Field assessment of weeds and associated aphids were conducted during the crop growing season from spring to autumn over a three-year period (2015–17) in Transylvania, Romania. Surveys were performed under the two aforementioned management regimes (low vs. high input fields), purposely selected to reflect similar geographic and climatic regions. The low-input (LIF) fields were characterized by traditionally managed fields of the Old Saxon region, some 7,440 km² in area at an altitude of 230–800 m above sea level (a.s.l.), the HIF by large monocultures with a low density of natural vegetation, situated about 5,500 km² in area at an altitude of 220–750 m a.s.l. (Szabó et al., 2019, 2020). Around 200 km separated the areas with no direct connections between them via main roads or railways.

Canadian goldenrod, S. canadensis (L.) is native to north-eastern and north-central America, grows in a diverse range of habitats, and is one of the most aggressive invasive weeds in Europe, resulting in significant crop losses in newly invaded areas (Anastasiu and Negrean, 2005; Abhilasha et al., 2008; Fenesi et al., 2015).

The weed assessment method here used has earlier been detailed by Szabó et al. (2019, 2020). In brief, two sample areas each of around 10m² were selected in both LIF and HIF regions. Inside each, 10 smaller 1×1 m subplots were assessed and the exact coverage of all plants (invasive and other weeds) identified. Bi-weekly assessments were made in the same subplots from the end of May until June during the 2-year vegetation growing period.

Canadian goldenrod together with aphid colonies were collected and stored at −20°C. Aphids were counted in the laboratory and species identified according to Blackman and Eastop (2000) and Blackman (2010). For each sample collection period, 100 samples were collected per transect and management system, i.e., 400 samples/management system/collection data = total 1,600 samples/year. A similar sample collection system was followed for crop plants around each weed transect. Because only potato, alfalfa and maize were cultivated, samples from these crops were collected and tested for viruses. All plant material was stored at 70°C prior to virus identification.

The two native and locally-distributed aphid species investigated, the leaf-curling plum aphid, B. helichrysi Kaltenbach, and the foxglove aphid, A. solani (Kaltenbach), are both considered to be highly polyphagous (Blackman and Eastop, 2000; Blackman, 2010), and as such, important virus vectors. In B. helichrysi, hosts include members of the Asteraceae, e.g., Chrysanthemum, as well as Prunus spp., Citrus, Fragaria, Medicago, Solanum, Trifolium and also maize (Popkin et al., 2017). The most important viruses transmitted are Beet mild yellowing virus, Potato virus Y and Sharka virus (= Plum Pox virus; Isac et al., 1998). Hosts of A. solani include several crop plants, e.g., celery, carrots, cucurbits, peppers, tomato, tobacco, tulip bulbs, and legumes (Jandricic et al., 2014), the aphid transmitting Potato viruses A, Y, and X, Potato leaf roll virus, Bean yellow mosaic virus, Cucumber mosaic virus, Soybean dwarf virus and Turnip yellows virus (Jandricic et al., 2010, 2014).

Analyses were made according to White and Kaper (1989) and Szabó et al. (2020). Material from both weed and crops was separately homogenized in an ice-cold mortar in 650 μl extraction buffer (100 mM glycine, pH 9.0, 100 mM NaCl, 10 mM EDTA, 2% SDS and 1% sodium lauroylsarcosine). Samples were mixed with phenol, centrifuged for 5 min., whereafter the aqueous phase was treated with a mixture of phenol, chloroform and isoamyl-alcohol (25:24:1). Following further treatment using chloroform: isoamyl-alcohol (24:1), the solution was precipitated with ethanol and re-suspended in sterilized water (White and Kaper, 1989; Szabó et al., 2020). Polyacrylamide gels from RNA pools were used for small RNA High Throughput Screening (HTS), separately for samples related to weed, crops, area and collection times. This approach allowed detection of any plant viruses present in samples at any given time during the survey. When investigating field crops belonging to different plant families (potato, alfalfa and maize) and also when the origin of cultivars was uncertain, their virus patterns were pooled. These pools were used for small RNA library preparation (2 libraries, one for weed and one for crop plants) using Truseq Small RNA Library Preparation Kit (Illumina, USA; Czotter et al., 2018). Thereafter, all samples were sequenced using HiScan2000 by UD Genomed (Debrecen, Hungary) 50 bp, single end procedures (Supplementary Figure S1). Fastq files of the sequenced libraries were deposited in the Gene Expression Omnibus (GEO) repository¹, accessible via series accession number GSE132755 (Szabó et al., 2019, 2020). Aphids from each plants (weed and crop) were also tested for viruses; the insects were carefully collected from plants during feeding, and stored in absolute ethanol together with the plant portion containing the aphids’ stylets, prior to virus testing.

The analyses were made in 2022. For POD enzyme analysis, weed and crop plant leaf samples were collected each year and date simultaneously with the aphid samples collected from the various fields and transects. In this way, aphid-plant interactions through high POD activity could be directly tested. Leaves and stems from the top of the plants containing aphids were sampled from Canadian goldenrod (n = 10 / 1 m² sub-transect) and crop plants (n = 7 / 1 m² sub-transect), using 20 samples from each transect (Supplementary Table S1), and stored at −20°C prior to enzyme extraction and activity assays. For this, ~ 400 mg of leaves were homogenized in 1 ml 50 mM phosphate buffer, pH 7.0, in a high-speed benchtop homogenizer (MP Biomedicals). The concentration of enzyme extract was determined using the Bradford method (1976; Szabó et al., 2019), whereafter POD activity was analyzed according to Németh et al. (2009). The increase in absorbance at λ 480 nm was followed in a spectrophotometer for 5 min., with POD activity separately defined as an absorbance change of 0.01 units.min ⁻¹ at both 5 and 10 μmol min⁻¹. mg protein.⁻¹

Besides the fact that only potato, alfalfa and maize crops were present in both years of the assessment of crops and aphids detected around stands of Canadian goldenrod, some viruses other than those infecting these crops were also identified. In light of this, a detailed survey of the scientific literature was performed to assess the virus infection potential of crops studied and also other cash crops. All viruses detected in Canadian goldenrod and the aforementioned crops were analysed and scientific references searched using WoS (Web of Science) for the years 2000–22 inclusive. From this, all possible crop plants with potential as aphid hosts in Europe (i.e., in which viral symptoms had at one time been found) were noted.

Firstly, for the Canadian goldenrod data, a mean coverage value in each 1 m² sub-quadrat was computed, involving averaging all the species relative abundance data from each 1 × 1 m plot per sampling date. As goldenrod was only present in low abundance under HIF conditions, its average coverage of 0.8% did not allow further analyses, so only data from LIF was considered. Next, analyses of inter-annual differences in coverage were compared using MANOVA, whereupon the mean values of weed coverage for one 1 m² quadrat (40 data / field type / collection dates) were used. No significant difference in coverage were detected between the 2 years of the survey (p = 0.34); therefore cumulated and averaged data were used for analyses.

Thereafter, aphid abundances were assessed on both weeds and crop plants. Aphid abundance on Canadian goldenrod did not meet the assumption of normality so that the Kruskal-Wallis test was used, followed by the Mann–Whitney U test to compare the abundance of the two aphid species’ on this host plant. Aphid density analyses of crop plants were performed by considering only the abundance of B. helichrysi. This was because the abundance of A. solani was too low in crop plants (except potato), so that no statistics were possible using this data set. As the B. helichrysi aphid data was normally distributed, ANOVA followed by Tukey testing were performed to compare abundances between treatments and crop plants. All analyses were done using R version 3.0.1 (R Core Team, 2013). Other aphid species were also detected, but in small numbers (e.g., Macrosiphum spp.), hence were not included in the analyses.

Virus diagnosis was determined by small RNA HTS for weed and crops. For bioinformatics analysis, CLC Genomic Workbench was used. For plant virus diagnostics two strategies were followed. Firstly, longer contigs from the non-redundant reads were built and the CLC assembler (de novo assembly) used. Next, directly mapped contigs were made using Reference Genomes of those viruses, which were characterized at least with one contig in any of the libraries (CLC Genomic workbench).Virus incidence in samples was claimed if at least two parameters (virus specific contig and/or normalized redundant virus specific read count) was >> 200, and/or coverage of the virus genome was > > 60%; (Szabó et al., 2019, 2020). Because some viruses detected from weeds could not be directly associated with those detected in crops in close proximity to Canadian goldenrod, a detailed analyses of scientific references was made to associate these viruses with other potential crops present in the area.

POD activity values for weed and crop plants at 5 and 10 μmol min⁻¹ were compared between years (n = 20 / year) using MANOVA, with values from 1 m² sub-transects per sampling period considered. Because no significant year effect was observed (p = 0.83), data between years were averaged and used for further analyses. Similarly with weed data, where only samples from LIF were considered. POD enzyme data at 5 and 10 μmol min⁻¹. mg protein⁻¹ unit were analyzed separately and compared between assessment data, when aphids abundances were the highest in weed and crop plants (June 05–15), and also when aphid abundance decreased (between June 15 and 24) each year (Supplementary Table S1). Because data were normally distributed, a complete randomized factorial ANOVA of POD specific activity values (5 and 10 μmol min⁻¹. mg protein⁻¹ unit) was used for comparison. The Tukey test with p < 0.01 and LSMEAN (Minimal quadratic means) test were performed using the statistic package SAS. For analyses, the average POD quantity / 20 plants / 1 m² sub-transect were used. Linear correlation analyses between POD activity level at both 5 and 10 μmol min⁻¹. mg protein⁻¹ unit and aphid species (B. helichrysi and A solani) abundances on Canadian goldenrod were computed for the same assessment period using the SPSS package, version 3.14.

The whole analyses was performed in 2022. To link virus infections with crop plants, four major factors influencing infections were considered: (1) the weed-aphid species interactions identified as in our assessments; (2), plant viruses identified in both weed and crop plants using new generation small RNA assessment; (3), virus incidences observed and identified suggesting a direct link between weeds and crop plants via aphid-viral interactions; and (4), management strategies formulated to control virus infections. From these, only papers dealing with at least three factors were considered. To link factors, the number of searches that connected these factors, i.e., weed-virus-crops (other than those assessed by us) were considered. Of 44 papers searched between 2000 and 2022, a strong link was found between factors connecting Canadian goldenrod and crop plants, such information being indicative of viral transmission in at least one direction (i.e., virus transfer from weed to crop via aphids; Reference list, Supplementary online materials). Strong and influential links were considered between two factors when at least 10 searches were found that connected these in one direction (and marked in Figure 1 as conceptual representations). This was done by placing the factors (weed, aphids and crop plants) in a Sankey diagram representing direct contact. The method used is rather similar to those used for factors modeling within the DPI (Driving force–Pressure–Impact) framework, reporting different environmental issues facilitating interrelations between these and human and/or social factors (Gabrielsen and Bosch, 2003).

Canadian goldenrod was present in LIF fields at an average coverage of 2.5%. In contrast, at the other site with high input management, its average coverage in transects was 0.8%; thus samples from this site were not used to test for the presence of viruses. A higher density of B. helichrysi was detected on goldenrod (U_1–40 = 3.4, p < 0.01) compared with A. solani. Other species (Macrosiphum spp) were observed at low density (12 individuals) on S. canadensis. Considering aphids on crop plants, the abundance of B. helichrysi was higher on alfalfa and potato and significantly lower on maize (F_1–40 = 4.5 p < 0.01; Figure 2A), while the abundance of A. solani was always low so that no statistics were possible, but a higher abundance on potato was clearly observed (Figure 2B).

The small RNA libraries sequencing resulted in more than 21 million raw reads for crop plants and more than 17 million for Canadian goldenrod. After trimming and quality control, > > 20 million raw reads were obtained for crops and > > 16 million for weeds (Table 1). Overall, the number of plant viruses detected in Canadian goldenrod from LIF was 18, of which 17 were plant viruses, and one an insect virus. Of the plant viruses detected, 15 were RNA and two DNA viruses (Table 2). Altogether 11 viruses were found to be directly associated with crop plants assessed during the study, mostly alfalfa and potato, whereas no particular virus associations were discovered with maize. The vast majority of the viruses detected were associated with plants from the family Solanaceae; all viruses that infect these plants (Potato virus Y, V, S, M, X, Wild potato mosaic virus and Bidens mosaic virus isolate SP01) were found in goldenrod and in potato crops all over the LIF area sampled. Furthermore, two viruses infecting sweet pepper and potato were detected (Pepper severe mosaic virus and Pepper yellow mosaic virus). Bean leaf roll virus and the Lucerne transient streak virus, both infecting plants of the family Fabaceae, were also detected in goldenrod and alfalfa at all sampling sites (Table 2). Amongst other viruses associated with Canadian goldenrod, some were highly plant pathogenic, including PPV (Plum Pox Virus), Sunflower ring blotch virus, Sunflower chlorotic mottle virus and Tobacco vein clearing virus. Two viruses associated with sweet-potato may be considered important in terms of potential virus vectoring (Sweet potato virus G isolate and Sweet potato mastrevirus), both present in goldenrod, although its presence in host plants other than sweet potato has yet to be assessed. The presence of the insect virus (Helicoverpa zea nudivirus-2) in Canadian goldenrod is a new finding, with no previous confirmed record of its occurrence in Europe (Table 2).

While considerable efforts were made to extract and detect these plant viruses from aphid stylets collected from weeds and crop plants, this proved technically impossible in our particular study due to the low titer of virus present.

Upon assessing POD enzyme activity, this was found to be detectable at both 5 and 10 μmol min⁻¹. mg protein⁻¹ unit, its activity increasing when aphids feeding was at its highest on the sampled plant (June 05–15 each year; Figure 3). A very similar trend was observed in crop plants. Furthermore, a high positive correlation between aphids abundance and POD activity level was detected on weeds (Figure 3), along with a very similar trend in crop plants (a Rho value of <0.79 at 5, and < 0.8 at 10 μmol min⁻¹. mg protein⁻¹ for B. helichrysi on potato, and < 0.77 at 5, and < 0.82 at 10 μmol min⁻¹. mg protein⁻¹ on alfalfa). Because of the low density of A. solani on crops, no comparison were made.

Conceptual analyses of viruses identified in the present assessment revealed that the most vulnerable crops appeared to be members of the Solanaceae. In the case of Canadian goldenrod, which belongs to the family Asteracae, this is currently expanding its range in mainland Europe, followed next by species from the family Fabaceae, e.g., alfalfa. In addition, the possibility of virus infection via aphid transmission from goldenrod to Helianthus, Chenopodiaceae and also Stone fruits (peaches, plums, cherries, etc.) are high. As Sweet potato cultivation is also expanding in Europe, viral infection of this crop is likely when Canadian goldenrod is present (Figure 1).