CTV isolates were obtained from the CCPP in planta Citrus Pathogen Collection at Riverside, CA. Source plants were graft-inoculated Madam Vinous sweet orange (C. sinensis L.) and represent isolates from imported citrus propagations from abroad as well as field sources collected in California over the past century. The CCPP CTV were separated in categories according to the following; location: SC; CC; excluded from California (XC) i.e., intercepted in California and eliminated so there was no field spread; CTV period: 1907–1957 (1), 1957–2009 (2); type of original source: A = CES and CCPP introductions, B = infected trees from urban landscapes or commercial groves; and C = interceptions by the CPDP (Table 1).

The MMM genotype analysis of CTV was performed using reverse transcription polymerase chain reaction (RT-PCR) with specific primers for the genotypes T30, VT, T36, T3, and B165/T68, as previously described (Hilf et al., 2005; Roy and Brlansky, 2010; Roy et al., 2010) (Table 2). Total RNA from ~0.2 g of bark tissue was extracted using the Spectrum Plant Total RNA kit (Sigma, Saint Louis, Missouri, USA) according to the manufacturer's instructions and eluted in 30 μl of RNase-free water. RT-PCR was performed using AMV Reverse Transcriptase Kit and GoTaq Hot Start Green Master Mix Kit (Promega, Madison, WI, USA) or Qiagen One-Step RT-PCR Kit (Qiagen, Germantown, MD, USA) using proper positive and negative controls as previously described (Hilf et al., 2005; Sharma et al., 2011). The RT-PCR products were analyzed using electrophoresis on 1% agarose gel and visualized over a UV transilluminator after ethidium bromide staining. All MMM reactions were repeated twice and at least two RT-PCR amplicons, per CTV isolate, were sequenced in order to verify homology with the corresponding CTV genome regions. Sequence analysis was performed with ClustalX (1.81), BioEdit (7.0.5.3), and GeneDoc (2.7.000) software (Nicholas and Nicolas, 1997; Thompson et al., 1997; Hall, 1999). The frequencies of the CTV genotypes were calculated as the sum of genotype counts in single, double, and triple mixtures.

A set of primers for universal CTV detection was designed from the genomic region of the CP gene (CP-Universal, CP-U) using the Primer 3 software (Rozen and Skaletsky, 2000) (Table 2). The RT-PCR products were purified using Zymo Research DNA Clean and Concentrator Kit (Zymo Research, Irvine, CA, USA), and then sequenced directly using the CP-U forward and reverse primers as previously described (Hajeri et al., 2011). For each of the 48 CTV isolates, the CP-U RT-PCR amplified products were sequenced in order to obtain the complete sequence of CTV CP gene (Table 2). Consensus sequences were assembled using DNA Dragon software (http://www.dna-dragon.com) with 2–3 × coverage per strand yielding a sequence of 672 nt in length corresponding to the CP gene. The CTV CP gene sequences acquired were deposited in GenBank (GenBank accession numbers KC841779-KC841826).

Thirty-four full-length CTV genome and CP gene sequences produced at previous studies available in the GenBank were analyzed phylogenetically (data not shown). Subsequently, eleven GenBank CTV sequences representatives of the identified phylogenetic clades, namely, T30 (AF260651), T36 (U16304), NUagA (AB046398), NZ-M16 (EU857538), VT (U56902), B165/T68 (EU076703), NZRB-M12 (FJ525431), A18 (JQ798289), SY568 (AF001623), HA16-5 (GQ454870), and T3 (KC525952) were selected for analysis with the CTV isolates in this study.

All topologies were reconstructed with neighbor-joining, maximum parsimony and maximum likelihood methods using MEGA5.1 software. The confidence level in tree topology was examined using bootstrap with 10,000 replicates (Tamura et al., 2011). The topologies from the three phylogenetic methods were similar so only the neighbor-joining phylogenic tree is presented.

CTV biological characterization was performed between 1970 and 2012 by graft-inoculation of Mexican lime [C. aurantifolia (Christm.) Swing.], Duncan grapefruit (C. paradisi Macf.), Eureka or Lisbon lemon (C. limonL. Burm.f.), sour orange (C. aurantium L.), and Madam Vinous or Pineapple sweet orange seedlings as previously described (Roistacher, 1991; Garnsey et al., 2005; Polek et al., 2005). CTV isolates were inoculated using at minimum six seedlings of each plant indicator, including negative and positive controls. Indicator plants were maintained under standard greenhouse conditions (24–28°C day and 17–21°C night temperatures), infection was confirmed by enzyme-linked immunosorbent assay (ELISA) and symptoms were rated 6–9 months post-inoculation. Symptoms were evaluated on a 0–5 scale (0 = negative; 1 = very mild; 2 = mild; 3 = moderate; 4 = severe; 5 = very severe) over several replicated experiments and the average score for each indicator among different experiments was used (Table 4). Finally, the sum score of the seedling yellows (SY) (0–15) and SP (0–10) indicators as well as the total score for all indicators (0–30) for each isolate were calculated and used for statistical analysis.

The biological characterization data were tested for normality using the Kolmogorov–Smirnov test. Normally distributed data were analyzed by One Way ANOVA, otherwise, the non-parametric Kruskal–Wallis test on ranks (ANOVA on Ranks) was conducted. Statistically significant differences among means were identified by the Holm-Sidak method. Normality test, ANOVA, and tests of significance were performed at p < 0.05 using the Sigma Plot 11.00 software (Copyright© 2008 Systat Software, Inc.; San Jose, CA, USA).

The MMM analysis is presented in Table 3. The identification of the T30 and VT genotypes in single infections were in general agreement (25 out of 26) between MMM-H and MMM-R (one isolate reacted differently). In contrast, seven CTV isolates reacted differently for MMM-H and MMM-R in the mixtures of T30 and VT genotypes (Table 3). The T36 genotype was identified only in mixtures with T30 and VT genotypes (Table 3).

Genotype T3 markers developed by Roy et al. (2010) are located in the exact same CTV genome area and contain the identical sequence of that of Hilf et al. (2005) except having a few extra nucleotides at the 3' end of primers (Table 2). Interestingly enough, the T3 markers reacted differently for the CTV isolates tested by MMM-H versus MMM-R. There was a general agreement (42 out of 48) on isolates identified as non-T3 but the two MMM systems identified six different isolates as T3 genotype (Table 3). Sequence analysis of the T3 MMM-H and MMM-R PCR products showed 85–89% similarity with the k17 region of ORF 1a of the newly characterized T3 representative (Harper, 2013). The highest sequence similarity (95–97%) was observed with the k17 region of the Indian isolates BAN-1 (AY285670), BAN-2 (AY285668), and B226 (AY285669) which have been reported as various mixtures of T30, VT, T36, and T3 genotypes (Roy and Brlansky, 2004). Hence, the determination of T3 genotypes in the CCPP CTV collection was inconclusive and the T3 genotype was excluded from further analysis.

Eighteen CTV isolates contained the single genotype T30 (37.5%) and eight isolates contained the VT (16.7%) genotype. No CTV isolate was found solely with the T36 genotype (Figure 1). The remaining 22 CTV isolates contained mixtures of two or three of T30, T36, and VT genotypes (45.8%). The most common genotype mixture was T30+VT identified in 16 isolates (33.3%). The genotype mixtures of T30+T36, VT+T36, and T30+VT+T36 represented 4.2% of the CTV isolates (Figure 1).

The CTV genotype frequencies and spatiotemporal distribution is presented in Figure 1. T30 and VT were identified at the same frequency for period 1. In period 2, the T30 genotype frequency increased while VT and T36 frequencies decreased (Figure 1). The T30 genotype frequency was similar for SC and CC (i.e., 79.4% and 87.5%, respectively). On the contrary, the VT genotype frequency was reduced to less than half in CC in comparison to SC while the T36 genotype was not detected in the CC isolates. All three genotypes were identified amongst the six California excluded (XC) isolates intercepted in the state but eliminated before field spread (Figure 1).

The diversity of the SC-2 CTV genotypes was reduced in comparison to the SC-1. Two genotype mixtures identified in SC-1 (VT+T36 and T30+VT+T36) were not identified in SC-2. In addition, the genotype frequency for T30 increased and T36 decreased in SC-2 in comparison to SC-1 (Figure 1). T30 and VT were the only genotypes identified in CC-2. The CC-2 genotypic diversity was not as big as the SC-1. For example, three genotype mixtures identified in SC-1 (T30+T36, VT+T36 and T30+VT+T36) were not identified in CC-2 while VT and T36 frequencies were reduced (Figure 1). Finally, the T30 genotype frequency was stable (~87%) for SC-2 and CC-2, however, VT and T36 frequencies were reduced in CC-2. All three genotypes were identified in CTV isolates originated from CES/CCPP introductions, urban/commercial groves, and CPDP interceptions (Figure 1).

GenBank contains a plethora of CTV sequences. In order to select appropriate CTV representative accessions for a meaningful analysis with the CCPP CTV isolates, GenBank CTV accessions were analyzed phylogenetically on their own. The analysis identified five distinct clades associated with the T30, T36, VT, B165/T68, and T3 CTV genotypes and three additional clades with isolates not related to any standard CTV genotype (data not shown). The CP gene sequence of 11 GenBank CTV accessions representing the eight identified phylogenetic clades was selected for analysis with the 48 CCPP CTV isolates in this study (Figure 2).

The 48 CCPP CTV isolates were clustered into seven distinct clades (Figure 2). SC and XC isolates were present in seven and four clades, respectively. In contrast, CC isolates were limited in two clades. California CTV isolates from periods 1 and 2 were well-distributed and present in five and six clades, respectively. CTV isolates from type A were present in six clades while types B and C were limited in three clades (Figure 2).

The majority of the CCPP CTV isolates (34 out of 48) clustered with representatives of the T30 and VT genotypes (clade 1 and 5). A total of six CCPP CTV isolates clustered with the representatives of T36, T3, and B165/T68 representatives (clade 3, 4, and 6). The remaining isolates clustered with the non-standard CTV genotypes NZRB-M12 (clade 2) and HA16-5 (clade 7) (Figure 2). It is worth noting here that NZRB-M12 and HA16-5 were strongly associated (bootstrap >85%) with CCPP CTV isolates of T30+T36 and VT genotype, respectively. (Vives et al., 2005; Harper et al., 2010; Melzer et al., 2010; Harper, 2013).

Results of biological indexing per isolate are presented in Table 4. Mexican lime reacted with all CTV isolates tested; three isolates did not induce SY; and 16 isolates did not induce SP reactions. Four CTV isolates reached the maximum SY sum score of 15 while nine additional isolates scored 12 and above. In contrast, no CTV isolate reached the maximum SP sum score of ten and only two isolates scored above seven (Table 4).

The biological characterization data were analyzed in relation to the location (SC, CC, and XC), period (1 and 2), type of original source (CES/CCPP introductions, urban/commercial groves, and CPDP interceptions), and genotype (T30, T36, and VT) of the CCPP CTV isolates (Tables 1, 4). Geographical location, time period, source type, genotype and CP gene phylogeny of the CTV isolates had no significant effect on the Mexican lime reactions (p > 0.05, ANOVA). In contrast, XC isolates had significantly higher biological activity scores than CC isolates (F₃₀ = 3.656, p = 0.039). Similarly, CTV isolates from period 1 induced more severe biological reactions than those from period 2 (F₃₀ = 5.802, p = 0.023). When geographical location and time period were co-analyzed, CTV isolates SC-1 and XC-2 induced significantly more severe biological reactions than SC-2 and CC-2 (F₃₀ = 6.324, p = 0.002). CTV isolates from CES/CCPP (type A) were significantly more virulent than the isolates naturally spreading in urban and commercial groves (type B) (F₃₀ = 5.008, p = 0.014). CTV isolates with T30 genotype induced significantly milder biological activity scores in comparison to VT and T30+VT+T36 (F₃₀ = 3.123, p = 0.025). The milder biological activity of T30 genotypes compared to VT genotypes was also supported by the significantly different biological score of CTV isolates that clustered in clade 5 (VT representatives) versus clade 1 (T30 representatives) (F₃₀ = 10.014, p < 0.001).

SY and SP severity scores were significantly correlated with time periods and geographical location of the CTV isolates, respectively. Period 1 isolates induced significantly more severe SY (H₁ = 6.165, p = 0.013) and XC isolates induced significantly more severe SP (H₂ = 7.636, p = 0.022). Spatiotemporal analysis of the SY and SP reactions indicated significantly higher SY reactions for SC-1 isolates in comparison to SC-2 and CC-2 (F₃₀ = 5.033, p = 0.007) and significantly higher SP reaction for XC-2 in comparison to SC-1 and -2 and CC-2 (H₃ = 8.035, p = 0.045). CTV isolates clustered in clade 5 (VT representative) had significantly more severe SY reactions in comparison to isolates from clade 1 (T30 representative) (F₂₇ = 8.297, p < 0.001). CTV source types and genotypes had no significant effects in SY and SP reactions (p > 0.05).

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.