to PVYO in Goldrush, or only mild mosaic symptoms similarly expressed across PVY strains in Russet Burbank (Nie et al. 2012). HR to PVYO is mediated by Ny, conferring partial resistance to the strain, which is not uncommon in North American potato varieties (Rowley et al. 2015).

All of these field trials were planted with inoculum containing the three major PVY strains present in the region (PVYO, PVYN:O and PVYNTN) in varying population ratios. The spread of these strains to initially virus-free plants was measured, as well as the rates of infection of individual tubers from those plants. These data were analysed to shed light on potential mechanisms to explain strain-related differences in spread in the PVY-aphid-potato pathosystem.

Materials and Methods

Field Trial Setup and Sample Collection

From 2014 to 2016, five controlled and replicated experimental trials using Goldrush and Russet Burbank varieties were undertaken to measure the effects of various management techniques and virus strains on spread of PVY in the field. These are among the most common of over 200 varieties grown in New Brunswick for seed production, together amounting to 27% of total acreage planted between 2014 and 2016. In 2014, commercial Goldrush and Russet Burbank seed lots, sourced from New Brunswick with relatively high levels of PVY, were used to plant both trials. PVY strains in these seed lots represented all three main strains present in the province, though in slightly different proportions and dominated by PVYNTN. In 2015 and 2016, however, field trials were planted with very-low to zero PVY seed, and proportions and spatial distribution of each PVY strain were controlled by hand-planting known infected tubers harvested from the 2014 (for planting in 2015) and 2015 (for planting in 2016) trials. Thus, ultimately, all PVY inocula used through 2014 to 2016 originated from New Brunswick irrespective of the location of the field trial. Details of potato variety, trial location and inoculum, and subsequent spread of PVY are given in Table 1.

Table 1 Summary of PVY trials conducted in seasons 2014 through 2016, including potato variety, location, initial PVY inoculum level and strains and PVY spread to test plants (initially virus free plants marked and tracked through the season)
Full size table

In the different trials, plot size, potato variety, management treatments, number of inoculum tubers and number of virus-free test plants varied, but the general procedure to sample plants, quantify PVY spread through the trial plot and identify PVY strains in newly infected plants was identical. In 2014, 1000–1500 potato plants chosen approximately equidistantly across the study field were marked with a unique survey flag and had a leaf sample taken shortly after emergence. Samples testing positive for PVY at this early stage of development, approximately 4 weeks after planting, were considered seed-borne infections and categorized as part of the PVY inoculum in the field. All other flagged plants testing negative were categorized as “test plants” and were monitored through the season up until harvest to quantify spread of PVY. Similarly in 2015 and 2016, 1000 test plants were flagged and monitored through the season in each field, however, essentially all of these plants were virus-free test plants as these trials were planted with near-zero PVY levels. Of all flagged plants in 2015 and 2016 (3000 across all trials) only 4 test plants were found with seed-borne PVY, and they were immediately rogued from the field after detection. Inoculum in these fields was provided by hand planting known-infected tubers (tested for PVY status and strain identity prior to planting) among the clean seed, separately marking the location of each inoculum tuber. Known infected tubers were planted in a square pattern, alternating between strains, to ensure the identical and even distribution of viral inoculum and strain in each plot. In rare cases when infected seed failed to emerge, the tuber was replaced by a greenhouse-raised transplant from the same seed source. Plants from all inoculum tubers were tested to confirm infection status and strain at this stage.

Approximately 3–4 weeks after initial leaf sampling in each trial, all flagged test plants were again sampled to quantify PVY spread to mid-season. After a further 4–5 weeks, following top-kill of the potato crop, three tubers from each test plant were dug and stored in individually identifiable bags for later PVY testing. All samples testing positive for PVY, at initial or mid-season leaf sampling or post top-kill tuber harvesting, were subsequently tested to determine PVY strain.

PVY Detection and Strain Determination

PVY was detected in leaf and tuber sprout samples using the double-antibody sandwich enzyme-linked immunosorbent assay (DAS-ELISA) method of Singh et al. (2003). Sometimes, collected leaf samples were kept refrigerated up to three days prior to ELISA testing. Tuber samples were stored for ca. 2 months, then dormancy was broken using rindite (Gugerli and Gehriger 1980) and two sprouts/tuber were tested by ELISA similarly to leaf samples. Sap from PVY-positive leaf or tuber samples was extracted with Tri-Reagent (Molecular Research Center, Inc., Cincinnati, OH)following manufacturer’s instructions. Following RNA extraction, complementary DNA (cDNA) was synthesized as described by Nie and Singh (2001), then a strain-specific multiplex PCR reaction was performed following Lorenzen et al. (2006) with products resolved by electrophoresis through a 2.5% agarose gel in 0.5× Tris-Borate-EDTA buffer. A subset of samples identified to strain by this Lorenzen PCR method were tested by the multiplex strain differentiation protocol of Chikh-Ali et al. (2010, 2013) and by cDNA sequencing to more narrowly characterize PVY strain. A region encompassing the coat protein was amplified using primers 5’-GCTTTCACTGAAATGATGGT-3′ and 5’-TTTTTTTTTGTCTCCTGATTGAAG-3′ (Nie and Singh 2003) then sequenced by The Centre for Applied Genomics of Sick Kids Hospital (Toronto, Canada). Resulting sequences were compared to the comprehensive list of PVY genomes published by Green et al. (2017) to establish precise strain affiliation.

Statistical Analysis

Statistical tests on differences of tuber yield and number between infections of the three PVY strains within each variety were performed with ANOVA and post-hoc t-tests using R (version 3.4.1 “Single Candle”, 30 June 2017) (R Core Team 2017).

Results

Recent Changes in Proportions of PVY Strains in NB, Canada

Recently, there has been a dramatic shift in proportions of the three common PVY strains, PVYO, PVYN:O and PVYNTN in New Brunswick. When collected in 2010, samples were dominated by the traditionally common PVYO strain, representing ca. 80% of all PVY infections at the time (Fig. 1). By 2014, PVYO proportion had dropped more than three-fold, being almost exclusively replaced by PVYNTN, which rose from 12% to 67% between 2010 and 2014. Since 2014, PVYO has continued to drop to less than 14% of the population with this remainder being displaced by a doubling of PVYN:O abundance, while PVYNTN remained dominant but stable.

Fig. 1
figure 1

Changes in proportions of PVY strains PVYO (circle, solid line), PVYN:O (square, dashed line) and PVYNTN (triangle, alternating dashed line) in annual surveys of infected plants and tubers from New Brunswick, Canada, 2010 to 2016

Full size image

A subset of samples from 2015 and 2016 identified by PCR as belonging to the three different strains were tested by the improved multiplex PCR method of Chikh-Ali et al. (2010) and cDNA sequenced in the coat protein region of the genome to better characterize strain affiliation. These analyses confirmed that the three strains belonged to the PVYO (identical to PVYO isolate ‘WI120092’ accession KY848029), PVYN:O (99.9% similar to PVYN:O isolate ‘ND23’ accession KY847997) and PVYNTNa (identical to PVYNTN isolate ‘ME4’ accession KY847973) strains according to Green et al. (2017). .

Differential Spread of PVY Strains during the Crop Season

Trials with Goldrush and Russet Burbank planted in 2014 had initial PVY inoculum levels of 2.3% and 3.3%, respectively, which was relatively high in recombinant strains of PVY representative of the strain proportions seen among the commercial fields in the province. The strain composition of PVY inoculum in two trials consisted of 9% PVYO, 15% PVYN:O, 76% PVYNTN(Gold rush) and 19% PVYO, 19% PVYN:O, 62% PVYNTN (Russet Burbank). PVY spread to 10.3% and 36.8% of initially virus-free test plants in the Goldrush and Russet Burbank trials, respectively (Table 1). Despite starting with relatively high proportions especially of PVYNTN, newly infected plants were dominated by an even greater proportion of PVYNTN (84% in Gold rush, 82% in Russet Burbank). Proportions of PVYO and PVYN:O generally dropped, especially PVYN:O in Russet Burbank, which dropped from 19% of inoculum to 5% of new infections.

In all trials, mid-season leaf samples taken approximately half-way through the growing season (late July) showed some PVY spread to virus-free test plants. While the predominant strain in these early detections was the same as that post-harvest (PVYNTN), the rate of PVY detected at this time, however, was an order of magnitude lower than that found in the tubers after harvest (data not shown). Presumably the low rate of detection of PVY at mid-season is due to slow transduction of the virus from the site of infection through to the rest of the plant, and thus the majority of recent infections likely went undetected at this time.

In 2015 and 2016, trials were planted with more consistent levels and strain proportions of PVY inoculum to quantify strain-specific differences in PVY spread in a more controlled manner. Fields with 0% PVY were hand-planted with known PVY infected tubers in a consistent pattern across each field, to 3% overall inoculum with equal proportions of all three strains planted in the 2015 and 2016 Goldrush trials, and 2.5% with a 30:20:30% ratio of PVYO, PVYN:O and PVYNTN in the 2015 Russet Burbank trial (Table 1). Fewer PVYN:O tubers were planted in the latter trial due to a limited supply of infected tubers with known PVYN:O status from the previous year’s harvest. As in 2014, PVY spread to a large number of the initially virus-free test plants in 2015 and 2016, and these new infections were again dominated by PVYNTN. Indeed, PVYNTN spread to five to seven times as many plants as PVYO across the three trials, despite being equal in proportion in the inoculum plants. PVYN:O also spread to relatively more plants than PVYO in these latter trials, when accounting for variation in inoculum proportion. In the 2015 and 2016 Goldrush trials in which it was planted proportionately equal to PVYO, it spread to two to four times as many plants as PVYO, and in the 2015 Russet Burbank trial, in which PVYO inoculum was twice that of PVYN:O, PVYO only infected about 1.6 times as many plants. These data, summarized in Table 2, clearly demonstrate that plant-to-plant spread of PVYNTN is most effective, followed by PVYN:O then PVYO in these two important potato varieties.

Table 2 Changes in PVY strain proportions from inoculum to new infections of test plants, proportion of tubers infected in test plants and relative infection potential of inoculum plants of each strain. The percentage of spread of each strain to test plants represents the proportion of all occurrences of each strain determined after harvest in the initially virus free plants compared to the total of all three strains. Occasionally, two strains were found co-infecting a single test plant; these were counted as two separate infection events contributing to the total spread of PVY, regardless of them happening to occur to one plant. See text for calculation of “Infection potential relative to PVYO”
Full size table

Strain-Specific Effects on Tuber Size, Yield and Distribution of PVY in Tubers

PVY infection showed variety- and strain-specific effects on yield and number of tubers produced by potato plants in these trials. After top-kill and harvest, uninfected Russet Burbank plants averaged more (ca. 15 versus 12) tubers per plant though less overall yield (ca. 1800 g versus 2400 g per plant) of tubers than Goldrush (Fig. 2). Goldrush plants infected with PVY showed some weak variability by strain, with PVYN:O and PVYNTN plants similar to uninfected controls while PVYO plants had slightly fewer tubers and lower overall yield, though only the difference in yield was statistically significant (t-test, p = 0.003) (Fig. 2a,c). Strain-specific differences in Russet Burbank plants were more complex, with PVYO and PVYN:O tuber numbers similar to uninfected control and PVYNTN significantly fewer than control (t-test, p = 0.03). Infection by any strain showed significantly less yield than control plants, though PVYO and PVYNTN were similar and more severe than PVYN:O (t-test, p < 0.01 all strains) (Fig. 2b,d). One peculiarity in Russet Burbank infected with PVYO, however, was that while the total number of tubers per plant was nearly equal to control, the size distribution of them was dramatically different. Very small tubers (less than 60 g or ca. 2 oz.) less likely to be harvested or survive to planting next season, were over 50% more numerous in PVYO than control, and nearly double that of PVYNTN plants. Similarly, the number of more marketable large tubers (>120 g or ca. 4 oz.) of PVYO were nearly ten-fold fewer than control, less than one per plant on average, and about three-fold and four-fold fewer than PVYNTN and PVYN:O, respectively. No such remarkable difference with the size spectra of tubers occurred in Goldrush plants, with numbers of large and small tubers generally reflecting the total tuber number when compared across the three strains.

Fig. 2
figure 2

Number and yield of tubers per Goldrush (A, C) and Russet Burbank (B, D) plant in uninfected plants and plants infected with PVYO, PVYN:O and PVYNTN. Displayed are means ± SEM; letters above values indicate significance groups determined by ANOVA with post-hoc t-test (threshold p = 0.05)

Full size image

After harvest, storage and sprouting, not all tubers from an infected plant had detectable PVY. On average, approximately 73% of Gold rush and 60% of Russet Burbank tubers showed PVY in sprouts detectable by ELISA (Table 2). While there appears a marked difference in rates of tuber infection between Goldrush and Russet Burbank varieties, this may be exaggerated somewhat, as the extraordinary PVY spread in the 2016 Goldrush trial suggests that many plants were infected two or more times, which likely increased the rate of tubers bearing detectable PVY. Nevertheless, when averaging only the 2014 and 2015 trials the Goldrush plants still showed 67.5% of tubers infected, more than 7% greater than Russet Burbank. Looking at rates of tuber infection by PVY strain within each variety, it is not convincing that there was a substantial or significant strain-specific effect. While, on average, PVYNTN showed the highest rate of tuber infection in each variety, the statistical significance of differences between strains in each trial varied (see Table 2) and when averaged within varieties, was only about 4% (Gold rush) or 7% (Russet Burbank) higher for PVYNTN than PVYO.

By quantifying the number of inoculum plants of each PVY strain, the number of newly infected plants of each and the number and proportion of infected tubers of those, it is possible to calculate the relative infection potential of each strain in terms of production of infected progeny tubers. For both varieties, individual PVYNTN-infected inoculum plants had far greater potential for passing on the virus to the harvested tubers. For each PVYNTN inoculum plant, nearly 7 times as many Goldrush and over 3 times as many Russet Burbank tubers were infected relative to PVYO. Interestingly, there was a great difference in the relative infection potential of PVYN:O between the potato varieties. While in Goldrush, PVYN:O-infected inoculum plants would spread to produce 2.5 times the number of infected tubers as a PVYO-infected plant, in Russet Burbank that number was slightly lower (0.79) that of PVYO. Most of the difference in the infection potential of the strains relative to PVYO was due to differences in plant-to-plant transmission (at least 80% of effect) rather than transduction of virus from infected plant to developing tubers (less than 20%).

Discussion

In most potato growing regions of the world in the past 10 to 30 years, including New Brunswick, Canada, the traditional strain PVYO has been rapidly replaced by recombinant strains of the virus. Since their discovery several decades ago, recombinant PVY strains such as PVYN:O, PVYN-Wi and PVYNTN have quickly grown to dominate PVY populations in different regions. Replacement of PVYO, primarily by PVYNTN, began occurring about 20 to 30 years ago in continental Europe (Schubert et al. 2007) and the UK (Davie et al. 2017), the mid-east and northern Africa (Djilani-Khouadja et al. 2010) and Indonesia (Chikh-Ali et al. 2016) among other regions. Recombinant strains were discovered in North America in the 1990’s (Crosslin et al. 2006, Lorenzen et al. 2006a) and they grew to begin displacing PVYO there by the late 2000’s (Gray et al. 2010, Karasev and Gray 2013, this study). Most recently, in the important potato producing Columbia Basin of the USA, PVYO dropped from 63% to 7% of the strain population from 2011 to 2015, with replacement mainly by PVYN-Wi (53%) and less so PVYNTN (24%) and PVYN:O (9%) (Funke et al. 2017).

In New Brunswick in recent years, average incidence of PVY has dropped dramatically - ca. 12-fold from 2010 to 2016 in post harvest virus testing - likely due to the establishment of a stringent seed certification, and development and adoption of evidence-based best management practices for crop management (Fageria et al. 2013; MacKenzie et al. 2014, 2016, 2017). In 2009 (Nanayakkara et al. 2012) and 2010 (this study), PVYO clearly dominated at 82% of the PVY population. Although there has been an overall drop in PVY levels in New Brunswick in recent years, proportionally, PVYO has given way to recombinant strains here as it has in other regions. Here though, it has largely been replaced by PVYNTN (64% of all PVY in 2016) and to a lesser degree PVYN:O (22%).Results from our study are more similar to those observed over the past two decades in Scottish seed production fields, in which PVYNTN became dominant over PVYO, with a relatively low population of PVYN:O (Davie et al. 2017). Although in that region, the non-recombinant NA-PVYNTN was also a major component of the PVY population, this strain was absent in New Brunswick. Aside from variations in the time of arrival of different strains, it is unknown what other factors may precipitate the rapid population shift to recombinant strains in these different regions. Different species of aphids can transmit different PVY strains with varying efficiencies. Within New Brunswick a recent study showed that over two years at least 65 different species, mostly non-colonizers of potato, were found carrying PVY in their mouthparts (Pelletier et al. 2012). We are unaware of evidence that changes in aphid abundance or species is responsible for shifts in PVY strain populations in New Brunswick or any other region. Indeed, within our field trials, proportionately similar shifts in strain proportions occurred during the growing season in widely separated field trials over multiple years, despite substantial inter-annual and regional differences in aphid populations.

During the current study period, traditional non-recombinant strains (other than PVYO) such as PVYN and PVYC were not detected at all in New Brunswick, similar to other regions where they have also become rare or absent. While the Lorenzen et al. (2006) method was not designed to identify PVYC, the primers used in that method would indeed amplify a single 266 bp product similar to PVYO, but lack the corresponding 689 bp product. No such peculiar band pattern was encountered in our samples, and from the group of samples sequenced, nothing remotely similar to published PVYC sequences were found.

Our results from five controlled experimental field trials, in three field seasons and with two different potato varieties, clearly show that the recombinant strains PVYN:O and even more so PVYNTN spread more quickly than PVYO during the growing season in most cases. Several mechanisms that could contribute to this include strain-specific PVY resistance in the recipient plants, differential efficiency of aphid acquisition, retention or transmission of strains of the virus, differences in viral titre in the inoculum plants or translocation ability in the recipient plants. Other cultural or management factors may also play a role in the in-field or season-to-season spread of different PVY strains based upon the visibility of infected plants in the field.

Plants of both varieties show strongest foliar symptoms when infected with PVYO, including substantial stunting and early senescence, whereas PVYN:O and PVYNTN-infected plants showed little effect of infection (Nie et al. 2012). Anecdotally, it was noticed in our field trials that PVYO infected inoculum plants tended to grow more slowly and senesced earlier than plants infected with PVYN:O or PVYNTN, or uninfected plants. Later in the season, PVYO plants had far less foliar area available to aphids to land or feed upon, and were often completely overgrown and obscured by neighboring uninfected plants. It is possible that this effect could reduce the rate of acquisition of PVYO by aphids relative to other strains and contribute to the observed differences in strain-specific spread, particularly later in the season. In this study, however, we did not quantify these differences in plant growth or senescence with infection by the different strains.

Data from our controlled field trials can isolate some of these factors relating to differential spread of PVY strains, and clearly show that strain-specific differences in PVY spread occurred primarily at the plant-to-plant transmission stage. While PVY affects the yield and number of tubers, and does not infect all tubers in a primary-infected plant (Fageria et al. 2013; MacKenzie et al. 2014), differences in these rates between PVY strains is small relative to differences in the transmission from inoculum plant to recipient plant. Thus, the majority of the difference in transmission between strains in our study (ca. >80% of effect) occurred during transmission, whether at acquisition and/or infection.

Recently, a number of studies have investigated strain-specific differences in infection rates under standardized conditions, with particular interest in how genetic resistance to PVY infection in some potato varieties may play a role in population-scale shifting of PVY strain proportions (e.g. Tian and Valkonen 2015; Funke et al. 2017; Davie et al. 2017). Many potato varieties exhibit a HR upon exposure to PVY, which halts replication or transduction of the virus in the plant. HR induction by PVYO is mediated by Ny in the potato, which interact with the HC-Pro region of the PVYO genome; recombinant strains gaining HC-Pro from the PVYN parent do not trigger the Ny-conferred HR response (Tian and Valkonen 2015). Thus, the HR resistance mechanism is preferentially more effective against PVYO, and restricts its transmission in these varieties more so than recombinant strains like PVYN:O and PVYNTN. In the USA, this was convincingly shown with the HR varieties Alturas, Umatilla Russet and Ranger Russet, in which transmission rate of PVYO was two to three times lower than PVYN-Wi or PVYNTN, compared to similar transmission rates across all strains in the non-HR variety Russet Burbank (Funke et al. 2017). In the UK, Davie et al. (2017) observed a similar shift toward serotype-N PVY, later identified as being mostly Eu-PVYNTN, from 1993 to 2015; field trials in their study showed that transmission efficiency was highest in the recombinant Eu-PVYNTN, followed by non-recombinant NA-PVYNTN and lastly PVYO.

In our study, both Goldrush and Russet Burbank varieties demonstrated strain-specific differences in PVY transmission, even though PVYO elicits HR only in Goldrush. In both varieties, PVYNTN spread far more effectively than PVYO, as did PVYN:O in Goldrush trials. In Russet Burbank trials, however, the relative transmission rate of NTN was only about 3-fold greater than PVYO compared to nearly 7-fold in Goldrush. Also in Russet Burbank, and unlike Goldrush, PVYN:O transmission was actually lower than PVYO. Mondal et al. (2016) showed significant differences in transmission efficiencies of PVY strains between Russet Burbank plants in controlled greenhouse studies. These trials included both potato colonizing and non-colonizing aphids and showed highest transmission efficiency with PVYNTN, followed by PVYO and PVYN:O in the same pattern as observed in the field trials of the present study. Carroll et al. (2016) also observed more efficient transmission of PVYNTN than PVYO by Myzus persicae in both HR and non-HR varieties, while Mello et al. (2011) found no difference in transmission between PVYN:O and PVYO in Yukon Gold potatoes by five aphid species.

Relatively less-studied have been the strain-specific differences in virus acquisition prior to transmission, and virus multiplication and transduction in recipient plants following infection. Mondal and Gray (2017) have shown that PVYNTN and PVYN:O are preferentially transmitted over PVYO in greenhouse studies in which aphids were allowed to acquire multiple strains of the virus concurrently. This preferential transmission was independent of the order in which the virus strains were acquired by the aphid, and may indicate a competitive ability of these recombinant strains to bind to the aphid stylet. Carroll et al. (2016) showed that PVYNTN maintained a higher viral titre in plants co-infected with PVYO, and PVYNTN was more likely to be transmitted from plants with such a co-infection. Also, Myzus persicae carrying combinations of PVYNTN along with PVYO or PVYN:O most often infected plants with both strains or only PVYNTN, rather than single infections of PVYO or PVYN:O, which would be expected to lead to an increased population proportion of PVYNTN over time (Srinivasan et al. 2012). Conversely, in a detailed examination of relative transmission efficiencies of PVY strains across several aphid species, Fox et al. (2017) concluded on the whole that PVYO and PVYNTN were transmitted with similar efficiencies. This, however, was very species-specific; while the model aphid Myzus persicae showed no difference between PVYO and PVYNTN, others such as Acyrthosiphon pisum and Sitobion avenae showed nearly twice the efficiency transmitting PVYO over PVYNTN, and still others including Aphis fabae, Cavariella aegipodii and Macrosiphon euphorbiae showed higher efficiency with PVYNTN over PVYO.

In this study, we observed a higher proportion of infected tubers on average from plants with PVYNTN than other strains, though this difference was small in both varieties. Preferential transduction of PVYN-serotype strains has been reported previously (Beemster 1976, Basky and Almasi 2005), but given the small differences between strains observed in the present study, it is unlikely to contribute as much to the rapid population changes in PVY strains as differences in strain transmission between plants. Typically, closely related viruses or strains are antagonistic when infecting or multiplying within plant host cells (Syller and Grupa 2016), and any such antagonism or synergism between PVY strains has been little studied in the PVY-potato pathosystem.

In the commercial setting, other management factors may affect spread of PVY strains in the field, but these were controlled for in our field trials. For example, greater rogueing success of PVYO plants due to the typically cryptic visual symptoms of the other strains could compound the already lower rate of PVYO spread. It is possible there are strain-specific effects on tuber longevity in storage or survival until planting that may bias the strain proportions in following year plantings. Also, tuber symptoms such as necrotic ringspots caused by PVYNTN in some susceptible potato varieties may make seed unmarketable and thus cause higher rates of destruction of PVYNTN-bearing seed lots in these varieties. Any of these factors could impact the proportions of PVY strains, but they have yet to be studied.