RAPD and Internal Transcribed Spacer Sequence Analyses Reveal Zea nicaraguensis as a Section Luxuriantes Species Close to Zea luxurians

Genetic relationship of a newly discovered teosinte from Nicaragua, Zea nicaraguensis with waterlogging tolerance, was determined based on randomly amplified polymorphic DNA (RAPD) markers and the internal transcribed spacer (ITS) sequences of nuclear ribosomal DNA using 14 accessions from Zea species. RAPD analysis showed that a total of 5,303 fragments were produced by 136 random decamer primers, of which 84.86% bands were polymorphic. RAPD-based UPGMA analysis demonstrated that the genus Zea can be divided into section Luxuriantes including Zea diploperennis, Zea luxurians, Zea perennis and Zea nicaraguensis, and section Zea including Zea mays ssp. mexicana, Zea mays ssp. parviglumis, Zea mays ssp. huehuetenangensis and Zea mays ssp. mays. ITS sequence analysis showed the lengths of the entire ITS region of the 14 taxa in Zea varied from 597 to 605 bp. The average GC content was 67.8%. In addition to the insertion/deletions, 78 variable sites were recorded in the total ITS region with 47 in ITS1, 5 in 5.8S, and 26 in ITS2. Sequences of these taxa were analyzed with neighbor-joining (NJ) and maximum parsimony (MP) methods to construct the phylogenetic trees, selecting Tripsacum dactyloides L. as the outgroup. The phylogenetic relationships of Zea species inferred from the ITS sequences are highly concordant with the RAPD evidence that resolved two major subgenus clades. Both RAPD and ITS sequence analyses indicate that Zea nicaraguensis is more closely related to Zea luxurians than the other teosintes and cultivated maize, which should be regarded as a section Luxuriantes species.


Introduction
The genus Zea has been classified into two sections [1,2], section Luxuriantes, which is composed of Zea diploperennis, Zea luxurians and Zea perennis, and section Zea, which contains four subspecies: Zea mays ssp. mays, Zea mays ssp. mexicana, Zea mays ssp. parviglumis and Zea mays ssp. huehuetenangensis [3]. In genus Zea, both wild taxa that have the common name ''teosinte'' and cultivated maize are diploid (n = 10) with the exception of tetraploid Z. perennis (n = 20). As the closest wild relative of maize, teosinte, which is indigenous to Mexico and Central America [4], is a potentially important resource for the study of maize genetics and evolution and for plant breeding.
A new teosinte recently discovered from Pacific Coastal Nicaragua, named Zea nicaraguensis, occurs at 6-15 m above sea level, a very low elevation for teosinte, and has the unusual ability to grow in as much as 0.4 m of standing or slowly moving water [5]. Thriving in flooded conditions, this fresh teosinte has a high capacity to form root aerenchyma and adventitious roots. Z. nicaraguensis, noteworthy for its tolerance to a waterlogged environment and for its stable expression of a reproductive pathway [6], could be a useful source of germplasm for maize breeding, specifically, for breeding flooding-tolerant maize accessions through wide cross [5,7]. The closest relative to Nicaraguan teosinte, Z. nicaraguensis, is probably Z. luxurians of southeastern Guatemala [5]. Z. nicaraguensis and Z. luxurians show close morphological resemblance, and also display considerable differences in developmental behaviour, supporting a taxonomic segregation [8]. Nicaraguan teosinte has much longer and more abundant tassel branches, a larger number of spikelets per branch, and longer, more visibly transversely rugose outer glumers, as well as a habitat different from its Guatemalan counterpart [5]. However, the actual genetic relationship of Z. nicaraguensis in Zea species is uncertain until now. Iltis and Benz [5] considered Z. nicaraguensis as a new species based on differences in ecology and tassel and plant morphology. The chromosome number of Z. nicaraguensis is 2n = 20, and the C-banding pattern shows that Z. nicaraguensis is very similar to Z. luxurians and more similar to Z. luxurians than to Z. diploperennis and cultivated maize [8]. Nevertheless, the genetic relationship of Z. nicaraguensis in Zea species should be further investigated systematically.
Recently-developed molecular genetic techniques have provided another opportunity to assess the degree of genetic relatedness between maize and teosinte. RAPD marker has proven quite useful in genetic study of many plant species [9]. This marker system has the ability to amplify DNA from dispersed polymorphic loci and has its power to detect small genetic differences [10]. To our knowledge, such molecular marker is only an indirect DNA sequence analysis technique. To infer a more accurate conclusion on genetic relationships among species, it is necessary to combine with direct DNA sequence analysis techniques. Previous studies on the internal transcribed spacer (ITS) region of the rDNA in nuclear genome showed promising results for the phylogenetic study of grasses [11,12]. We have chosen ITS as a phylogenetic marker, because it is appropriate for investigating species-level relationships and it is a nuclear marker that can be useful for detecting reticulate phenomena [13]. ITS sequences evolve rapidly, but size and functional constraints permit comparison of homologous sequences between taxa in genus or subgenus [14].
In this report, we present the first molecular marker analyses for the genetic relationship of Z. nicaraguensis in Zea species by using randomly amplified polymorphic DNA (RAPD) markers and the internal transcribed spacer (ITS) sequences of nuclear ribosomal DNA (nrDNA), which have both been used to resolve relationships among closely related taxa [15], and a sample of 14 maize and teosinte species. Comparisons between clustering trees inferred from both indirect and direct DNA sequence analysis techniques would provide a better assessment of the true species relationships.

RAPD band polymorphism
A total of 4500 polymorphisms out of 5303 repeatable products ranging from 500 to 2000 bp were obtained from 136 decamer primers screened from 340 primers among maize and teosinte species (Figure 1). The ratio of polymorphism was 84.86% and each primer generated an average of 38.99 bands and 33.09 polymorphic bands, indicating extensive genetic diversity existing in Zea species.

Genetic similarities among Zea species by RAPD analysis
The RAPD data were used to estimate the genetic similarities and the phylogenetic relationships among the 14 Zea species genotypes. A similarity coefficient matrix showed the genetic distance among all Zea species in the present study ( Table 1). The similarity values ranged from 0.585 to 0.809 with an average similarity value 0.671. For the same species from different geographical regions compared, the highest similarity index (0.809) was observed between the Tz01 and Tz02 genotypes, followed by the similarity index between Tz05 and Tz06 (0.796). For the different species compared, the highest similarity index (0.745) was observed between the Tz04 and Tz07 species, and the lowest one (0.585) was recorded between Tz01 and Tz11. A high degree of genetic similarity was revealed among the same species from different geographical regions. For Z. nicaraguensis, the similarity index between Tz10 and Tz06 was the highest (0.736), and the one between Tz10 and Tz02 was the lowest (0.593), indicating the close relationship between Z. nicaraguensis and Z. luxurians.

RAPD cluster analysis
The similarity matrix was then used to construct a dendrogram which showed two major clusters within 14 species of Zea at the level of 0.619 with the UPGMA (unweighted pair group method with arithmetic average) method ( Figure 2). The first major cluster included Z. diploperennis (Tz09), Z. perennis (Tz01 and Tz02), Z. luxurians (Tz05 and Tz06) and Z. nicaraguensis (Tz10). The second major cluster consisted of Z. mays ssp. parviglumis (Tz03 and Tz04), Z. mays ssp. mexicana (Tz07 and Tz08), Z. mays ssp. huehuetenangensis (Tz11) and Z. mays ssp. mays (Tz12,Tz13 and Tz14). The results were generally consistent with the opinion that the genus Zea was divided into sections Luxuriantes and Zea according to Dobeley et al. [1] and Iltis et al. [2]. The new teosinte Z. nicaraguensis was classified into the first major cluster and thus considered to be a section Luxuriantes species. It formed a subgroup with Z. luxurians, indicating their closest relationship.
Within the two major clusters, the section Luxuriantes clade indicated that Z. diploperennis and Z. perennis formed a subgroup at the level of 0.626, which contained two genotypes only belonging to the perennis tesionte species. The section Zea was divided into three subgroups at the level of 0.686, with the first subgroup including Z. mays ssp. parviglumis and Z. mays ssp. mexicana, the second subgroup consisting of three cultivated maize inbreds, and the third subgroup only containing Z. mays ssp. huehuetenangensis.

ITS sequence variation
The PCR-amplified DNA fragments of all samples showed a clean single band product when examined on an agarose gel ( Figure 3).
The boundaries of ITS1, 5.8S and ITS2 regions were identified by comparison with published ITS sequences of maize and teosinte available from GenBank referring to Buckler and Holtsford [16]. The ITS sequences in the present study were highly homologous (95-99%) with Zea ITS sequences reported by Buckler et al. [16], demonstrating that our sequences are accurate and reliable. Characteristics of the ITS1, 5.8S and ITS2 sequences of Zea species were summarized in Table 2. The lengths of the entire ITS regions of the 14 taxa analyzed varied from 597 to 605 bp. The ITS1 sequences which varied in size from 211 to 218 bp were only slightly shorter than ITS2 (221-229 bp), and the 5.8S gene was uniformly 164 bp in length. The average GC content of the total ITS region was 67.8%; ITS1, 5.8S, and ITS2 regions had average GC contents of 70.4%, 56.6%, and 73.5%, respectively. All of the Zea alleles had similar base composition. When the outgroup taxa (Tripsacum dactyloides L.) was taken into account, the alignment of the entire ITS sequences resulted in 614 characters, of which 10 positions (of 78 variable sites) were potentially phylogenetically informative sites. Five sites (of 47 variable sites) were informative in ITS1, 5 (of 26 variable sites) in ITS2, and no informative site was recorded in the 5.8S gene which contained 5 variable sites ( Table 2).
For the combined data set (ITS1, 5.8S, ITS2), pairwise nucleotide sequence divergence based on Kimura two-parameter distance method [17] ranged from 0.17 to 2.92% among taxa of Zea, with a mean value of 1.29%; between species of Zea and the outgroup taxa, it ranged from 0.17 to 9.54% ( Table 3). The divergence values from pairwise comparisons between Tz03 and Tz04, Tz06 and Tz10, and Tz05 and Tz10 genotypes were the lowest, and the greatest divergence was found between Tz02 and Tz13. The sequence divergence between Z. nicaraguensis and the other taxa in Zea ranged from 0.17 to 2.20%, among which the distances between Tz10 and Tz05, and Tz10 and Tz06 were the lowest, confirming that Z. nicaraguensis is more similar to Z. luxurians than to the rest wild and domesticated species.
Gaps due to insertion/deletion events were introduced to align the sequences of the ITS1 and ITS2 regions. The largest gap Table 1. Similarity coefficients of 14 accessions/species of genus Zea by using RAPD primers.

Tz01
Tz02 Tz03  Tz04  Tz05  Tz06  Tz07  Tz08  Tz09  Tz10  Tz11  Tz12 Tz13 Tz14  (9 bp) was in the ITS2 region of the T. dactyloides sequence. Hsiao et al. [12] indicated gaps were correlated with particular species groups and were potentially phylogenetically informative. Baldwin [18] also suggested that length mutation in the ITS region can be of potential value for phylogeny reconstruction.
In addition to the insertion/deletions, the most common polymorphisms in the ITS region were base substitutions. The numbers of transitions/transversions by pairwise comparisons of nucleotide substitutions of 14 taxa from Zea species and the outgroup, T. dactyloides, were shown in Table 4. There were only a few base pair differences between Z. luxurians and Z. nicaraguensis. Z. luxurians (Tz06) and Z. nicaraguensis (Tz10) had only 1-bp transition, and Z. luxurians (Tz05) and Tz10 had only 1-bp transversion, indicating their identical ITS sequences. The similarity in ITS sequences between the two species was expected since analysis of C-banding patterns [8] have suggested that they are apparently closely related.

Phylogenetic inference based on ITS data
Treating gaps as missing data, the phylogenetic trees of 14 taxa in Zea based on ITS sequence variation were constructed, and rooted by Tripsacum species (Trip). As shown in the NJ (neighbor-joining) tree (Figure 4), 14 samples of Zea were also grouped into two major clusters similar to the result of RAPD analysis, among which the taxa from section Zea composed a cluster supported with a high bootstrap support (BS) of 92%, and the taxa from section Luxuriantes including Z. nicaraguensis were all members of the other cluster (BS = 71%). The topology of the MP (maximum parsimony) tree was almost identical to that of the NJ tree, except for a few minor differences at the bootstrap values of the branches (results not shown). Several subclades appeared further within the two major clusters in the NJ tree. The subspecies of Z. mays clade formed sister groups with bootstrap support of about 70% which were distinct from the outgroup. Z. diploperennis, together with the two Z. perennis species, were in the moderately supported group (BS = 67%), which was a sister group to Z. luxurians and Z. nicaraguensis. From the ITS phylogenetic tree (Figure 4), it is apparently reconfirmed that Z. nicaraguensis located in section Luxuriantes clade had the closest association with Z. luxurians.
As defined in Figure 5, the phylogenetic tree reconstructed combining our ITS sequences with data from GenBank (Table 5) was rooted with Tripsacum species (including Trip and tripd1tripd 4). Our ITS sequences, such as Z. luxurians (Tz05 and Tz06), diploperennis (Tz09), Z. perennis (Tz01 and Tz02), Z. mays ssp. huehuetenangensis (Tz11), Z. mays ssp. mays (Tz12, Tz13 and Tz14), Z. mays ssp. parviglumis (Tz03 and Tz04) and Z. mays ssp. mexicana (Tz08), were grouped completely consistently with the corresponding taxa on the GenBank database. But Z. mays ssp. mexicana (Tz07) gathered with GenBank sequences from Z. mays ssp. mays (zmays); GenBank sequences from Z. mays ssp. mexicana (zmm) and Z. mays ssp. parviglumis (zmp) didn't form stable monophyletic clades, few of which mingled with Z. mays ssp. mays (zmays). These may be due to introgression between cultivated maize and the annual teosinte (Z. mays ssp. mexicana or Z. mays ssp. parviglumis). The result indicates that the ITS sequences obtained in our laboratory possess high value, phylogenetic analysis of which can reflect genetic relationships of maize and teosinte in Zea.

RAPD-based UPGMA analysis
The subgeneric boundaries of Zea proposed by Doebley and Iltis [1,2] are well defined on UPGMA-based dendrogram derived from RAPD analysis (Figure 2). The members of each section share a suite of morphological and genetic features [3]. Interestingly, the grouping of populations representing the Zea species in the subgenus interior clades is more congruent with grow habitats as well as their geographical locations than morphological data according to Wilkes [19], a little bit different with the system of classification proposed by Doebley and Iltis [1,2]. In our study, there are eight recognized species of teosinte Z. diploperennis, Z. perennis, Z. luxurians, Z. nicaraguensis, Z. mays ssp.
mexicana, Z. mays ssp. parviglumis, Z. mays ssp. huehuetenangensis and cultivated maize, which can be basically divided into two sections, annual species and perennial species, based on RAPD cluster analysis. The annual species have been then classified as annual teosinte and cultivated maize. The annual teosintes were then further divided into two parts, one part including Z. mays ssp. mexicana, Z. mays ssp. parviglumis, and Z. mays ssp. huehuetenangensis, and the other including Z. luxurians and Z. nicaraguensis. For the perennial section, it has been classified as Z. perennis (2n = 40) and Z. diploperennis (2n = 20). In 1967, Wilkes had presented a system of classification for teosinte which furnished the different geographic populations with separate racial designations [3,19]. As treated by Wilkes [19], Zea is divided into two sections: section Eucblaena comprising Zea perennis and Zea mexicana which includes races Table 3. The distance matrix based on Kimura two-parameter distance method.

ITS variation
The high GC content and secondary structure were presented in this study for the ITS region of Zea. We used high success-rate DNA polymerase KOD FX to circumvent these PCR amplification problems. KOD FX is effective for the amplification of GCrich targets and crude samples, and the PCR error ratio is about 10 times less than that of Taq DNA polymerase. Considering the conditions used here, the error rate should be less than 0.02 bases per ITS clone [16]. Therefore, the nucleotide variation in Zea ribosomal ITS sequences was hardly a PCR artifact.
The level of genetic variation observed in our ITS data set (0.17-2.92% sequence divergence among taxa of Zea, only 10 informative sites) was compared with ITS variation in other angiosperms studied. ITS sequence divergence ranged from 0.4 to 15.6% and 247 sites were informative within the alpine Saxifraga, sect. Ligulatae (Saxifragaceae) [20], and the corresponding values were 0.5-17% and 238 in Iris series Californicae (Iridaceae) of the North American Pacific Coast [21]. It was shown that our intrageneric variation of ITS sequences was much lower than those in these angiosperms. However, the ITS variation value and the number of informative positions in the present data set were comparable to those found in some other angiosperms. For example, 0-1.48% sequence divergence and 10 informative characters were recorded in the Oxytropis campestris and O. arctica (Fabaceae) complexes in Alaska [15]; pairwise nucleotide sequence differences within the European montane/alpine endemic Soldanella (Primulaceae) ranged from 0 to 2.1%, with a mean value of 1.0% [22]; only 9 informative characters were detected in the circumboreal species Saxifraga oppositifolia and some related taxa [23].
ITS phylogeny of the Zea ITS phylogeny depicted two major groups of the genus Zea. These two groups are distinct and consistent with the RAPD tree, which are supported by classification of Zea according to Doebley and Iltis [1,2]. The phylogenetic tree of Zea (Figure 4) inferred from the ITS sequences shows that each subgenus is monophyletic. When Tripsacum was used as an outgroup, the taxa in section Luxuriantes were consistently placed as the sister groups, and albeit with not very strong support, they were close to each other in space position at a major cluster located on the base of the phylogenetic tree and clearly differentiated from the monophyletic Z. mays clade. The moderate bootstrap values for interior clades suggest that there is abundant sequence polymorphism or introgression among species or subspecies in Zea [16]. Figure 5 indicates that our ITS data largely correspond to the previously published ITS sequences of Zea, although there are a few samples that are not always well grouped. Our ITS data favor previous phylogenies based on chloroplast and isozyme evidence, which significantly supported a Z. luxurians, Z. diploperennis and Z. perennis clade [3,24,25] and a Z. mays clade. Z. perennis and Z. diploperennis are essentially indistinguishable, supported by previous reports from rDNA sequences [16] and microsatellites [26], probably reflecting a recent divergence. The ribosomal ITS relationships between Z. mays. ssp. parviglumis, Z. mays. ssp. mexicana, Z. mays. ssp. huehuetenangensis and Z. mays. ssp. mays are complicated by the recent divergence of these taxa [16]. The present result is corroborated by microsatellite evidence [26] of admixture between Z. mays ssp. mays and the Mexican annual teosintes (ssp. mexicana and ssp. parviglumis) ( Figure 5); however, it may also be a function of the recent divergence between these taxa such that their gene pools are not yet completely differentiated. A likely explanation is that maize and annual teosinte are very close geographically and cytogenetically, and except for rare chromosomal rearrangements, their basic genomes are essentially homologous [27]. Another explanation is probably genetic exchange between conventional maize landraces and teosinte owing to crossability among them [28][29][30]. Fukunaga et al. [26] identified 56 out of 117 teosinte plants containing 20% or more of maize germplasm. In their opinion, teosinte grows near maize in most locations and is capable of hybridizing with maize, allowing admixture between maize and the teosintes to occur [26]. Gene flow among maize (Zea mays ssp. mays) and teosinte (Zea spp.) populations has occurred readily since maize's domestication 9,000 years ago [31][32][33]. In spite of crossability between maize and teosinte, more studies suggest that gene flow is mainly unidirectional from teosintes to maize, with either insignificant introgression from maize to teosintes or none at all in either direction [25,34]. For example, the Teosinte crossing barrier-1 (Tcb1-s) or Gametophyte-1 (Ga1) incompatibility allele in teosinte can prevent crossing in the direction of maize to teosinte [35,36]. However, gene flow or introgression among different teosinte types has not been studied thoroughly until now, and the genetic affinities between maize and teosinte need to be better understood. Furthermore, the phylogenetic position of Z. nicaraguensis, a newly discovered teosinte, in Zea species should be further investigated systematically. The results of the present study indicate that it has been clearly defined as being a taxon of section Luxuriantes (Figures 4-5).

Relationship of Z. nicaraguensis in Zea
As expected, both RAPD cluster and ITS sequence analyses showed that the new teosinte species found on the Pacific cost of Nicaragua, Z. nicaraguensis, was clearly classified into section Luxuriantes, which was most closely allied to Z. luxurians. Z. luxurians, an annual with similar, strongly two-nerved lower glumes, is known mostly from a restricted region of southeastern Guatemala, although it has outlier populations in Honduras and Nicaragua [5]. Both Z. nicaraguensis and Z. luxurians possess unique flooding-related traits such as the capacity to form root aerenchyma under non-flooding conditions and the ability to form adventitious roots at the soil surface under flooding conditions [37]. SSR data confirmed that Z. nicaraguensis was not strongly differentiated from Z. luxurians [26]. The similarity in Cbanding pattern most likely reflected a closer relationship between Z. nicaraguensis and Z. luxurians [8]. This has been confirmed by the highest RAPD genetic similarity and the lowest pairwise nucleotide sequence divergence between the two species as revealed in this study. Whether or not Z. nicaraguensis and Z. luxurians should be regarded as subspecies or separate species still needs to be considered with the cytological meiosis evidence. Fukunaga et al. [26] proposed that the status of Z. nicaraguensis should be investigated by determining its cross compatibility with Z. luxurians. If they are interfertile, then it would be best to treat Z. nicaraguensis as a subspecies of Z. luxurians. Cultivated maize (Z. mays ssp. mays) has mostly subterminal knobs and only a few terminal ones, which makes it different from the wild teosinte species in section Luxuriantes. All chromosomes of Z. nicaraguensis except for  chromosome 10 have terminal knobs [8], demonstrating that Z. nicaraguensis is more similar to section Luxuriantes species than to maize. However, no published reports of natural or artificial hybridization between the other teosintes and the Nicaraguan teosinte Z. nicaraguensis are known.

Conclusions
In conclusion, the genetic relationships among Zea species were reported in many previous studies on the basis of morphological, cytogenetical, chemical, and molecular data [4]. The present work provided the first molecular evidence for systematic assessment of genetic relationship of Z. nicaraguensis in Zea species. The results of the analyses using two types of genetic markers, RAPD and ITS, indicate that Z. nicaraguensis has been clearly defined as being the taxon of section Luxuriantes. The Guatemalan teosinte Z. luxurians is the closest relative to the Nicaraguan teosinte Z. nicaraguensis.

Plant materials and DNA extraction
The plant materials used in the present study are listed in Table 6, which were obtained from the International Maize and Wheat Improvement Center (CIMMYT) and United States Department of Agriculture (USDA). Tripsacum dactyloides L. (2n = 72, sample No. was Trip) from USA was selected as the outgroup for ITS sequence analysis according to the phylogenetics of Zea and Tripsacum reported by Buckler and Holtsford [16]. Total DNA was extracted from young leaves of each of the genotypes using a CTAB method (modified after Saghai-Maroof et al. [38]).

RAPD and ITS sequences amplifications
RAPD reactions were performed using 136 random decamer primers with stable and readily reproducible banding patterns screened from a total of 340 primers purchased from Beijing SBS Genetech Co., Ltd (for sequences of the 136 primers, see Table  S1). Amplification conditions were 25 mL reactions with ddH 2 O 16.8 mL, 106PCR buffer 2.5 mL, MgC1 2 (25 mmol/L) 1.5 mL, dNTPs (10 mmol/L) 2.0 mL, primer (10 pmo1/mL) 1 mL, Taq polymerase (5 U/mL) 0.2 mL, and genomic DNA (50 ng/mL) 1 mL. The PCR parameters were 94uC for 5 min, 38 cycles of 94uC for 1 min, 38uC for 1 min, and 72uC for 1.5 min, followed by a final extension step at 72uC for 7 min. Amplification products were separated by size on 1.5% agarose gels, stained with ethidium bromide (EB), and visualized under UV (ultraviolet) light. Pictures were taken with an Electrophoresis Systems Photo Documentation Camera.
The ITS region (including ITS1, 5.8S and ITS2) of each sample was amplified with primers P1: 59-TCGTAACAAGGTTTCCG-TAGG-39 and P4: 59-TCCTCCGCTTATTGATATGC-39 [39], synthesized by Shanghai Invitrogen Biotechnology Co., Ltd. The DNA fragment amplified using these two primers is approximately 700 bp long. Amplification conditions were 25 mL reactions by adding 12.5 mL 26PCR buffer for KOD FX, 4.5 mL ddH 2 O, 5.0 mL dNTPs (2 mmol/L), 0.75 mL P1 (10 pmo1/mL), 0.75 mL P4 (10 pmo1/mL), 0.5 mL KOD FX (1 U/mL) (Toyobo Co., Ltd.), and 1 mL genomic DNA (50 ng/mL). The PCR parameters were 94uC for 4 min, 38 cycles of 94uC for 1 min, 61uC for 1 min, and 72uC for 1.5 min, followed by a final extension step at 72uC for 8 min. The PCR products were separated on 1.5% agarose gels and visualized with EB. Due to the smooth terminal of target bands amplified by using high fidelity PCR enzyme KOD FX, it was necessary to add ''A'' tail to the terminal of the PCR products before TA cloning. Then PCR products were purified with TIANgel Midi Purification Kit (TIANGEN, Beijing, China) following the protocol from the manufacturer, and ligated into a pMD19-T vector (TaKaRa, Dalian, China). Both gel-purified PCR products and cloned inserts which were confirmed in the range of about 700 bp were used as templates for sequencing. Difficulties with obtaining sequence information from PCRamplified products from some samples made it necessary to derive the complete sequence from cloned templates. In this case, at least two different clones were sequenced to minimize the impact of errors caused by PCR amplification. Sequencing was conducted on an ABI 3730 automated sequencer (Invitrogen, Shanghai, China).

Analysis of RAPD and ITS data
Photographs were used to score the data for RAPD analysis, and DNA fragment sizes were estimated by comparisons with DNA size markers run on the same gel. RAPD bands were scored as present (1) or absent (0) in each genotype for each set of primers to create binary data matrices and only strong, clear and reproducible bands were used in this study. The data matrices were entered into the NTSYS-pc2.10 software and pairwise distance matrices were computed based on Dice similarity coefficients using the SIMQUAL (similarity for qualitative data) routine [40]. Similarity coefficients were used to construct the UPGMA (unweighted pair group method with arithmetic average) dendrograms [41] using the SHAN (sequential, hierarchical, agglomerative, and nested cluster methods) clustering and a tree was displayed through the tree plot routine in the software. Sequences of the entire ITS region from 14 Zea species together with that of the outgroup species were used to determine ITS1, 5.8S and ITS2 boundaries by homologous blast with published ITS sequences on the GenBank database, and were subsequently aligned with the Clustal X program [42]. Then, phylogenetic analyses of the aligned ITS sequences (the entire ITS1-5.8S-ITS2 region) was performed in MEGA 4.0 (Molecular Evolutionary Genetics Analysis, version 4.0) to calculate the characteristics of the ITS regions and to construct the phylogenetic trees employing neighbor-joining (NJ) and maximum parsimony (MP) methods. Calculations were made after considering gaps as missing data [22]. A preliminary analysis including all sequenced clones and PCR products was conducted. If all the clones from one sample were supported as monophyletic or as components of the same polytomy, a representative sequence showing the fewest autapomorphies was chosen for all further analyses. MEGA 4.0 was also used to calculate the proportion of nucleotide sequence differences (adjusted for missing data and gaps), and Kimura two-parameter distances as the measure of sequence divergence. In addition, bootstrap analysis (1000 replications) was performed to obtain estimates of support for clades of the ITS trees. To verify our ITS data further, GenBank data were used in sequence analysis together with the ITS sequences determined in our laboratory, all of which came from studies by Buckler and Holtsford [16] ( Table 5). Four GenBank sequences from species within Tripsacum (Tripsacum dactyloides) were used as the outgroup for phylogenetic analysis.