Introduction

Calophyllum brasiliense (Calophyllaceae) is a native tree species from Central and South America. Its distribution range extends from southern Mexico to northeast Argentina (Reitz and Klein 1978; Rodríguez et al. 2009). It grows exclusively in hygrophyle forests and in lowlands that are almost permanently flooded (Marques et al. 2003). It offers many ecosystem services as a component of riparian forests, and several active compounds of medicinal interest have been recently isolated from different plant tissues. These compounds have inhibitory activity against some tropical pathogens that cause regional and worldwide diseases such as schistosomiasis, Chagas disease and tuberculosis (Gasparotto et al. 2005; Rea et al. 2013; Pires et al. 2014).

The chloroplast genome (cpDNA) is now the most widely used to understand phylogenetic issues among plant families and genera (Shaw et al. 2014). In the past, the use of cpDNA for population analysis was discouraged because of its low mutation rate within species. However, since low-cost sequencing technologies have become routine, many phylogeographic studies in plant species have been based on a variable number of regions (Olmstead and Palmer 1994; Dumoulin-Lapègue et al. 1997; Huang et al. 2004). At first, the few cpDNA regions commonly used for higher taxonomic levels sometimes did not serve the desired purpose, but once the first complete genomes were sequenced, the conservative nature and cpDNA synteny (Palmer 1987; Dumoulin-Lapègue et al. 1997) allowed researchers to test many potential regions. Shaw et al. (2005, 2007, 2014) compared the variability rates of chloroplast regions among distinct clades and found that the cpDNA could be much more informative than earlier thought. Nevertheless, the most suitable chloroplast regions for a population genetic analysis usually will depend on the target species. In this context, the aim of this study was to characterize 8 cpDNA regions among C. brasiliense samples derived from populations from Argentina, Paraguay and Mexico by using 10 previously reported primer pairs and 1 specific primer pair to evaluate their utility in phylogeographic analysis, information that could also help in conservation decisions within the most reduced populations.

Materials and methods

Material and study site

From 2 to 10 individuals from each of the populations in three regions were sampled and stored in Ziploc bags: (1) Misiones (San Ignacio POP—SI) and Corrientes (Puerto Valle POP—PV; Rincón Ombú POP—RO) Argentinean Provinces; (2) Misiones Paraguayan Department (Ayolas POP—AY and Isla Yacyretá POP—IY); and (3) a Mexican sample from the Ecological Reserve Jaguaroundí Park in Veracruz-Llave (Mexico POP—MX). Table 1 summarizes the geographic coordinates and the number of individual collected per site.

Table 1 Location of Calophyllum brasiliense populations sampled
Full size table

DNA extraction

Once leaves were completely dried, total genomic DNA was isolated from these tissues following the protocol of Stange et al. (1998) with modifications made by Percuoco et al. (2014), which involved the addition of antioxidant reagents (2% w/v polyvinylpyrrolidone [PVP], 5 mM ascorbic acid, 4 mM sodium diethyldithiocarbamatetrihydrate [DIECA] and 1.2% β-mercaptoethanol) to the digestion buffer and the extension of incubation steps (digestion step: 0.5 vs. 3 h; precipitation step: 0.5 vs. 1 h).

cpDNA amplification and sequencing

Seven chloroplast intergenic regions were amplified: trnH-psbA, psbI-trnS (ccmp2), trnS-trnG-trnG, psbC-trnS, petA-psbJ, petG-trnP and rpl32-trnL. The eighth region was the trnL intron. Eight heterologous and 2 universal primer pairs obtained from the literature and an intraspecific pair that we designed were used for the PCR reactions (Table 2). The PCR mix composition and cycling profiles are detailed in Tables 3 and 4, respectively. PCR products were purified using GFX PCR DNA #28-9034-70 (GE) and sequenced by Macrogen Inc. using both forward and reverse primers.

Table 2 Details of the primers used for the PCR amplification in Calophyllum brasiliense chloroplast
Full size table
Table 3 PCR mix composition for each of the 8 noncoding cpDNA regions amplified in Calophyllum brasiliense
Full size table
Table 4 PCR profiles for 8 noncoding cpDNA regions amplified in Calophyllum brasiliense
Full size table

Sequence analysis

The quality control of each electropherogram and all sequence editing were done using BioEdit v.7.0.9.0 (Hall, 1999). Consensus sequences were created for each individual, and their identity was verified using BLASTn 2.2.25 (Zheng et al. 2000).

C. brasiliense cpDNA descriptive data

The percentage sequence coverage for C. brasiliense cpDNA was estimated using the genome of Jatropha curcas (NC_012224) as reference. We chose this genome arbitrarily among four available within the order Mapighiales. We also mapped the approximate physical position of the 8 reported regions. Three other chloroplast sequences of C. brasiliense available in GenBank were included in this cpDNA illustration generated with DNA Plotter, free on-line software available at http://www.sanger.ac.uk/science/tools/dnaplotter (Carver et al. 2009).

Results

The 8 regions amplified showed single fragments of approximately the expected size in a variable number of individuals and populations. The editing and analysis of the obtained sequences allowed the characterizing of different noncoding cpDNA regions in C. brasiliense. Three intergenic spacers (see below) and the trnL intron had putative informative sites for C. brasiliense. The length of the sequences obtained and the GenBank accessions for each of the 8 regions analyzed are described in Table 2.

  1. (1)

    trnH-psbA From the 500-bp products separated by agarose gel electrophoresis, a 409-bp consensus sequence was obtained with several microsatellite motifs included in it. These mono- and dinucleotide tandem repeats made the sequencing reaction difficult in some samples, even using microsatellite capillary electrophoretic conditions. The only alignment obtained showed 100% identity between 2 individuals from different Argentinean populations (SI and RO).

  2. (2)

    psbI-trnS (ccmp2) Amplicons of ≈ 200-bp were obtained from samples from 2 Argentinean populations (SI, RO), whereas a smaller band size was observed in the Mexican samples. A 194-bp sequence was edited based on the SI and RO populations. This intergenic fragment corresponded to a chloroplast microsatellite that had an interrupted motif in C. brasiliense. The sequence shared 63% identity with Nicotiana tabacum (Z00044) and maximum identity (73%) with Mallotus paniculatum (AY159390). The sequence could not be submitted to GenBank due to the minimal length of 200-bp required by the database (see http://www.ncbi.nlm.nih.gov/genbank/submit_types).

  3. (3)

    trnS-trnG and trnS-trnG intron-trnG Two primer pairs were tested to amplify this region: (a) Hamilton’s (1999) primers amplified the expected band size of ≈ 600-bp and two extra bands; and (b) The primers of Shaw et al. (2007) that amplified a unique, expected band of ≈ 2000-bp. From the latter, an 1897-bp consensus sequence was edited based on an individual from the Isla Yacyretá population, which shared 86% identity with Jatropha curcas. However, most accessions in the BLAST alignment showed coverage only up to 43% of the query sequence, except when aligned with Licania heteromorpha (KJ414481), another tree species from the family Chrysobalanaceae within the order Malpighiales, with which coverage was 66%. Four more samples (9%) were sequenced and partially aligned, but no polymorphic site was identified.

  4. (4)

    psbC-trnS A unique amplicon of ≈ 1500-bp was amplified in all individuals from SI and RO (22/51%) but failed to be amplified among samples derived from the other populations (21/49%). Once the reliable electropherograms were edited, we obtained a 1514-bp consensus sequence, from which 1307 bp matched the psbC gene 3′ end encoding photosystem II CP43 chlorophyll apoprotein, and only 207-bp corresponded to the noncoding intergenic portion in C. brasiliense. Simultaneously, we carried out an additional digestion assay in this region with MboI restriction endonuclease and compared the profiles after standard agarose electrophoresis. In this assay, 30 individuals of the SI and RO POP were analyzed, showing a unique profile for both populations.

  5. (5)

    trnL intron An amplicon of ≈ 700-bp was obtained in all populations. A first 611-bp consensus sequence was obtained for this region, but difficulties in sequencing of most of the amplicons required us to design a new specific primer pair for the species, which improved the electropherogram quality. The new consensus sequence was 450-bp long, and the alignment of the 43 sequences (100%) showed 4 informative sites, 2 SNPs, a 2-bp inversion and 1 indel 7-bp long. Three haplotypes were then identified. SSRs previously reported for this intron in other species (Hale et al. 2004) were interrupted and monomorphic in C. brasiliense.

  6. (6)

    petA-psbJ Two primer pairs were assayed to amplify this region. Both pairs amplified the expected band of ≈ 1200-bp, but the primers designed by Huang et al. (2004) yielded no amplicon in several individuals (7/18%); thus, we chose the primers published by Shaw et al. (2007). The primer petA yielded overlapped products and low signal after 140-bp in all individuals analyzed, although the reverse primer psbJ allowed us to obtain 1200-bp readings. Once editing was completed, sequences from 1137 to 1174-bp long were aligned. Two indels, a substitution, and a (T) n microsatellite allowed discriminating 2 haplotypes exclusive to the southernmost populations and Mexican populations respectively.

  7. (7)

    petG-trnP An amplification product of ≈ 500-bp was obtained from all samples; 19 (44%) of these were sequenced and resulted in a 449-bp consensus sequence. An 11-bp inversion was the only polymorphism found, which discriminated 2 monohaplotypic clusters, 1 from the Mexican population samples and the others from the other 5 populations.

  8. (8)

    rpl32-trnL Bands of ≈ 1000-bp were observed in all populations. Once the sequences were edited, we obtained a fragment in the range of 811–890-bp. The variation in length was partially due to the inferior quality of the consensus sequence obtained and to a variable number of the (AAATT) n motifs that differentiated the Argentinean and Paraguayan from the Mexican populations.

Three other partial C. brasiliense coding sequences for cpDNA are available in GenBank. These correspond to ndhF (HQ331853), matK (HQ331550) and 8 accessions of rbcL (JQ591092–JQ591094; KC493366–KC493368; KC570911; KF981208). The alignment including all 8 rbcL sequences showed complete identity among them.

The 8 sequences reported here totaled 7149-bp, which increased to 9529 bp when ndhF, matK and rbcL (JQ591094) available sequences were added. Considering that the Jatropha curcas genome is 163,856-bp, approximately 6% of C. brasiliense cpDNA is now accessible, of which ≈ 4.5% corresponds to noncoding regions. The 11 regions (our 8 regions plus the 3 previous deposited in GenBank) now available for C. brasiliense were mapped over the reference genome (Fig. 1).

Fig. 1
figure 1

Sequences of C. brasiliense cpDNA herein described mapped over the Jatropha curcas genome (NC_012224) used as reference

Full size image

Discussion

All 8 regions analyzed in this work have been previously used in studies of several tree species (Dumoulin-Lapègue et al. 1997; Huang et al. 2004; Ramos et al. 2007), showing different informative rates. Eight heterologous, 2 universal and 1 specific primer pair developed in this work allowed amplicons to be obtained from many of the C. brasiliense samples analyzed. Three intergenic regions (petA-psbJ, petG-trnP, rpl32-trnL) and the trnL intron were informative for the species. Particularly, two indels, a substitution and a mononucleotide cpSSR in petA-psbJ, an 11-bp inversion in petG-trnP and a 5-bp repeat motif within trnL-rpl32 were the polymorphisms that allowed 2 haplotypes to be identified, 1 common to the southernmost C. brasiliense populations and the other restricted to the samples derived from Mexico. We can thus affirm that they are suitable chloroplast noncoding regions for population genetic approaches in C. brasiliense.

In the case of the trnL intron a 7-bp indel, one 2-bp inversion and 2 transversions were analyzed. These discriminated 3 haplotypes among C. brasiliense populations, 1 common haplotype distributed in Argentina and Paraguay, and 2 other haplotypes, exclusive to Paraguayan and Mexican populations respectively. Therefore, this region was used in a phylogeographic analysis (Percuoco et al. 2015). The trnL intron was used for several phylogenetic analyses (Hale et al. 2002; Hollingsworth et al. 2011) although some authors have discarded it for intraspecific studies due to its low resolution at low taxonomic levels (Shaw et al. 2005). This region was also considered as a putative barcode region in plants, because of its high conservation rate within species and the robustness of the amplification process (Taberlet et al. 2007).

The primers used to amplify the psbC-trnS intergenic spacer in C. brasiliense showed that this region contains a long psbC fragment, while the spacer is very short for the species. Although we did not sequence the fragment in many samples, the enzymatic digestion revealed no difference among the C. brasiliense samples, in accordance with the expected low level of variation because of the selection pressure exerted over coding regions. However, Ramos et al. (2007) used psbC-trnS in an intraspecific study of 17 populations of Hymenaea stigonocarpa from Brazil and found 23 haplotypes based on 13 substitutions and 4 indels in the alignment of a 524-bp fragment. Unfortunately, these sequences were not deposited in GenBank, and the electropherograms were not included as supplementary material for comparison with our results to enrich the present analysis.

The trnH-psbA intergenic spacer showed low-quality electropherograms due to several polyA/T repeats. There is, however, only one published phylogeographic analysis of C. brasiliense populations from Brazil where the use of this intergenic spacer seems to have been successful (Salgueiro et al. 2011). Although these authors used the same primers that we tested in this work, they reported a 263-bp sequence, a fragment much shorter than the one we obtained (409 bp). Surprisingly, they identified 28 informative sites within the 263 bp of the trnH-psbA intergenic spacer based on 192 individuals of C. brasiliense; however, again, the electropherograms are not available to compare with our data.

The high number of repeats observed in the trnH-psbA region of C. brasiliense is in accordance with the common mono- and dinucleotide SSRs usually found in cpDNA, but these informative polymorphisms could easily arise as parallelism or homoplasies (Petit and Vendramin 2007), leading to wrong phylogeographic conclusions. Therefore, this type of mutation is often discarded in phylogeographic analysis (Huang et al. 2004; Lihová et al. 2010).

The trnH-psbA spacer has also been proposed as a potential DNA barcoding region (Kress et al. 2005; Chase et al. 2007), its high variability in length across larger angiosperm families and the ambiguity added by the obligatory manual editing because of multiple cpSSRs, as those observed in C. brasiliense, make it a problematic if it were the standard plant barcode alone (Chase et al. 2007).

Primer mismatches, a frequent problem when heterologous or universal primers are used, likely were responsible for the difficulties in the sequencing reaction for the trnH-psbA, psbI-trnS, trnS-trnG-trnG and psbC-trnS intergenic spacers amplified in C. brasiliense.

The chloroplast genome remains the most appropriate target in plant phylogeography studies due to its uniparental inheritance and its haploid nature. In this sense, cpDNA “records” historical fluctuations in populations for each species with high fidelity, providing a historical glance to improve conservation decisions. Although many cpDNAs have been completely sequenced (Malé et al. 2014), various nonmodal tropical tree species remain to be studied. Moreover, universal primers are frequently tested in several species but only in a few individuals within each species; therefore, the intraspecific performance and features of the primers are unknown. Particularly for our native tree, all cpDNA regions were amplified with diverse levels of difficulty in the subsequent analysis of the sequences obtained. We improved our results when we designed a pair of primers for C. brasiliense as we showed for the trnL intron. Thus, we strongly recommend designing specific primer pairs to ensure the quality of the data even for those regions difficult to analyze. Finally, using more than one cpDNA region increases the number of informative sites and takes advantage of genome fitness. Thus, the 8 regions described in this work contribute to the knowledge of C. brasiliense cpDNA, but based on our results, we suggest petA-psbJ, petG-trnP, rpl32-trnL and the trnL intron analyzed in this work as suitable noncoding regions that could be useful for phylogeographic studies of this species.

Conclusions

All 8 chloroplastic regions were amplified, and a sequence of 7149-bp, which represents approximately 6% of C. brasiliense cpDNA is now available in GenBank. Three intergenic regions (petA-psbJ, petG-trnP, rpl32-trnL) and the trnL intron were informative and will be useful for population analyses of C. brasiliense.