- Research article
- Open Access
Do potatoes and tomatoes have a single evolutionary history, and what proportion of the genome supports this history?
© Rodriguez et al; licensee BioMed Central Ltd. 2009
- Received: 2 February 2009
- Accepted: 7 August 2009
- Published: 7 August 2009
Phylogenies reconstructed with only one or a few independently inherited loci may be unresolved or incongruent due to taxon and gene sampling, horizontal gene transfer, or differential selection and lineage sorting at individual loci. In an effort to remedy this situation, we examined the utility of conserved orthologous set (COSII) nuclear loci to elucidate the phylogenetic relationships among 29 diploid Solanum species in the sister clades that include tomato and potato, and in Datura inoxia as a far outgroup. We screened 40 COSII markers with intron content over 60% that are mapped in different chromosomes; selected a subset of 19 by the presence of single band amplification of size mostly between 600 and 1200 bp; sequenced these 19 COSII markers, and performed phylogenetic analyses with individual and concatenated datasets. The present study attempts to provide a fully resolved phylogeny among the main clades in potato and tomato that can help to identify the appropriate markers for future studies using additional species.
Among potatoes, when total evidence is invoked, one single predominant history is highlighted with complete resolution within and among the three main clades. It also supports the hypothesis of the North and Central American B-genome origin of the tuber-bearing members of Solanum sect. Petota and shows a clear division between A genomes in clades 3 and 4, and B genomes in clade 1+2. On the other hand, when a prior agreement approach is invoked other potato evolutionary histories are revealed but with less support. These alternative histories could be explained by past hybridization, or fast rates of speciation. In the case of tomato, the analyses with all sequence data completely resolved 19 of 21 clades, for the first time revealed the monophyly of five clades, and gave further support for the recent segregation of new species from the former Solanum peruvianum. Concordance analyses revealed and summarized the extensive discordance among COSII markers. Some potential reasons for discordance could be methodological, to include systematic errors due to using a wrong model of sequence evolution, coupled with long branches, or mixtures of branch lengths within COSII, or undetected paralogy or alignment bias. Other reasons could be biological processes such as hybridization or lineage sorting.
This study confirms and quantifies the utility of using DNA sequences from different parts of the genome in phylogenetic studies to avoid possible bias in the sampling. It shows that 11–18 loci are enough to get the dominant history in this group of Solanum, but more loci would be needed to discern the distribution of gene genealogies in more depth, and thus detect which mechanism most likely shaped the discordance.
- Total Evidence
- Consensus Approach
- Concordance Analysis
- COSII Marker
- Intron Content
The basic chromosome number in potatoes (sect. Petota), tomatoes (sect. Lycopersicum), and the most closely related outgroups (sects. Etuberosum, Juglandifolium, and Lycopersicoides) is 2n = 2x = 24. Potatoes are alone in the group in possessing polyploids. Approximately 70% of the over 100 potato species are diploids, with most of the rest tetraploids (2n = 4x = 48) and hexaploids (2n = 6x = 72), with rare triploids and pentaploids .
Potatoes, tomatoes, and outgroups are characterized by relatively small chromosomes. Karyotypic and genomic analyses have included crossability success of interspecific combinations, hybrid sterility, hybrid viability, pollen fertility and in the degree of chromosomal homology [2–9]. Chromosome pairing relationships in interspecific hybrids and in polyploid potato species have been interpreted by genome formulae, although authors gave them different symbols. According to these hypotheses, five genomes (A, B, C, D and P) are recognized in the tuber-bearing species of section Petota. Genome symbol E was given to species of section Etuberosum based on the specificity of meiotic behavior and sterility of their diploid hybrids with the A-genome tuber-bearing potato species [7, 10]. Symbol L was proposed for tomato (section Lycopersicon) on the bases of preferential chromosome pairing and clear-cut parental genome discrimination by using genomic in situ hybridization (GISH) and LLEE or artificial amphidiploids of tomato and S. etuberosum .
Wild and cultivated potatoes (Solanum L. sect. Petota) (see Additional file 1 for authors of taxa) grow from the southwestern United States to southern Chile. Hawkes  recognized 232 species in section Petota, and divided it into 19 tuber-bearing and two non-tuber-bearing series. He further divided these 19 tuber-bearing series into two superseries based on corolla morphology (superseries Stellata with stellate corollas and superseries Rotata with rotate corollas). He distinguished "primitive" and "advanced" forms of each superseries and recognized four groups based on morphological characters: primitive Stellata, advanced Stellata, primitive Rotata, and advanced Rotata. He hypothesized the evolution of the advanced Rotata morphology from primitive Stellata-like ancestors. He postulated that the ancestral wild potato species were diploid, possessed B genomes, produced white stellate corollas, and originated in North or Central America in the late Cretaceous to Eocene eras. Subsequent dispersal of one or more of these species to South America took place before the sinking of the Central American land bridge during the mid-Eocene to Pliocene eras, gradually leading to the evolution of species with A genomes and rotate corollas. He also postulated that when the bridge was restored in Pliocene times, a remigration of one or more of these diverged A genome species back into North and Central America allowed the hybridization and allopolyploidization with the native Mexican or Central American B genome taxa. This produced the tetraploid members of series Longipedicellata (AB). A second, comparatively recent migration of a second species (S. verrucosum, A genome) from South America formed the Demissa hexaploids (A1A4 [B,C,D])  by crossing with series Longipedicellata tetraploids and possibly other series. In North and Central America are also another group of allopolyploids, members of series Conicibaccata (AC), but only series Longipedicellata was designated AB, and the source of the C and D genome donors of series Conicibaccata and Demissa is unknown. However, recent molecular clock data indicated that eggplant and tomato/potato shared a common ancestor approximately 14 MYA and potato and tomato 7 MYA . So it seems unlikely that the timing for the events proposed by Hawkes really took place in the Cretaceous – Eocene eras but the events are still possible.
The latest taxonomic interpretation  recognized fewer species (190) and predicted even further reductions in species. Spooner et al.  used plastid DNA restriction site data and morphological data to reinvestigate the relationships of potatoes, tomatoes (former genus Lycopersicon), farther outgroups in Solanum, and other genera of the Solanaceae. Their results confirmed placement of all members of Hawkes's  tuber-bearing species into sect. Petota (90% bootstrap), but the non-tuber-bearing species were shown to be outgroups although with very low support (68%). Yet other plastid DNA restriction site marker data and DNA sequence data from plastids and single-copy nuclear DNA supported these relationships [15–18].
Taxonomic interpretations in tomato (Solanum sect. Lycopersicon) differed widely depending upon whether morphological or biological species concepts were used . Rick et al.  recognized nine species of tomatoes, based mainly by their ability to intercross, and divided them into two crossing groups. Peralta et al.  summarized morphological studies of tomatoes [18, 26–30], increased the traditional nine species of Rick et al.  to 13 species, and recognized tomatoes within Solanum. Tomato has two endemic species in the Galápagos Islands, and weedy escaped forms of S. lycopersicum are distributed worldwide. The remaining species and immediate tomato outgroup sections Juglandifolia (two species) and Lycopersicoides (two species) are endemic to western South America from Ecuador to northern Chile.
Peralta et al.  interpreted the many and sometimes conflicting morphological and molecular data sets to recognize four informal "species groups" within sect. Lycopersicon: the "Lycopersicon species group" including the four red to orange colored fruited species S. cheesmaniae, S. galapagense, S. lycopersicum, and S. pimpinellifolium; the "Neolycopersicon species group" containing only S. pennellii, the "Eriopersicon species group" containing S. chilense, S. corneliomulleri, S. habrochaites, S. huaylasense, S. peruvianum and, the "Arcanum species group" including S. arcanum, S. chmielewskii, and S. neorickii. They recognized sect. Juglandifolia as the sister group to sect. Lycopersicon.
In the past, comparative anatomy, cytogenetics, ecology and morphology had been used to infer phylogenetic relationships among taxa. Lately, with the development of molecular techniques, many fragments of the genome have been used in phylogenetic inference and multiple data sets have become available. As a result, phylogenetic studies now usually involve data sets from different gene regions, or different marker systems, or molecular and morphological data sets. There has been an active and controversial debate over the best ways to derive a phylogeny that incorporates all existing data. There are three different traditional approaches that have been used to analyze data from different sources: taxonomic congruence, character congruence and the prior agreement approach. They all seek to maximize evidence, the first one from the relationships (cladograms) using the summaries of data as evidence, and the other two derived directly from characters using data observation as evidence .
Proponents of the consensus approach argue that partitions should not be combined before estimating the phylogenetic tree; rather, the trees should be estimated separately from each partition and combined using taxonomic congruence . The decision to keep data sets separate generally reflects a hypothesis that either 1) different evolutionary processes are acting on different data sets, or 2) different data sets reflect different evolutionary histories .
Character congruence, also known as the total evidence approach, involves all the data available for a group of terminal taxa. The goal is to seek a single, best-fitting hypothesis, which in cladistics involves maximizing character congruence . The prior agreement approach or "conditional combination" tests for character incongruence before analysis, and only combines the data when no significant character incongruence exists between partitions.
Farris et al.  proposed the Incongruence Length Difference (ILD) test that many researchers use to assess whether or not data sets should be combined. The idea is to determine if there is more incongruence between data sets than is expected between similarly sized random partitions of a homogeneous data set.
Baum  proposed that the identification of the primary concordance tree (PCT) could be a valuable summary of the dominant phylogenetic history among a group of organisms. The PCT is composed of clades with higher concordance factors (CF, proportion of the genome for which a given clade is true) than any contradictory clade. Ané et al.  developed a Bayesian-based methodology that estimates the distribution of evolutionary histories within a multi-gene data set and summarizes the results allowing genealogical information from one gene to influence our estimates of another gene's genealogy. It also allows the identification of genes with outlier gene-tree topologies that permits estimation of the proportion of the genome that was transferred during a possible introgression event. Baum  pointed out the possible outcomes that can arise if reticulation events had happened during the evolutionary history of a group of organisms. If a single event of horizontal gene transfer (HGT) had occurred, one primary history and one minor history are expected. Under lineage sorting or introgression, one primary history and co-minor histories are expected. But, if hybridization occurs two co-primary histories are expected. Thus, the concordance approach is considered a modified consensus tree method that allows the identification of different phylogenetic histories in the face of reticulation, where estimated CFs are metrics that give an indirect measure of reticulation. Furthermore, it can be assumed that high CFs provide evidence that relationships among the organisms under study have been mostly divergent, and that low CFs are indicative of reticulate evolution. Baum  stressed that the acquisition of a divergence structure increases gradually as a result of gene lineage extinction in reproductively isolated populations or demes. Thus, the speciation process reaches its ultimate point when a CF of 1 is reached. Finally, he stated that a clade which has a CF that is greater than all contradictory clades has some genealogical unity that indicates a history of genetic isolation and suggested that this clade can be formally named.
Recently, single- to low-copy nuclear DNA markers have been explored for phylogenetic reconstruction. Here we test a specific subset of these markers called conserved ortholog set II (COSII) markers in the sister clades potato and tomato . Orthologs are genes sharing a common ancestor by descent, in contrast to paralogs that are duplicated copies within a genome through polyploidization or tandem duplications [38, 39].
The purpose of our study is to test a diverse range of COSII markers and data analyses for phylogenetic reconstruction in the sister clades tomato and potato. The phylogenetic relationships between and among species from these two groups have been the subject of study by many researchers, but these relationships remain unresolved because of the use of a limited number of molecular markers. Our study was greatly aided by COSII conversion into a set of consensus PCR primers that amplify these orthologues in the Solanaceae, referred to as universal primers for euasterid I (UPA) . UPAs are designed to amplify either intronic regions (iUPA primers) that provide the polymorphism needed for the proposed tomato and potato study, or exonic regions (eUPA primers) that are more appropriate for comparisons of more distantly related species. Our study uses many more molecular markers than prior phylogenetic studies in the tomato and potato clades. It does not attempt to provide a definitive study of the evolutionary relationships within these clades, but rather to identify appropriate markers for future studies that should incorporate a wider species sampling. Our approach consists of examining 1) diverse markers on all 12 linkage groups, 2) amplification of a single band (in diploids) of representative taxa, 3) markers with length between 600–1200 bp that amplify with single pass sequencing, and 4) markers with over 60% intron content appropriate for this taxonomic level of closely related species.
Allelic variation and sequence alignment
Of the 30 species analyzed, four tomato species had two alleles: S. chilense and S. huaylasense in COS5c and COS13b, S. peruvianum and S. habrochaites in COS7b, S. huaylasense in COS10b, S. peruvianum in COSX2 and COS11b, S. chilense in COS1b; S. chilense, S. peruvianum and S. habrochaites in COS8b, and S. chilense, S. peruvianum and S. huaylasense in COS15b (16 cases in total). In four of these 16 cases alleles from the same species form a clade; in four cases alleles formed polytomies; and in eight cases alleles from the same species formed a clade with another species .
Complete sequences (total missing data less than 0.3%) were generated for the 19 COSII. In general, the farthest outgroups, Datura inoxia and S. dulcamara, showed the longest sequence and the tomato species and their closest outgroups showed more gaps in aligned sequences in comparison with potato .
In general, likelihood models that account for rate variation, allowing for gamma-distributed rate variation among sites (Γ), resulted in the greatest increase in Maximum Likelihood (ML). In more than 90% of the cases the HKY model that includes five parameters best fitted the data . In cases where the two criteria for selection chose different models, the model that is implemented in Mr. Bayes was preferred.
Phylogenetic analysis of individual loci
Analysis of individual COSII sometimes showed incongruent results. To illustrate this we discuss our data using maximum parsimony analysis in potato and direct the reader to  for details in the tomato and complete data sets and with maximum likelihood and Bayesian inference. In the potato data set the number of parsimony informative sites per individual COSII varied from 13 to 64; in tomato, from 20 to 286; and for the potato and tomato data set from 46 to 317. All data sets showed high consistency, retention and rescaled consistency indices, with consistency indices ranging from 0.753 to 1.000, and rescaled consistency indices ranging from 0.612 to 1.000. There was a moderate level of homoplasy (calculated by H = 1-CI) with less than 25% for all loci.
In the potato data set seven of the 12 COSII sequenced supported the three main potato clades as a monophyletic group. In eight of the 12 COSII clade 1+2 (S. bulbocastanum, S. polyadenium, S. stenophyllidium, S. trifidum) was monophyletic with bootstrap support over 54%; in three of the 12 COSII, clade 1+2 failed to resolve into a distinct clade; one COSII placed S. polyadenium outside of clade 1+2. When we forced the monophyly of clade 1+2 for this COSII, the shortest tree was ten steps longer than the unconstrained tree, which was significant by the Templeton Test (p = 0.0016–0.0075) and the monophyly of this clade was rejected .
Clade 3 (S. albornozii, S. andreanum) had bootstrap support more than 50% in nine of the 12 COSII; one COSII failed to group these species in their own clade, and two COSII grouped one of the two species in clade 4. Forcing these species to be monophyletic resulted in trees only three and two steps longer than the unconstrained tree, and the Templeton test failed to reject their monophyly (p = 0.3657 and p = 1). Thus, stochastic noise alone can explain the misplacement of these species.
Clade 4 (S. brevicaule, S. raphanifolium, S. verrucosum) had bootstrap support more than 59% in ten of the 12 COSII; in the remaining two COSII S. raphanifolium did not form a clade with the other two species, and trees forcing all three species to group together were only one and three steps longer than the Maximum Parsimony (MP) unconstrained tree. The Templeton test failed to reject their monophyly and revealed that the entire clade 4 became unresolved in trees that were one and three steps longer than the MP tree. This indicates that the observed non-monophyly is not significant (p = 0.8635 and p = 0.0833–0.3173 respectively) .
Regarding the relationships among the three main clades, four of the 12 COSII supported clade 1+2 as sister to the rest of potatoes, with bootstrap support ranging from 59–100%. Three COSII placed clade 3 as sister to the rest of potatoes and only two of these three COSII gave bootstrap support of 69% and 90%. Two COSII placed clade 4 as sister to the rest of potatoes with a bootstrap support of 60% and 100% respectively. The remaining three COSII did not support any of the three basal relationship alternatives in their most parsimonious trees. Attempts to force species to follow either of these three possibilities were not rejected for COS2c, because constrained trees were only four or five steps longer than the MP unconstrained tree. In contrast, COS5 could not reject clade 1+2 as sister to the other potatoes, and COS5c could not reject clade 3 species as sister to the others.
Combined data analyses
We explored three approaches to combine the data: total evidence, prior agreement, and the consensus approach.
Total evidence, potato data set
Total evidence, tomato data set
Total evidence, potato and tomato data set
Prior agreement, potato data set
A prior agreement approach combines only congruent data and excludes data partitions with a significant level of incongruence, as they can introduce error that can obscure reliable data and lead to erroneous topologies [41, 42]. We explored the congruence of the different gene phylogenies in potato using ILD tests. From the 12 original COSII in potatoes, 66 pairwise comparisons were made with a level of significance of 0.00076 (= 0.05/66), and nine of them showed a high level of discordance (p = 0.0006). Three of these COSII (COS2c, COS5c and COS7b) counted for eight of these nine discordances (p = 0.0004), and the last one was between COS11 and COS10b (p = 0.0006).
Through reiterative analyses we proceeded to eliminate discordant COSII and constructed a set of six COSII that were highly congruent and produced highly resolved topologies similar to the entire concatenated data set and named this "6COScon" (COS 9, COS 9b, COS 11, COS 3, COS 8 and COS 3c) Rodriguez . MP, ML, and BI of 6COScon gave the same phylogeny as the 12 COSII combined with 100% of bootstrap in all nodes except one that was supported at 99% (Figure 2b).
We further explored a minimum number of COSII to be used in a large number of species without losing resolution by further reducing the "6COScon" to three COSII by choosing the best three by two criteria: 1) length of the sequence, and 2) number of parsimony informative sites and called this "3COSa" (COS9, COS9b, COS11). MP, ML, and Bayesian analyses produced phylogenetic results almost identical to the 12 COSII set (Figure 2c).
Consensus approach, potato data set
Consensus approach, tomato data set
Results from the second tomato concordance analysis placed the clade consisting of S. habrochaites and S. pennellii as sister to the rest of tomatoes, indicating that those are the most divergent group; the CF was 30% (5.38 COSII) with α infinite, 63% (11.34 COSII) for α = 10 and 66% (12 COSII) for α = 1, and the credibility intervals for the CF of conflicting clades did not overlap. Thus, we conclude that S. habrochaites and S. pennellii are the most divergent tomatoes and are monophyletic (Figure 7). This analysis also placed S. corneliomulleri with S. peruvianum in all the runs with CF of 33% to 47% of COSII, and S. chilense outside of the clade conformed by S. corneliomulleri-S. lycopersicum-S. peruvianum, with CF ranging from 18% to 32% (Figure 7).
In the last concordance analysis with the tomato data set, all 18 COSII supported S. huaylasense as the sister to two clades, one formed by the red-orange colored fruit tomatoes (S. cheesmaniae, S. galapagense, S. lycopersicoides, S. pimpinellifolium) and the other formed by three species, S. arcanum-S. chmielewskii-S. neorickii; it also resolved S. chmielewskii with S. arcanum in all three α values (Figure 7).
As was pointed out in the background, all three approaches summarizing multiple data sets seek to maximize evidence, but there are questions about when each approach should be applied. Combined analysis of multiple data sets is justified when each data set has evolved under the same underlying history, in which differences in the estimated tree are due only to sampling error or model mis-specification . Under this assumption, the combined phylogenetic analysis improves the signal to noise ratio. It potentially allows for more accurate estimation of the single shared genealogy. It also produces a more accurate phylogeny than from a consensus approach. But, in cases where genes have tracked more than one underlying history, some of the differences among data sets would not be due to sampling error, but to genealogical discordance. In this case we may wish to estimate the primary history with information quantifying the extent to which different genes have followed that history. That primary history can be obtained using a combined analysis if there are similar amounts of phylogenetic signal in each sampled gene. But, because combined analysis assumes that there is only one evolutionary history, the discordance is assumed to be caused by homoplasy, and when reticulation has occurred the inferred tree may not resemble any of the underlying histories. In such a case, Bayesian concordance analysis (BCA) would be a better procedure to use, because it does not assume one single evolutionary history. Rather, it allows us to make statements of statistical confidence in the estimated concordance factors by taking into account the prior evidence of discordance, uncertainty in gene tree estimates, and the number of genes sampled. It also allows the investigation of potentially interesting biological processes, such as incomplete lineage sorting, hybridization, and lateral gene transfer . For this reason, BCA is becoming a tool to analyze genealogies that are neither fully reticulate nor fully divergent . Therefore, we followed all three approaches to get insights into the evolutionary processes in tomato and potato.
COSII analyses support many aspects of prior phylogenetic interpretations of potato and tomato, show some new relationships, but most notably, when total evidence is invoked, one single predominant history is highlighted with complete resolution in potato (Figure 2a) and nearly so in tomato (Figure 3). In potato, the four clades discovered from plastid DNA restriction site data [19, 21] is supported except, like waxy sequence data  and NIA sequence data  clades 1 and 2 are united into a single clade. Results from total evidence highly support the hypothesis of a North and Central American B-genome origin of the tuber-bearing members of Solanum section Petota, as it was shown to be sister of the rest of potatoes. It resolves the controversy among all previous individual gene sequencing studies that showed different relationships among main clades. On the other hand, when a prior agreement or consensus approach was invoked, other evolutionary histories in potatoes are revealed but with less support (Fig 2d, 2e, Figures 6, 7). Of interest is that the branches showing these two alternative histories (Figure 2d, 2e) are shorter than the branch showing the predominant evolutionary history (Figure 2a). These alternative histories could be explained by past hybridization events or fast rates of speciation.
In tomato, 19 of the 21 nodes are completely resolved and for the first time this study highly supports the monophyly of five clades: 1) a clade that includes S. arcanum-S. chmielewskii-S. neorickii, 2) a clade conformed by S. chilense-S. corneliomulleri-S. peruvianum, and the sister relationship between S. corneliomulleri and S. peruvianum; 3) a clade formed by S. habrochaites-S. pennellii as basal in tomatoes, 4) a clade that includes S. cheesmaniae and S. galapagense, and 5) a clade formed by S. lycopersicum and S. pimpinellifolium (Figure 3). COSII data also clearly support the segregation of S. peruvianum sensu lato into at least three species, S. arcanum, S. huaylasense, S. peruvianum+S. corneliomulleri (on the same clade) as a possible single species, or all four of these species, following Peralta et al. . The data show S. arcanum and S. huaylasense to be very distinct from S. corneliomulleri and S. peruvianum that were on the same clade (Figure 3). The two branches that remain unresolved are the placement of the clade S. arcanum-S. chmielewskii-S. neorickii and the relationship among these three species. Both of these branches are very short (Figure 3), indicating the possible need for additional markers as was pointed out by Zou et al. , who showed the need for 142 gene sequences to resolve two small branches in the phylogeny of rice. The sister group relationships of sect. Lycopersicon to sect. Juglandifolia and to sect. Lycopersicoides were completely resolved thereby clarifying ambiguities in previous studies.
Concordance analyses in tomatoes reveal the same relationships as total evidence, but the monophyly of the clade with S. chilense-S. corneliomulleri-S. peruvianum is not supported and the placement of sect. Juglandifolia or sect. Lycopersicoides as the closest tomato outgroup remained unresolved. Furthermore, concordance analyses reveal and summarize the extensive discordance among COSII markers. This discordance is precisely why a large sample of nuclear loci was necessary for the reconstruction of a robust phylogeny in potato and tomato. Some potential reasons for discordance could be methodological, to include systematic errors due to using a wrong model of sequence evolution, coupled with long branches, or mixtures of branch lengths within COSII, or undetected paralogy or alignment bias. Other reasons could be biological processes such as hybridization or lineage sorting or fast speciation.
In the complete data set, where potatoes and tomatoes are analyzed together, the sister relationship between the potato and tomato clades is highly supported in all analyses and sect. Etuberosum is supported as sister to both (100%, 100% and 1), confirming these controversial relationships as was first shown by Spooner et al. . Additionally, our study highlights the importance of high intron content (more than 60% in this study), and PCR amplification length between 600 and 1000 bp to be useful to investigate relationships among potato and potato. It shows the importance of intensive screening of COSII primers to maximize their utility in potato and tomato and allows us to propose the use of a reduced set of COSII primers for maximum efficiency. For example, for the major clades of sect. Petota, three COSII (COS9, COS9b and COS11) were enough to reveal the predominant history with high support, and we suggest beginning with these COSII in future studies in potato. This suggestion has to be taken with caution because when more taxa and more accessions per taxon are sampled the resolution can be reduced. Finally, the utility of gaps as additional characters in phylogenetic studies is not questionable but it is advisable to use them when their inclusion does not increase homoplasy. However, in this study, gap characters usually increased phylogenetic signal.
COSII sequence data are useful for potato and tomato phylogenetic studies. Intron contents more than 60% were the best to investigate relationships within this group of closely related species. When total evidence is invoked, a well-supported predominant history is revealed in both potato and tomato. In the case of potato, with the prior agreement and consensus approaches, two additional possible histories were revealed, although with less statistical support. We show that 11–18 loci are enough to get the dominant history in potato and tomato respectively, but a reduced set of three COSII provide the same phylogeny in potato. Finally, we determined that there is not a single evolutionary history for potato and tomato, as at least three different possible histories were revealed. More loci would be needed to discern the distribution of gene genealogies in more depth, and thus detect which mechanism most likely shaped discordance among individual COSII.
Plant Materials and DNA Isolation
One diploid genotype from nine wild potato species (sect. Petota) representing all four plastid clades, all 13 wild tomato species (sect. Lycopersicon), both species of sect. Juglandifolia, both species of sect. Lycopersicoides, two species of sect. Etuberosum, and two farthest outgroups, Datura inoxia and S. dulcamara were analyzed (30 species in total; see Additional file 1). These species were chosen to cover most of the genomic groups and clades in potato, all of the species recognized in tomato , and appropriate outgroups. Leaf total DNA from each species was isolated using a standard CTAB protocol . DNA quality and quantity were estimated by comparison with CsCl-purified λ DNA digested with PstI on ethidium bromide-stained agarose gels.
Choice, amplification, and sequencing of COSII
We tested 40 COSII loci that were putatively single copy in tomato and potato, from all 12 tomato and potato chromosomes (see Additional file 2). We screened these 40 COSII in genomic DNA of four tomato and four potato species. PCR amplifications were performed using 20 μL reactions consisting of 0.1 μM final concentration of each primer (see Additional file 2) and 20 ng of template genomic DNA. Amplifications were carried out in an MJ Research DNA Engine Dyad® Peltier Thermal Cycler (Watertown, Massachusetts) using an initial denaturation at 94°C for 5 min, followed by 35 cycles of 94°C for 30 s, 55°C for 1 min, 72°C for 2 min and with a final elongation at 72°C for 10 min. The reactions were run on a 1.5% agarose gel with 1× TBE buffer for 3 hours. The above PCR conditions worked for most COSII but in some cases we had to modify these PCR conditions (see Additional file 3).
We selected 19 of these 40 loci based on single band amplification and length of the PCR product mostly between 600–1200 bp and with more than 60% intron content (see Additional file 3), and amplified DNA in all 30 species. To clean the product for sequencing the reaction was digested with EXO-SAP-IT® (USB Corp., Cleveland OH) following the manufacturer's instructions except that reaction volumes were halved. One μL of this product was sequenced with the same primers in a 5 μL reaction using the ABI Big Dye dideoxynucelotide termination kit (Applied Biosystem, Foster City, California). Amplifications were carried out in an MJ Research DNA Engine Dyad® Peltier Thermal Cycler (Watertown, Massachusetts) using an initial denaturation at 95°C for 3 min, followed by 30 cycles of 96°C for 25 s, 50°C for 20 s, 60°C for 5 min and with a final elongation at 72°C for 7 min. Excess of dye terminators were removed using CleanSeq magnetic bead sequencing reaction clean up kit from Agencourt Biosciences (Beverly, MA). Sequences were resolved on an ABI 3730xl capillary-based automated DNA sequencer (Applied Biosystems) with 50 cm POP-7 polymer capillaries at the Biotechnology Center of the University of Wisconsin-Madison. When a faint second band appeared or when the sequence was unreadable, the PCR product was cloned and five positive colonies were sequenced. Subsequently we learned that 18 of these 19 COSII are single copy on tomato, 12 in potato and 11 in both potato and tomato (see Additional file 3).
Sequence editing and alignment
Sequences were edited with Staden package version-2003.0-beta  and aligned using ClustalX version 1.81  at default parameters, except for the "percentage of delay divergence sequences" which was set to 15% after tests of various parameters. Further manual alignments were done in MacClade 4.06 PPC  minimizing the number of gaps and preferring transitions to transversions. Indels were scored by the simple gap scoring method  using SeqState 1.40 . DNA sequences were deposited in GenBank (see Additional file 4).
All analyses were conducted on three data sets: 1) a "potato data set" of 12 COSII across 11 species, nine of sect. Petota and two of sect. Etuberosum, 2) a "tomato data set" of 18 COSII across 19 species that included all species except members of sect. Petota and sect. Etuberosum, 3) the "complete data set" of 11 COSII across all 30 species. Phylogenetic analyses based on maximum parsimony were performed using PAUP* 4.0b10 . The most parsimonious trees were found by heuristic searches under Fitch criteria and with equal weight for all characters. A heuristic search was performed using TBR branch swapping on 100,000 random taxon addition sequences. A rooted strict-consensus tree was obtained using S. etuberosum and S. palustre as outgroups for the potato data set, and Datura inoxia and S. dulcamara for the other two data sets. Bootstrap support  was estimated with 10,000 bootstrap replicates performing a TBR branch swapping on 100,000 random taxon addition sequences.
Maximum likelihood analysis was used to assess 56 models evaluated in Modeltest ver. 3.07  using the Akaike Information Criterion (AIC) and hierarchical likelihood ratio test (hLRT) at α = 0.01. ML phylogeny was estimated using RAxML 7.0.3  that allows each partition (each COSII) to have its own model and parameters. To evaluate the stability of clades on the optimal tree, a bootstrap analysis was performed with 100 bootstrap replicates.
Bayesian Markov chain Monte Carlo (MCMC) phylogenetic analysis  also was performed using MrBayes version 3.1.2 [56, 57]. Model parameters for DNA data were chosen according to the criteria described above, they were estimated separately for DNA and gaps; for gaps the restriction site (binary) model was used (lset coding = variable) which accounts for the ascertainment bias produced by characters that are constant (either state 0 or 1) in all taxa and are not observed. Tree searching using MrBayes was performed by four runs of four linked chains ("temp" was empirically determined) for 1,100,000 or 2,200,000 generations with trees sampled every 100 generations. At the end of the analysis we checked that the average standard deviation of split frequencies was below 0.01 to ensure that convergence occurred properly. A conservative burn-in period was determined, and only post-burn-in trees were saved. The three sets of post-burn-in trees were then combined to form a majority rule consensus tree, and this pool was taken as the best representation of the posterior distribution of tree topology and model parameters . The proportion of searches in which any given node is found during the post burn-in portion of the chain constitutes the Bayesian posterior probability (PP) for that node.
The ILD test  was used to explore the difference in phylogenetic signal between and among data partitions, and in the potato data set was used to implement the prior agreement approach, as implemented through the partition homogeneity test in PAUP* 4.0b10 . Multiple ILD tests comparing all COSII sequenced in each data set were performed and the sequential Bonferroni correction was used to control for multiple comparisons [34, 58]. First, we performed one ILD test to see if all COSII sequenced in each data set can be combined together. Given that they did not pass the test, we proceeded to make all the possible pairwise comparisons within each data set to identify the COSII that were congruent. A total of 66 pairwise comparisons were made for the potato data set, 153 for the tomato data set and 55 for the potato and tomato data set. The Templeton test  as implemented in PAUP* 4.0b10  was used to explore alternative topologies in a parsimony framework.
To assess the total evidence approach we performed a single analysis with the concatenated data set. For the prior agreement approach we were able to identify groups of COSII that gave congruent results only with the potato data set. For the consensus approach, Bayesian concordance analysis (BCA)  was performed in the potato and tomato data sets separately, to identify the best estimate of the phylogeny of each COSII and to create a consensus of these separate point estimates. Three different prior levels of discordance: α = 1, 10, and infinite were used. For instance, an α value of zero corresponds to the total evidence approach that insists there is a single tree for all genes, and an infinite α value corresponds to an assumption of independence between gene trees. With the potato data set four independent runs with four linked chains were performed for all three α values; in each run with 1,100,000 generations, 100,000 of which were discarded as the burn-in period. For heating chains, the option used was m = 50. Runs showed good mixing and converged to the same results since the standard deviation of concordance factors was < 0.001 in all cases.
For tomato concordance analysis, it was impossible to perform a single concordance analysis with all of the species because the trees sampled during the Bayesian analysis of four COSII (COS13b, COS1b, COS4 and COS8) were all distinct, and it was necessary to prune some species to reduce the number of trees sampled in the Bayesian analysis to assure convergence to stationary distribution within an acceptable error. For this reason, we performed three independent analyses: 1) to determine the placement of the sections that are sister to tomatoes, a concordance analysis was performed with sequence of 18 COSII in 12 OTUs that included species or allelic variants of Datura inoxia, S. dulcamara, S. lycopersicoides, S. sitiens, S. juglandifolium, S. ochranthum, S. pennellii, S. corneliomulleri, S. huaylasense-1, S. huaylasense-2, S. arcanum and S. lycopersicum. For all three α values, four independent runs with four linked chains were performed. In each run of 22 million generations, 10% of the generations were discarded as the burn-in period; and for heating chains we used the option m = 50. Runs showed good mixing and converged to the same results since the standard deviation of concordance factors was < 0.005 in each case. 2) The second concordance analysis was conducted to determine if S. habrochaites and S. pennellii are the most divergent tomato species and to determine the placement of S. chilense, S. corneliomulleri, and S. peruvianum. We conducted a concordance analysis with all 18 COSII and with ten OTUs (S. juglandifolium, S. pennellii, S. habrochaites-1, S. habrochaites-2, S. chilense-1, S. chilense-2, S. corneliomulleri, S. peruvianum-1, S. peruvianum-2, and S. lycopersicum) using the same parameters that were used for the outgroups. Runs showed good mixing (SD of concordance factors was < 0.005 in each α) 3) The last concordance analysis with the tomato data set with all 18 COSII was conducted to determine the placement of S. huaylasense, and to determine relationships among S. arcanum, S. chmielewskii, and S. neorickii. Four independent runs of four linked chains were performed for each α with seven OTUs (S. pennellii, S. huaylasense-1, S. huaylasense-2, S. neorickii, S. arcanum, S. chmielewskii and S. lycopersicum).
This paper represents partial fulfillment of a PhD Degree for FR in Plant Breeding and Plant Genetics at the University of Wisconsin-Madison. We thank Ph.D. committee members Dennis Halterman, Shelley Jansky, and Kenneth Sytsma; and also David Baum for advice and review, and the staff of the US Potato Genebank and the C.M. Rick Tomato Genetics Resource Center for germplasm. This research was supported by the USDA and by NSF DEB 0316614 and USDA National Research Initiative Grant 2008-35300-18669 to David Spooner.
- Hijmans R, Gavrilenko T, Stephenson S, Bamberg J, Salas A, Spooner DM: Geographic and environmental range expansion through polyploidy in wild potatoes (Solanum section Petota). Global Ecol Biogeogr. 2007, 16: 485-495. 10.1111/j.1466-8238.2007.00308.x.View ArticleGoogle Scholar
- Marks GE: Cytogenetic studies in tuberous Solanum species: I. Genomic differentiation in the Group Demissa. J Genet. 1955, 53: 262-269. 10.1007/BF02993980.View ArticleGoogle Scholar
- Marks GE: Cytogenetic studies in tuberous Solanum species: III. Species relationships in some South and Central American species. New Phytol. 1965, 64: 293-306. 10.1111/j.1469-8137.1965.tb05398.x.View ArticleGoogle Scholar
- Hawkes JG: Taxonomy, cytology and crossability. Kartoffel. Edited by: Kappert H, Rudorf W. 1958, Berlin: Paul Parey, 2: 1-43. 2Google Scholar
- Hawkes JG: The potato. Evolution, biodiversity and genetic resources. 1990, Washington, DC: Belhaven PressGoogle Scholar
- Irikura Y: Cytogenetic studies on the haploid plants of tuber bearing Solanum species: II. Cytogenetic investigations on haploid plants and interspecific hybrids by utilizing haploidy. (In Japanese, with English summary). Res Bull Hokkaido Nat Agric Exp Station. 1976, 115: 1-80.Google Scholar
- Ramanna MS, Hermsen JGT: Structural hybridity in the series Etuberosa of the genus Solanum and its bearing on crossability. Euphytica. 1981, 30: 15-31. 10.1007/BF00033655.View ArticleGoogle Scholar
- López LE, Hawkes JG: Cytology and genome constitution of the tuber-bearing Solanum species in series Conicibaccata. Solanaceae III, Taxonomy, Chemistry and Evolution. Edited by: Hawkes JG, Lester RN, Nee M, Estrada-R N. 1991, Kew, UK: Royal Botanic Gardens, 327-346.Google Scholar
- Matsubayashi M: Phylogenetic relationships in the potato and its related species. Chromosome engineering in plants: genetics, breeding, evolution, Part B. Edited by: Tsuchiya T, Gupta PK. 1991, Amsterdam: Elsevier Science Publishers BV, 93-118.Google Scholar
- Ramanna MS, Hermsen JGT: Genome relationships in tuber-bearing Solanums. The biology and taxonomy of the Solanaceae. Edited by: Hawkes G, Lester RN, Skelding AD. 1979, Linnean Society of London Ser. 7. London: Academic Press, 647-654.Google Scholar
- Gavrilenko T, Thieme R, Rokka V-M: Cytogenetic analysis of Lycopersicon esculentum (+) Solanum etuberosum somatic hybrids and their androgenetic regenerants. Theor Appl Genet. 2001, 103: 231-239. 10.1007/s001220100626.View ArticleGoogle Scholar
- Wang Y, Diehl A, Wu F, Vrebalov J, Giovannoni J, Siepel A, Tanksley SD: Sequencing and comparative analysis of a conserved syntenic segment in the Solanaceae. Genetics. 2008, 180: 391-408. 10.1534/genetics.108.087981.PubMed CentralView ArticlePubMedGoogle Scholar
- Spooner DM, Salas A: Structure, biosystematics, and genetic resources. Handbook of potato production, improvement, and post-harvest management. Edited by: Gopal J, Khurana SMP. 2006, New York: Haworth's Press, Inc. Binghampton, 1-39.Google Scholar
- Spooner DM, Anderson GJ, Jansen RK: Chloroplast DNA evidence for the interrelationships of tomatoes, potatoes, and pepinos (Solanaceae). Amer J Bot. 1993, 80: 676-688. 10.2307/2445438.View ArticleGoogle Scholar
- Olmstead RG, Palmer JD: Implications for the phylogeny classification, and biogeography of Solanum from cpDNA restriction-site variation. Syst Bot. 1997, 22: 19-29. 10.2307/2419675.View ArticleGoogle Scholar
- Olmstead RG, Sweere JA, Spangler RE, Bohs L, Palmer JD: Phylogeny and provisional classification of the Solanaceae based on chloroplast DNA. Solanaceae IV: Advances in biology and utilization. Edited by: Nee M, Symon DE, Lester RN, Jessop JP. 1999, Kew, U.K.: Royal Botanic Gardens, 111-137.Google Scholar
- Bohs L, Olmstead RG: Solanum phylogeny inferred from chloroplast DNA sequence data. Solanaceae IV: Advances in biology and utilization. Edited by: Nee M, Symon DE, Lester RN, Jessop JP. 1999, Kew, U.K.: Royal Botanic Gardens, 97-110.Google Scholar
- Peralta IE, Spooner DM: GBSSI gene phylogeny of wild tomatoes (Solanum L. section Lycopersicon [Mill.] Wettst. subsection Lycopersicon). Amer J Bot. 2001, 88: 1888-1902. 10.2307/3558365.View ArticleGoogle Scholar
- Spooner DM, Sytsma KJ: Reexamination of series relationships of Mexican and Central American wild potatoes (Solanum sect. Petota): Evidence from chloroplast DNA restriction site variation. Syst Bot. 1992, 17: 432-448. 10.2307/2419483.View ArticleGoogle Scholar
- Rodríguez A, Spooner DM: Chloroplast DNA analysis of Solanum bulbocastanum and S. cardiophyllum, and evidence for the distinctiveness of S. cardiophyllum subsp. ehrenbergii (sect. Petota). Syst Bot. 1997, 22: 31-43. 10.2307/2419676.View ArticleGoogle Scholar
- Spooner DM, Castillo R: Reexamination of series relationships of South American wild potatoes (Solanaceae: Solanum sect. Petota): evidence from chloroplast DNA restriction site variation. Amer J Bot. 1997, 84: 671-685. 10.2307/2445904.View ArticleGoogle Scholar
- Spooner DM, Rodríguez F, Polgár Z, Ballard HE, Jansky SH: Genomic origins of potato polyploids: GBSSI gene sequencing data. Pl Genome, suppl Crop Sci. 2008, 48 (S1): S27-S36.Google Scholar
- Rodríguez F, Spooner DM: Nitrate reductase phylogeny of potato (Solanum sect. Petota) genomes with emphasis on the origins of the polyploid species. Syst Bot. 2009, 34: 207-219.View ArticleGoogle Scholar
- Peralta IE, Spooner DM, Knapp S: The taxonomy of tomatoes: a revision of wild tomatoes (Solanum section Lycopersicon) and their outgroup relatives in sections Juglandifolium and Lycopersicoides. Syst Bot Monogr. 2008, 84: 1-186.Google Scholar
- Rick CM, Laterrot H, Philouze J: A revised key for Lycopersicon species. Tomato Gen Coop Rep. 1990, 40: 31-Google Scholar
- Peralta IE, Spooner DM: Morphological characterization and relationships of wild tomatoes (Solanum L. Section Lycopersicon). Monogr Syst Bot, Missouri Bot Gard. 2005, 104: 227-257.Google Scholar
- Palmer JD, Zamir D: Chloroplast DNA evolution and phylogenetic relationships in Lycopersicon. Proc Natl Acad Sci USA. 1982, 79: 5006-5010. 10.1073/pnas.79.16.5006.PubMed CentralView ArticlePubMedGoogle Scholar
- Miller JC, Tanksley SD: RFLP analysis of phylogenetic relationships and genetic variation in the genus Lycopersicon. Theor Appl Genet. 1990, 80: 437-448.PubMedGoogle Scholar
- Marshall JA, Knapp S, Davey MR, Power JB, Cocking EC, Bennett MD, Cox AV: Molecular systematics of Solanum section Lycopersicum (Lycopersicon) using the nuclear ITS rDNA region. Theor Appl Genet. 2001, 103: 1216-1222. 10.1007/s001220100671.View ArticleGoogle Scholar
- Spooner DM, Peralta IE, Knapp S: Comparison of AFLPs with other markers for phylogenetic inferences in wild tomatoes. [Solanum L. section Lycopersicon (Mill.) Wettst.]. Taxon. 2005, 54: 43-61.View ArticleGoogle Scholar
- Williams DM: Combining trees and combining data. Taxon. 1994, 43: 449-455. 10.2307/1222721.View ArticleGoogle Scholar
- Huelsenbeck JP, Bull JJ, Cunningham : Combining data in phylogenetic analysis. Tree. 1996, 11: 152-158.PubMedGoogle Scholar
- Mason-Gamer RJ, Kellogg EA: Testing for phylogenetic conflict among molecular data sets in the tribe Triticeae (Gramineae). Syst Biol. 1996, 45: 524-545.View ArticleGoogle Scholar
- Farris JS, Källersjö M, Kluge AG, Bult C: Testing significance of congruence. Cladistics. 1994, 10: 315-319. 10.1111/j.1096-0031.1994.tb00181.x.View ArticleGoogle Scholar
- Baum DA: Concordance trees, concordance factors, and the exploration of reticulate genealogy. Taxon. 2007, 56: 417-426.Google Scholar
- Ané C, Larget B, Baum DA, Smith SD, Rokas A: Bayesian estimation of concordance among gene trees. Mol Biol Evol. 2007, 24: 412-426. 10.1093/molbev/msl170.View ArticlePubMedGoogle Scholar
- Wu F, Mueller LA, Crouzillat D, Pétiard V, Tanksley SD: Combining bioinformatics and phylogenetics to identify large sets of single-copy orthologous genes (COSII) for comparative, evolutionary and systematics studies: A test case in the Euasterid plant clade. Genetics. 2006, 174: 1407-1420. 10.1534/genetics.106.062455.PubMed CentralView ArticlePubMedGoogle Scholar
- Gogarten JP, Olendzenski L: Orthologs, paralogs and genome comparison. Curr Opin Gen Develop. 1999, 9: 630-636. 10.1016/S0959-437X(99)00029-5.View ArticleGoogle Scholar
- Sonnhammer ELL, Koonin EV: Orthology, paralogy and proposed classification for paralog subtypes. Trends Gen. 2002, 18: 619-620. 10.1016/S0168-9525(02)02793-2.View ArticleGoogle Scholar
- Rodríguez F: Molecular phylogeny of Solanum Subgenus Potatoe, Sections Lycopersicon and Petota. PhD thesis. 2008, University of Wisconsin; Horticulture Department, Plant Breeding and Plant Genetics Program; Madison, WisconsinGoogle Scholar
- Bull JJ, Huelsenbeck JP, Cunningham CW, Swofford DL, Waddell PJ: Partitioning and combining data in phylogenetic analysis. Syst Biol. 1993, 42: 384-397.View ArticleGoogle Scholar
- Miyamoto MM, Fitch WM: Testing species phylogenies and phylogenetic methods with congruence. Syst Biol. 1995, 44: 64-76.View ArticleGoogle Scholar
- Peralta IE, Knapp S, Spooner DM: New species of wild tomatoes (Solanum section Lycopersicon: Solanaceae) from northern Peru. Syst Bot. 2005, 30: 24-434. 10.1600/0363644054223657.View ArticleGoogle Scholar
- Zou X-H, Zhang F-M, Zhang J-G, Zang L-L, Tang L, Wang J, Sang T: Analysis of 142 genes resolves the rapid diversification of the rice genus. Genome Biol. 2008, 9: R49-10.1186/gb-2008-9-3-r49.PubMed CentralView ArticlePubMedGoogle Scholar
- International Potato Center (CIP): Molecular Biology Laboratory Protocols: Plant Genotyping. Edited by: Ghislain M, Zhang D, Herrera MR. 1999, Crop Improvement and Genetic Resources Department Training Manual. Lima, Peru, SecondGoogle Scholar
- Staden R: The Staden sequence analysis package. Mol Biotech. 1996, 5: 233-241. 10.1007/BF02900361.View ArticleGoogle Scholar
- Thompson JD, Gibson TJ, Plenwniak F, Jeanmougin F, Higgins DG: The ClustalX Windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools. Nuc Acids Res. 1997, 25: 4876-4882. 10.1093/nar/25.24.4876.View ArticleGoogle Scholar
- Maddison DR, Maddison WP: MacClade 4.03: Analysis of phylogeny and character evolution. 2001, Suderland, Massachusetts: SinauerGoogle Scholar
- Simmons MP, Ochoterena H: Gaps as characters in sequence-based phylogenetic analyses. Syst Biol. 2000, 49: 369-381. 10.1093/sysbio/49.2.369.View ArticlePubMedGoogle Scholar
- Müller KF: SeqState – primer design and sequence statistics for phylogenetic DNA data sets. Appl Bioinformatics. 2005, 4: 65-69. 10.2165/00822942-200504010-00008.View ArticlePubMedGoogle Scholar
- Swofford DL: PAUP*: Phylogenetic analysis using parsimony (*and other methods). Version 4.0b3a PPC. 2002, Sunderland, Massachusetts: Sinauer AssociatesGoogle Scholar
- Felsenstein J: Confidence limits on phylogenies: an approach using the bootstrap. Evolution. 1985, 39: 783-791. 10.2307/2408678.View ArticleGoogle Scholar
- Posada D, Crandall KA: MODELTEST: testing the model of DNA substitution. Bioinformatics. 1998, 14: 817-818. 10.1093/bioinformatics/14.9.817.View ArticlePubMedGoogle Scholar
- Stamatakis A: RaxML-VI-HPC: Maximum Likelihood-based phylogenetic analysis with thousands of taxa and mixed models. Bioinformatics. 2006, 22: 2888-2690. 10.1093/bioinformatics/btl446.View ArticleGoogle Scholar
- Yang Z, Rannala B: Bayesian phylogenetic inference using DNA sequences: a Markov Chain Monte Carlo method. Mol Bio Evol. 1997, 14: 717-724.View ArticleGoogle Scholar
- Huelsenbeck JP, Ronquist F: MrBayes: Bayesian inference of phylogenetic trees. Bioinformatics. 2001, 17: 745-755. 10.1093/bioinformatics/17.8.754.View ArticleGoogle Scholar
- Ronquist F, Huelsenbeck JP: MrBayes 3: Bayesian phylogenetic inference under mixed models. Bioinformatics. 2003, 19: 1572-1574. 10.1093/bioinformatics/btg180.View ArticlePubMedGoogle Scholar
- Holm S: A simple sequentially rejective multiple test procedure. Scand J Stat. 1979, 6: 65-70.Google Scholar
- Templeton AR: Phylogenetic inference from restriction endonuclease cleavage site maps with particular reference to the evolution of humans and the apes. Evolution. 1983, 37: 221-244. 10.2307/2408332.View ArticleGoogle Scholar
- Castillo R, Spooner DM: Phylogenetic relationships of wild potatoes, Solanum series Conicibaccata (sect. Petota). Syst Bot. 1997, 22: 45-83. 10.2307/2419677.View ArticleGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.