Introduction

Sweet potato virus disease (SPVD) is probably the most devastating disease constraint of sweet potato [Ipomoea batatas (L.) Lam] worldwide [9]. It is caused by co-infection of the aphid-transmitted sweet potato feathery mottle virus (SPFMV, family Potyviridae, genus Potyvirus) and the whitefly transmitted sweet potato chlorotic stunt virus (SPCSV; genus Crinivirus; family Closteroviridae) [17, 20, 45]. Single infections of SPFMV usually show mild or no symptoms, and no appreciable yield reduction can be observed [12, 20, 33]. However, co-infection with SPCSV causes SPVD, which is characterized by very severe symptoms such as general chlorosis, stunting, leaf strapping, leaf crinkling and even plant death [17, 23, 45], and yield losses ranging from 70 to100% [20, 33, 38, 40]. Molecular studies have shown that co-infection of SPCSV enhances SPFMV RNA viral titers by at least 600-fold, whereas SPCSV titers remain equal or are reduced as compared to single infection [23, 24, 38]. The severity of SPVD, and the degree of SPFMV titer increase, depends on the strain of SPFMV involved in the double infection [20, 24]. Besides SPFMV, several other potyviruses, as well as other, unrelated viruses, can cause synergistic diseases when co-infecting with SPCSV [24, 38, 51], although the importance of these interactions for yield losses in the field are not well known.

SPFMV is one of the most widespread viruses infecting sweet potato [36]. The CP genomic region of SPFMV has been used in previous studies to establish phylogenetic relationships among SPFMV samples. It can be divided into four phylogenetic lineages [5, 21, 25, 38, 49, 50]: East Africa (EA), constituted by East African samples; Russet Crack (RC), comprising samples from Australia, Africa, Asia and North America; Ordinary (O) containing samples from Japan, China, Korea, Niger, Nigeria and Argentina; and Common (C) including samples from USA, China, Australia, East Africa and Argentina. Unlike the geographically unrestricted C, RC and O strains, the EA strain has almost exclusively been detected in the East African countries, the possible exceptions being two sequences available from the GenBank from Spain and Portugal [53] (Table 1). SPFMV has been reported from sweet potato in Peruvian fields since 1987 [30], but it was only after 1998 that the prevalence of SPVD emerged, possibly due to the increase of whitefly populations during the exceptionally strong El Niño phenomenon of that year [20]. Although the impact of SPVD on the yield of Peruvian sweet potato cultivars has been assessed [20], no molecular studies have been carried out concerning variability of its two causal agents. Cedano et al. [10] reported that some SPFMV isolates differed in the severity of symptoms induced in Ipomoea nil. SPFMV-C1, was collected during the 1980s [39] and shown to be closely related to the C strain [22, 42], whereas three other isolates, M2-41, M2-44 and C-18, were obtained from SPVD-infected sweet potato plants and differed in their symptomatology and serological reaction with monoclonal and polyclonal antibodies [20].

Table 1 SPFMV isolates and samples used in this study
Full size table

Studies on other potyviruses infecting sweet potato are less abundant. These include sweet potato latent virus (SPLV), found in Asia, Africa and Peru [20, 29], sweet potato mild speckling virus (SPMSV) from Argentina and Peru [15, 20], sweet potato virus G (SPVG), identified in China, Egypt, Ethiopia, Europe and the United States [3, 6, 12, 13, 21, 47] and a potyvirus first reported as sweet potato virus II [43], and later named ipomoea vein mosaic virus [47], sweet potato virus Y [4] or sweet potato virus 2 (SPV2) [49], has been identified from the United States [47], Africa, Taiwan, China, Portugal [49] and Australia [6]. Since the proposal to refer to this new potyvirus species as SPV2 has been favorably considered by the International Committee on the Taxonomy of Viruses, this name is used here. SPLV and SPMSV have been reported in Peru at low frequency [20].

Understanding the molecular variation of viruses is essential to design knowledge-based strategies to control them. In the present study, we determine the nucleotide sequence of the region encompassing the 3′-terminal ∼1,800 nts of 17 potyvirus samples mostly collected from SPVD-affected plants from the major sweet potato-producing areas in Peru. Most of the viruses were identified as SPFMV , but we also report for the first time the occurrence of SPV2 and SPVG in Peru and South America. Phylogenetic analysis of SPFMV sequences indicates a variable population of SPFMV in Peru, including EA, C and RC strain groups, and provides evidence for the existence of recombinants between strains.

Materials and methods

Virus-infected plant samples and virus isolates

One symptomless sweet potato plant and 14 with SPVD-like symptoms were collected at random from six main sweet potato-producing areas in Peru (Fig. 1). Plants with SPVD-like symptoms were found in all locations except for the Chira Valley in Piura, where only symptomless plants were collected. Details of the samples and isolates, their names, province of origin and year of collection are shown in Table 1 (Fig. 4 for SPV2 and SPVG). Stem cuttings of collected plants were maintained in an insect-proof greenhouse at CIP headquarters, Lima, Peru, for at least 3 weeks before analysis. Asymptomatic plants were grafted onto the indicator plant I. setosa, which was observed for symptom development and analyzed by serological means. The presence and identity of sweet potato viruses were confirmed using antisera included in the NCM-ELISA sweet potato virus detection kit from the International Potato Center (CIP, Lima, Peru) [51], according to the manufacturer’s protocol. A number of SPFMV isolates kept in desiccated I. nil leaves for as long as 20 years, as well as the Peruvian isolate of SPFMV, C1 [39, 42], and the North American isolate, YV [35], maintained in Nicotiana benthamiana and I. nil, respectively, as part of the CIP virus collection, were also included in the study. Isolation of these viruses was done by three consecutive single-lesion transfers on Chenopodium amaranticolor [10, 30, 35, 39].

Fig. 1
figure 1

Locations of the sweet potato fields surveyed in Peru

Full size image

RNA extraction, RT-PCR and cloning

Total RNAs were extracted from approximately 0.2 g leaves of SPVD-infected sweet potato or SPFMV-infected indicator plants using TRIZOL reagent (Invitrogen, CA, USA) according to the manufacturer’s procedure. For lyophilized samples, the initial amount of tissue was 0.02 g. Although this provided sufficient amounts of total RNA, a further purification of high-molecular-weight RNA with 4M LiCl4 significantly improved the subsequent RT-PCR reaction. The integrity of the isolated RNA was visually verified after electrophoresis in a standard formaldehyde agarose gel and staining with ethidium bromide. Reverse transcription was performed on extracted RNA using AMV reverse transcriptase (Promega, WI, USA) according to the manufacturer’s recommendations, with the primer FMV10820 (Table 2), corresponding to the last 20 nucleotides of the virus genome excluding the poly A tail of the ‘SPFMV’ subgroup (see Discussion) of potyviruses [50]. A fragment comprising the 3′-terminal ∼1,800 nts of the potyvirus genome including the 3′-terminal part of the NIb gene, the complete CP gene and the 3′ UTR was amplified by PCR using the potyvirus-specific forward primer PVD-2 (Table 2) [16] and the reverse primer FMV 10820. PCR amplifications were carried out in 25-μl volumes containing a 1× PCR buffer (Promega), 0.4 mM each of dGTP, dATP, dTTP and dCTP, 0.3 μM of each primer, 1 U of Taq DNA polymerase (Promega), and 2.5 μl of the reverse transcription reaction. The cycling conditions were set as follows: 95°C for 5 min, followed by 35 cycles at 95°C for 15 s, 52°C for 20 s and 72°C for 90 s , and then one final elongation at 72°C for 10 min. PCR products were separated on 1% agarose gels and fragments of interest recovered by using the Wizard SV gel extraction kit (Promega) according to the manufacturer’s recommendation. The eluted DNA was ligated into plasmid vector pCR 2.1 (Invitrogen) according to the manufacturers instructions and cloned in Escherichia coli strain DH5α.

Table 2 List of synthetic oligonucleotide primers used in this work
Full size table

DNA sequencing, sequence analysis and phylogenetic relationships

Plasmids containing the amplified viral sequences were sequenced in both directions (Macrogen, Seoul, Korea). Internal primers were designed for sequencing as shown in Table 2. At least two individual clones per sample were sequenced, and if inconsistencies were detected, then further clones were sequenced. In one case in which an unexpected gap was identified, the fragment was re-amplified from a new RNA extraction and re-sequenced for confirmation. Sequences were assembled using ContigExpress, included in the Vector NTI software package (Invitrogen). The alignments and phylogenetic analyses were performed with the MEGA 4 software package [27] and included a number of sequences downloaded from the GenBank database (Table 1, Fig. 4). Distances were calculated using the Kimura 2-parameter model, and trees were assembled using neighbour joining with 1,000 bootstrap replicates.

Recombinant analysis

Recombination events were suspected after visual examination of the nucleotide sequence alignments. To confirm the putative recombination events we used the Recombination Detection Program (RDP) 2.0 [31]. To further confirm the results of the predictions by RDP, sequence alignments were cut at the predicted recombination junctions, and phylogenetic analysis of the aligned sequences corresponding to each side of the junction was performed. Viruses grouping in different strain groups with strong (>90%) bootstrap support in the different phylogenic trees were considered true recombinants.

Results

Nucleotide (nt) sequences encompassing the 3′-terminal part of the NIb gene, the complete CP gene and the 3′UTR of 14 Peruvian samples infected with SPFMV, the SPFMV isolate C1 from Peru, and the isolate YV from USA were determined in this study, as well as those of one sample of SPV2 and SPVG. RNA was also obtained from 12 desiccated leaf samples of SPFMV-infected plants collected in Peru as far back as 20 years ago. Although this yielded apparently intact RNA as assessed by gel electrophoresis, RT-PCR amplicons of the correct size were obtained from only three of these samples, and only one of these amplified products could be cloned (SP-33) and proved to be an SPFMV sequence.

Analyses of SPFMV sequences

Analysis of SPFMV included more than 40 sequences taken from the GenBank database in addition to the ones determined in this study (Table 1). The deduced CP sequences of most SPFMV-infected samples were 315 aa in length with 155 aa showing variability (36.5%), 60 of which were located in the N-terminal region. With the exception of the type isolate C, all isolates belonging genetically to strain group C lacked two amino acid residues at positions 62 and 63, as reported previously by Tairo et al. [50]. In addition, however, a deletion of 14 aa was found in the CP N-terminal region (at aa position 42–55) of the Peruvian EA isolate M2-44. Nucleotide and amino acid sequence identities for different regions are presented in Fig. 2.

Fig. 2
figure 2

Mean nucleotide and amino acid (bold) inter- and intra-group identities (%) calculated for the C-terminal 208 aa of the NIb (a); the entire coat protein, CP (b); the N-terminal aa of the CP (c) and 3′UTR (d). Recombinant sequences were excluded

Full size image

Phylogenetic analysis of the aligned sequences split the SPFMV samples into four strain groups: C, RC, O and EA. Peruvian SPFMVs were found to correspond to groups EA, RC and C (Figs. 3, 4). Visual inspection of the alignments of the complete ∼1,800 nt fragments suggested that some SPFMV samples might be the result of recombination between strain groups; i.e. in isolates C and YV the CP-encoding region and 3′UTR seemed to be derived both from a C strain and a non-C strain virus, respectively, and in both of the Egyptian samples, Eg1 and Eg9, the central part of the NIb gene and the remaining 3′ part of the genome appeared to originate from viruses of strain groups EA and RC, respectively. To confirm this, recombination analysis was performed using the various algorithms of the RDP2.0 program, which all predicted, with high, but varying probability, the suspected recombination events in the strains in question. The algorithm Max-Chi Squared [32] provided by the RDP program predicted the following breakpoints, in accordance with the suspected points identified in the sequence alignments: the first one, located at nt position 9,597 (strain S [44]; nt 570 in our fragment) within the 3′-terminal part of the NIb gene was shared by SPFMV Eg-9 and SPFMV Eg-1, whereas the second one, located at nt position 10,500 (nt 1473 in our fragment) within the 3′ terminal region of the CP was present in SPFMV-YV and -C (Fig. 3a). Further confirmation of recombination was obtained by construction of phylogenetic trees using the segments corresponding to each side of the predicted recombination breakpoints (Fig. 3b–e). The topologies of the four trees produced were similar, distinguishing the four strain groups, except for the tree created using the sequences encompassing nts 1473-3′end, in which only two well-supported clades could be discriminated; one comprising viruses of the O, RC and EA strains, and one comprising C strain viruses (Fig. 3c). The conflicting affinity of the Egyptian viruses between the trees produced using the different genome regions clearly confirms the predicted recombination events; i.e. Eg1 and Eg9 are members of strain group RC when alignments comprising the nt positions 570-3′-end are analyzed (Fig. 3e), but belong to strain group EA on the basis of the nt positions 0–570 (Fig. 3d). Similarly, isolates C and YV belong to the strain group C when the region encompassing nts 0–1452 is analyzed (Fig. 3b) but group together with RC, EA and O when 3′ nts are analyzed (Fig. 3c).

Fig. 3
figure 3

Phylogenetic trees calculated from alignments of nucleotide sequences on either side of recombination breakpoints identified with the RDP program (see “Materials and methods”). a Schematic representation of the 3′ region of SPFMV used for analysis indicating the regions used and the corresponding trees matching to predicted recombination breakpoints in isolates C and YV, and Eg1 and Eg9. b–e Phylogenetic trees of SPFMV sequences encompassing the regions indicated in a, arrowheads indicate the conflicting groupings of sequences in which recombination events were predicted. The scale bar in each figure indicates Kimura nucleotide distances. In b the sequences corresponding to the NIb genes were not included because they are not available for isolate C. Similarly, the sequence of the 3′UTR is not included in the analysis in (e) as this region is not available for the isolates Eg1 and Eg9. The percentage of bootstrap support out of 1000 replicates is given at each of the major nodes in the trees

Full size image
Fig. 4
figure 4

Phylogentic tree of CP nucleotide sequences of potyviruses. The Ipomoea-specific ‘SPFMV’ subgroup, as well as previously identified subgroups are shaded in grey. Sweet potato-infecting viruses are in bold, proposed virus strain groups are shaded in light grey with roman numerals, except for SPFMV, which is according to (25). The scale bar indicates Kimura nucleotide distance. The percentage of bootstrap support out of 1,000 replicates is given at the major nodes in the trees where they exceeded 80%. GenBank accession numbers: SPFMV sweet potato feathery mottle virus (see Table 1); SPV-Zim: (AF016366); SPLV sweet potato latent virus (Ch: X84011, Jap: E15420, Tai: X84012); SPMSV sweet potato mild speckling (U61228); SPVG sweet potato virus G (Ark15: Ref 3, CH2: X76944, CH: Z83314, Eg: AJ515380, Henan: DQ399861, Hua2: EU218528); SPV2 sweet potato virus 2 (Aus54: AM050884, Hua3: EU218529, MD2: AY459606, PD12: AY459607, Thomas16A: AY459608, VTSBT-Tschilombo: AY459609, XN3: AY459611, Zambia: AY459610); agropyron mosaic virus (NC_005903); bean common mosaic virus (NC_003397); bean common mosaic necrosis virus (NC_004047); bean yellow mosaic virus (NC_003492); beet mosaic virus (NC_005304); chilli veinal mottle virus (NC_005778); cocksfoot streak virus (NC_003742); cowpea aphid-borne mosaic virus (NC_004013); daphne virus Y (NC_008028); East Asian passiflora virus (NC_007728); hordeum mosaic virus (NC_005904); Japanese yam mosaic virus (AB027007); johnsongrass mosaic virus (NC_003606); konjac mosaic virus (NC_007913); leek yellow stripe virus (NC_004011); lettuce mosaic virus (NC_003605); lily mottle virus (NC_005288); maize dwarf mosaic virus (NC_003377); narcissus degeneration virus (NC_008824); onion yellow dwarf virus (NC_005029); papaya leaf-distortion mosaic virus (NC_005028); papaya ringspot virus (NC_001785); pea seed-borne mosaic virus (NC_001671); peanut mottle virus (NC_002600); pennisetum mosaic virus (NC_007147); pepper mottle virus (NC_001517); pepper severe mosaic virus (NC_008393); Peru tomato mosaic virus (NC_004573); plum pox virus (NC_001445); potato virus A (NC004039); potato virus V (NC_004010); potato virus Y (NC_001616); scallion mosaic virus (NC_003399); shallot yellow stripe virus (NC_007433); sorghum mosaic virus (NC_004035); soybean mosaic virus (NC_002634); sugarcane mosaic virus (NC_003398); Thunberg fritillary virus (NC_007180); tobacco etch virus (NC_001555); tobacco vein mottling virus (NC_001768); turnip mosaic virus (NC_002509); watermelon mosaic virus (NC_006262); wild potato mosaic virus (NC_004426); wisteria vein mosaic virus (NC_007216); yam mosaic virus (NC_004752); and zucchini yellow mosaic virus (NC_003224)

Full size image

Analyses of SPV2 and SPVG sequences

A Blast search indicated that sequences closely related to SPVG and SPV2 had been amplified from two plants from Huaral showing typical symptoms of SPVD. Infection by these two viruses was confirmed by NCM-ELISA. Phylogenetic analysis of the SPV2 sequence showed it was closely related to sample Thomas 16A from South Africa, and confirmed the four genetically distinct lineages suggested previously by Ateka et al. [6, Fig. 4). The SPVG sequence determined here, however, was quite distinct from other reported SPVG samples, suggesting that SPVG-Hua2 might represent a novel strain. Thus, SPVG also can be divided into at least three genetically distinct lineages represented by the CH2, Hua2 and the remaining samples, respectively (Fig. 4). Inclusion of additional potyvirus CP sequences into the analysis further revealed that SPV2, SPVG, SPFMV and a potyvirus isolated from sweet potato in Zimbabwe are related, forming a well-supported separate phylogenetic lineage within the genus Potyvirus (Fig. 4).

Discussion

Our study, the first attempt to classify Peruvian sweet potato potyviruses at the molecular level, demonstrated the presence of SPFMV strains C, RC and EA, as well as SPV2 and SPVG in the main sweet potato-producing regions. This is the first time that SPFMV of strain group EA has been reported from the Americas and SPV2 or SPVG from South America. Two of the isolates, C1 and SP-33, corresponding to strain group C and EA, respectively, were collected in 1987, indicating that these viruses had been present in Peru before SPVD occurred at a high incidence in 1997.

Comparison of the SPFMV sequences with those available from the database enabled us to identify novel variations amongst SPFMV strains. The CP aa sequence of the EA sample M2-44, which, coincidentally, was reported as one of the most detrimental isolates identified in Peru [20], lacks 14 amino acids. Although deletions seem rare in the CP of SPFMV, deletions of 12 aa in the CP region are common in isolates of yam mosaic virus [2]. Besides the characteristic 42-nt (14-aa) deletion found in the C terminus of the CP of strain M2-44, we also obtained the first evidence for recombinants of SPFMV. Four samples containing recombinant segments were detected visually, and their occurrence was confirmed using specialized software and phylogenetic analysis (Fig. 3). The 3′-terminal sequences of the recombinant isolate YV were amplified and cloned from two individual RNA extractions to be sure that the recombination event detected was not an artifact of the PCR reaction. The fact that the same recombination is found in isolate C from the USA, which was cloned and sequenced using separate primers by a different laboratory, corroborates that this is unlikely to be a PCR artifact. The tentative recombinant viruses Eg1 and Eg9 [21] share the same recombination breakpoint in the NIb-encoding region and originate from the same geographic region in Egypt. It is therefore likely that they share a common evolutionary ancestor. The same can be argued for the two recombinant North American strains of SPFMV identified in this study. Recombination in evolutionary history of potyviruses is not a novelty and has been reported for a number of potyvirus species, such as Yam mosaic virus [8], Yam mild mosaic virus [7], Potato virus Y (PVY [18]), Plum pox virus [19], Turnip mosaic virus [41], Lettuce mosaic virus [26] and Sugarcane mosaic virus (SCMV [54]), but also between isolates of closely related species such as Bean common mosaic virus (BCMV), and Soybean mosaic virus [14], Bean common mosaic necrosis virus and BCMV [28] and those of other related viruses [52]. These reports are evidence for an important role for recombination in the evolution of potyviruses, although their frequency may vary significantly between virus species. Such recombination events may lead to more virulent virus strains [18, 54] and even new species [14, 52] or genera [52].

In a previous study by Gutiérrez et al. [20], the SPFMV samples M2-44, M2-41 and C18, here shown to correspond to strains EA, RC and C, respectively, were compared and found to vary in reaction to different antisera, as well as the severity of symptoms induced in I. nil and sweet potato. In both hosts, the EA isolate produced the most severe symptoms, whereas the RC strain produced the mildest symptoms. However, Moyer et al. [37] reported that isolate C (strain group C) caused milder symptoms than isolate RC (strain group RC), and in Japan an RC isolate (SPFMV-S) was reported to be the most severe [34]. Although an EA strain group isolate was not included in these studies, the contradictory results obtained for severity of C and RC isolates in the study of Gutiérrez et al. [20], and that of Moyer et al. [37], suggest that symptom severity may not necessarily be a characteristic of the strains, but rather that of individual isolates. Similarly, the ability to infect N. benthamiana appears to be an isolate-specific rather than a strain-specific characteristic, as only some isolates from different strain groups (EA and C) are able to infect this host [20, 37]. Nevertheless, the isolates belonging to strain group C are genetically distinct from the other strain groups of SPFMV. Therefore, biological differences are expected. It has been suggested that strain group C may represent a separate virus species [50]. Our extended analysis agrees with the percentage of CP nt and aa sequence identities previously reported for samples/isolates of strain group C (Fig. 2b: 75.6–78.7% nt, 79.5–85.2 aa) [25, 50]. However, these identities do not provide a clue for classifying the C strain as another viral species, because they are very close to the CP nt and aa sequence identities of 76–77 and 82%, respectively, used currently as threshold values for potyvirus species demarcation [1]. Similarly, the partial NIb sequences analyzed in this study are at the edge of the recommended species demarcation criterion of 75% identity (Fig. 2a). In contrast, the variability found in the 3′UTR (80.0–85.4% identity, Fig. 2c) appears well above that recommended to distinguish different potyvirus species (76% identity [1]). The definitive reclassification of SPFMV-C as a new species is something that can only be resolved with the help of the entire genome sequence of a C isolate. Preliminary data from the Peruvian C1 isolate indeed indicate that variability in other parts of the genome significantly exceeds those found for strains of the same species (our unpublished data). This and the fact that polyclonal or monoclonal antibodies to SPFMV may not recognize all isolates of strain group C [50] underlines the need to develop appropriate diagnostic methods for detection of these viruses. A PCR-RFLP-based method described by Tairo et al. [49] may be too expensive for routine detection purposes, especially in developing countries, which often lack facilities for using molecular techniques.

The identification of two of the potyvirus-specific amplicons as pertaining to SPV2 and SPVG presents the first report of the occurrence of these viruses in Peru and South America. Hence, all currently recognized sweet potato potyviruses are endemic to Peru. In routine testing from CIP’s germplasm collection and from seed production in Huaral and Cañete using specific antibodies, SPVG is frequently detected (c. 30% of symptomatic samples) and appears to be more prevalent than SPV2 (c. 3%). Despite this, SPFMV is by far the most common virus found (c. 90% of symptomatic samples), and consequently, the other potyviruses may contribute little to yield losses caused by SPVD. Although it has been shown that the titers of all these viruses increases upon co-infection with SPCSV, titers of SPFMV increase more than those of the other potyviruses [24, 47, 51]. These higher replication rates upon co-infection with SPCSV may provide a possible explanation for the prevalence of SPFMV over the other potyviruses, which are out-competed. This also implies that if cultivars with resistance to only SPFMV were deployed, the other potyviruses could rapidly replace it, causing similar synergistic virus diseases. Because available antisera to SPFMV show a weak serological cross-reaction in NCM ELISA with SPVG as well as SPV2, infection with these viruses may previously in many cases have been attributed to SPFMV, and their prevalence worldwide may be greater than previously known. The availability of SPVG- and SPV2-specific antibodies is now facilitating the detection of both viruses in samples from different countries, discriminating them from the presence of SPFMV.

Phylogenetic analysis of various samples of the three viruses identified in this study, together with a representative repertoire of other potyviruses, enabled us to show that these viruses form a well-supported phylogenetic subgroup within the genus Potyvirus (Fig. 4), together with an unknown virus reported from Zimbabwe [11]. Besides notable sequence similarity (including identical last 20 nts), this ‘SPFMV’ subgroup distinguishes itself by having a narrow host range, mainly confined to the Convolvulaceae, indicating a likely common evolutionary ancestor adapted to this family of hosts. On the other hand, SPLV and SPMSV, the two sweet potato-infecting potyviruses not belonging to the ‘SPFMV’ subgroup, are phylogenetically distantly related and have broader host ranges including Chenopodiaceae and Solanaceae [29]. Other potyvirus subgroups also have certain host specificities (Fig. 4), such as the ‘SCMV’, ‘BCMV’ and ‘PVY’ subgroups predominantly infecting gramineous, leguminous, and solanaceous plants, respectively [46, 48], suggesting a significant role for virus host co-evolution in potyvirus speciation. Further sequencing of SPV2, SPVG and SPFMV-C genomes may enable us to shed some light onto the specific characteristics required for adaptation to convolvulaceous hosts, although the identification of a highly conserved region in the P1 protein of SPFMV and the ipomovirus sweet potato mild mottle virus [52] already alludes to an important role for that protein in host specificity.