To dispel a laboratory or bioinformatics error or contamination, NGS-based identification of putative new viral sequence or new strains of existing viral species requires further confirmation by complementary methods, usually by (RT)-PCR using specific primers. If the concentration is very low, the sample can be enriched in viral particles through homogenization, filtration, and/or ultracentrifugation prior to nucleic acid extraction. Enriched viral particles can also be observed by electron microscopy, even if electron microscopy lacks sensitivity and viral particles can be missed. For DNA viruses, especially Caulimoviridae, additional research is also needed to evaluate whether the detected viral sequences exist in an episomal form and not only as the trace of a “fossilized” virus integrated in the plant genome (Liu et al., 2010; Aiewsakun and Katzourakis, 2015).

The assembled contig will generally allow for provisional taxonomic assignment of the viral sequence. Nevertheless, this can turn out to be a very difficult task, made even more complex by (i) the frequent incompleteness of the genome sequence data obtained, (ii) the variable and irregular coverage of viral genomes, (iii) the incompleteness of viral taxonomy, and (iv) the variability of species, genus, or family molecular discrimination criteria. Specific guidelines can be found in the latest ICTV report (http://www.ictvonline.org). The detected viral sequences may also correspond to a bona fide virus infecting other organisms associated with the sample, including bacteria, fungi or arthropods (Al Rwahnih et al., 2011; Marzano and Domier, 2015) or to viral sequences integrated into the plant genome. The future solution to this problem is to establish the baseline of virus presence through an improved knowledge of the viromes of agronomically important plants as well as viromes of natural ecosystems, including the integrated viral sequences.

Overall, provisional taxonomic assignment should always be considered with caution but can give first clues on the physical and/or chemical properties of a virus as well as on its biology, including (i) the putative host range for the tentative identification of alternative hosts, (ii) the prediction of the pathogenic potential in those hosts, (iii) the modes of horizontal and/or vertical transmission, including the identification of candidate vectors. These hypotheses have practical consequences such as the selection of appropriate hosts and experimental modes of inoculation, a first assessment of the possible epidemic potential of a new viral agent, or a tentative identification of possible alternative hosts in ecosystems. Altogether, this early step, which requires expertise in molecular biology, in plant virology, and basic bioinformatics skills, might lead to the elaboration of a draft epidemiological scenario (Loconsole et al., 2012; Adams et al., 2014). Nevertheless, this approach can be biased as for example in the genus Closteroviridae, which includes economically important viruses with similar virion morphology, genome size, and organization but with contrasted biological properties (Martelli et al., 2012).

In addition, all the metadata about the sample should be collected precisely to feed the risk assessment process. It is important to document symptoms (if any) and time of appearance, to identify the plant species and cultivar, to describe the sampled plant part (leaves, fruit, seeds, etc.), the geographical origin of the sample, its intended use etc…Any missing information should be completed at this stage so as to be included in the first notification.

The assembly of viral sequence(s) based on NGS reads can result in three scenarios of genome coverage: (i) complete, (ii) incomplete but continuous, or (iii) a set of scaffolds and contigs. The reconstruction of the full genome sequence of a candidate new virus is mainly based on targeted (RT)-PCR amplification and sequencing to fill the sequence gaps between de novo generated contigs, verify their exclusive viral origin and the absence of host-derived sequences. Taxonomic assignment can also be fine-tuned thanks to the whole genome sequence; but in a worst case situation, even the availability of a complete genomic sequence may be insufficient to settle such a question (Marais et al., 2016).

Identifying a new virus species, even from a partial genome sequence, triggers the immediate need to develop a diagnostic protocol. This is a fundamental step toward the management of viral diseases in cultivated crops or the unveiling of a new virus eco-epidemiological role in wild plants. The diagnostic of newly identified and confirmed viruses is usually done using PCR-based methods (the most popular ones for new viruses) or with the LAMP technology, as a recent alternative (Boonham et al., 2014). Nucleic acid-based methods are currently preferred to diagnose new viruses because they are quicker, easier and cheaper to develop than antibody production for ELISA.

Small-scale epidemiological surveys at the discovery location can be undertaken. Such surveys will also take into account the hypotheses that may have arisen after bibliographical research and should be focused on the plant species where the virus was detected; but they could also take into account, whenever possible, the potential host(s) and vector(s), their geographical spread and seasonality. Likewise, sampling methods enabling for the statistical linking of findings and field observations should be selected. Several scenarios are possible depending on the origin of the sample: field or trade/quarantine/certification.

The presence of symptoms in the initial field host has often been perceived as a starting point for further investigations about the presence of a virus, but asymptomatic plants must also be included in the sampling because symptomatology reflects the complex interplay of infectious agent(s), the host plant metabolism, its defense systems and varietal specificities, as well as abiotic factors. The advent of wide metagenomic studies of environmental viromes (Kristensen et al., 2010; Mohiuddin and Schellhorn, 2015) allows for the identification of new viruses from any sample, including asymptomatic field samples. Thus, even if focused symptom observations in the field cannot be neglected, a survey cannot be led by symptomatology alone. Hence, virus spread to other agro-ecological niches where they may encounter new hosts, co-infecting viruses or vectors, may cause a shift in their pathogenic potential and result in a serious disease.

The inoculation of a new virus to test plants in laboratory/greenhouse conditions is the basis for any biological characterization and may be particularly complex. Depending on the objective, the virus can be transmitted to indicator plants, host plant candidates or other cultivars of a same species. The modes of vertical and horizontal transmission of viruses are diverse, so that the integration of taxonomic (Figure 2 and Section Provisional Taxonomy Assignment), bibliographical (Figure 2 and Section Bibliography), and field survey data (Figure 2 and Section Field Surveys) is of utmost importance for the proper design of experimental transmission procedures in the laboratory or the greenhouse.

Plants can be inoculated by several techniques, the most common ones being mechanical inoculation, grafting, vector inoculation or the use of dodder, but all these techniques have potential limitations as they may not separate viruses in case of a mixture of viruses. A very interesting and universal technique for an in-depth characterization of a virus is the preparation of infectious clones. The infectious clones, requiring the complete viral genome sequence, offer valuable information about (i) symptomatology for individual viruses and for mixture of viruses, (ii) host range studies and transmissibility, (iii) the assessment of natural or induced mutation/recombination rates in host plants, and (iv) the effects of targeted modifications on virus-host interactions and symptomatology. Mastering virus transmission will also be key to the development of research efforts to understand the potentially positive effects of viruses on plants. Inoculated plants from various species or cultivars can be submitted to biotic or abiotic stresses, and their reaction can be studied in laboratory and/or in greenhouse experiments to provide evidence of beneficial interactions.

Large-scale surveys, organized on a national or international scale and based on the diagnostic technique(s) developed for the new agent, should be undertaken to evaluate its prevalence and distribution within a given ecosystem (Krenz et al., 2014). These studies will aim to monitor the candidate/demonstrated host plants/vectors. They should be carried out in agricultural ecosystems (fields, commercial orchards or vineyards, nurseries, and germplasm collections), and in natural ecosystems, to enable a holistic characterization of the pathosystems with particular attention to symptomatology. They will complement the preliminary information gained at a smaller scale (See Section Field/Batch Observation) or in the laboratories and greenhouses (See Section Transmission Experiments). The interplay between field surveys and laboratory experiments is particularly important at this stage because it will allow targeting the most appropriate host plant(s) and vector(s), and thus increases knowledge on virus biology.

Large-scale surveys will also provide insights into the genetic variability of a new virus within the surveyed territories. This knowledge will be key for efforts to optimize diagnostic techniques by selecting primers covering the genetic diversity of the species (Massart et al., 2014). In addition, the identification of new variants may bring new fundamental hypotheses on the pathogenicity of the viruses, and these new variants could be characterized under laboratory and greenhouse conditions (See Section Transmission Experiments).

The biological characterization of new viruses may be particularly challenging with complex of different viruses, where the combination of two or more viruses can significantly modify their pathogenic potential through synergistic or antagonistic interactions (Martin and Elena, 2009; Syller and Grupa, 2016). Mixture of virus species is indeed very frequent, especially in woody plants like grapevine (Jooste et al., 2015), often leading to unpredictable variations in symptoms, infectivity, accumulation, and/or vector transmissibility. Biological and technological solutions for isolating and determining the role of individual viruses are sometimes possible using selective experimental hosts followed by back-inoculation of “purified” isolates to the original host (Ali and Roossinck, 2008), partial sanitation (Maliogka et al., 2015), infectious clones (Nagyová and Šubr, 2007), or the use of ion-exchange chromatography on monolithic supports to separate viral particles (Ruščić et al., 2015). The biological properties of viruses alone or in mixture can therefore be compared in depth (see Section Field Surveys). The complexity of disease etiology in field conditions might pose additional challenges, as exemplified in the Case Study N°4 on carrot (See Supplementary Material).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.