- Record: found
- Abstract: found
- Article: found
Deficiency of a triterpene pathway results in humidity-sensitive genic male sterility in rice
research-article
Author(s):
Zheyong Xue
1 ,
Xia Xu
1
,
2 ,
Yuan Zhou
1
,
2 ,
Xiaoning Wang
3 ,
Yingchun Zhang
1 ,
Dan Liu
1
,
2 ,
Binbin Zhao
1 ,
Lixin Duan
1
,
4 ,
Xiaoquan Qi
1
,
Publication date (Electronic):
9 February 2018
Journal:
Nature Communications
Publisher:
Nature Publishing Group UK
Read this article at
There is no author summary for this article yet. Authors can add summaries to their articles on ScienceOpen to make them more accessible to a non-specialist audience.
Abstract
In flowering plants, the pollen coat protects the released male germ cells from desiccation and damage during pollination. However, we know little about the mechanism by which the chemical composition of the pollen coat prevents dehydration of pollen grains. Here we report that deficiency of a grass conserved triterpene synthase, OsOSC12/OsPTS1, in rice leads to failure of pollen coat formation. The mutant plants are male sterile at low relative humidity (RH < 60%), but fully male fertile at high relative humidity (>80%). The lack of three major fatty acids in the pollen coat results in rapid dehydration of pollen grains. We show that applying mixtures of linolenic acid and palmitic acid or stearic acid are able to prevent over-dehydration of mutant pollen grains. We propose that humidity-sensitive genic male sterility (HGMS) could be a desirable trait for hybrid breeding in rice, wheat, maize, and other crops.
Abstract
In flowering plants, the pollen coat surrounds the male germ cells and protects against dehydration, damage and pathogen attack. Here, the authors show that a deficiency in terpenoid synthesis results in rice pollen over-dehydration and leads to a humidity-sensitive conditional male sterile phenotype.
Related collections
Most cited references28
- Record: found
- Abstract: found
- Article: not found
Triterpene biosynthesis in plants.
Ramesha Thimmappa, Katrin Geisler, Thomas Louveau … (2014)
The triterpenes are one of the most numerous and diverse groups of plant natural products. They are complex molecules that are, for the most part, beyond the reach of chemical synthesis. Simple triterpenes are components of surface waxes and specialized membranes and may potentially act as signaling molecules, whereas complex glycosylated triterpenes (saponins) provide protection against pathogens and pests. Simple and conjugated triterpenes have a wide range of applications in the food, health, and industrial biotechnology sectors. Here, we review recent developments in the field of triterpene biosynthesis, give an overview of the genes and enzymes that have been identified to date, and discuss strategies for discovering new triterpene biosynthetic pathways.
- Record: found
- Abstract: found
- Article: not found
Genetic regulation of sporopollenin synthesis and pollen exine development.
Tohru Ariizumi, Kinya Toriyama (2010)
Pollen acts as a biological protector of male sperm and is covered by an outer cell wall polymer called the exine, which consists of durable sporopollenin. Despite the astonishingly divergent structure of the exine across taxa, the developmental processes of its formation surprisingly do not vary, which suggests the preservation of a common molecular mechanism. The precise molecular mechanisms underlying pollen exine patterning remain highly elusive, but they appear to be dependent on at least three major developmental processes: primexine formation, callose wall formation, and sporopollenin synthesis. Several lines of evidence suggest that the sporopollenin is built up via catalytic enzyme reactions in the tapetum, and both the primexine and callose wall provide an efficient substructure for sporopollenin deposition. Herein, we review the currently accepted understanding of the molecular regulation of sporopollenin biosynthesis and examine unanswered questions regarding the requirements underpinning proper exine pattern formation, as based on genetic evidence.
- Record: found
- Abstract: found
- Article: found
Genomic analysis of hybrid rice varieties reveals numerous superior alleles that contribute to heterosis
Xuehui Huang, Shihua Yang, Junyi Gong … (2015)
The phenomenon that the heterozygous first filial (F1) generation often has better performance than its homozygous parents is known as heterosis or hybrid vigour1 2 3. The development of heterotic crops, especially those for hybrid rice and maize, is one of the most important applications of genetics in agriculture, and now over half of the rice and maize production worldwide is from hybrid seeds that lead to tremendous increases in yield4. Although rice is a self-pollinated organism and nearly all traditional rice cultivars are inbred lines, a hybrid seed production mechanism has been developed using systems based on cytoplasmic (three-line hybrid system) and environmentally sensitive (two-line hybrid system) genetic male sterility since the 1970s5 6. The genetic mechanism of heterosis has been explained by three non-mutually exclusive hypotheses, including dominance (complementation)7 8, overdominance9 10 and epistasis11 12. Further molecular genetic and genomic approaches have been used to investigate the heterotic performances in plants13 14 15 16 17 18 19. Single-locus overdominance of heterozygous alleles has been shown to result in heterosis straightway in Arabidopsis 13, tomato14 and maize15. In rice, quantitative trait locus analysis in an indica–japonica hybrid suggested that dominance complementation was the major cause of heterosis16. Recently, the genetic dissection of yield traits using an ‘immortalized F2’ population from an indica–indica rice hybrid cross enabled the assessment of genetic composition of yield heterosis17, which showed that the relative contributions of the genetic components varied with different yield traits18. Moreover, a genomic and metabolic approach has been applied to predict complex heterotic traits in hybrid maize19. To elucidate the genetic basis of rice heterosis, we developed an integrated genomic framework that exploited population-scale genomic landscapes from a representative number of hybrid rice varieties and parental lines to map the heterotic loci at fine scales. We collected and sequenced 1,495 diverse varieties of hybrid rice (F1), which are grown on >15 million hectares per annum and contribute greatly to the agricultural production4. The hybrid rice varieties were extensively phenotyped for grain yield, grain quality and disease-resistance traits. This approach enabled us to analyse the genomic structures of the rice hybrids and to identify the heterotic loci and the genetic effects of both the homozygous and heterozygous genotypes. This research provides new insights into the principles of hybrid vigour and has implications for rice breeding. Results Genomic architecture and heterozygosity of rice hybrids In an attempt to investigate as many rice hybrid combinations as possible, we sampled a total of 1,495 diverse varieties of hybrid rice (Supplementary Data 1), many with publicly available agricultural production statistics and pedigrees. Nearly all elite rice hybrids that are widely cultivated during recent years were included in the collection. The hybrids were sequenced with twofold genome coverage (an average of 2.2x), and a total of 1.2 Tb of genome sequence was generated (Supplementary Fig. 1). After sequence alignment of all the paired-end reads against the rice reference genome sequence, we called genotypes of the hybrid rice at 1,654,030 single-nucleotide polymorphism (SNP) sites (>4 SNPs per kb on average). Using the software Beagle20, a fine-scale genome map for all the hybrid rice varieties was therefore generated. To estimate the accuracy of the inferred genotypes, four hybrid rice varieties were sequenced independently at a high coverage (~40x genome coverage for each), and genotype calls from the deep-sequencing data were consistent with the imputed genotypes at a specificity of over 97.8% (Supplementary Table 1). We also sequenced 90 inbred lines (an average of 14.5x genome coverage) that were commonly used as the parents of hybrid rice (Supplementary Table 2). To evaluate the data sets from a number of hybrid combinations, we further analysed the genomes of 35 parents–child trios, in which the F1 hybrids and both their parents were sampled and sequenced in this study. In the 35 trios, the experimentally determined genotypes from both the parents are highly concordant (97.5%) with the imputed genotypes of the corresponding F1 hybrids (Supplementary Fig. 2). Hence, the independent data sources confirmed the accuracy of the genotypes of the rice hybrids sequenced. We used the SNP data to investigate the population structure of the rice hybrids. Indica and japonica are two major subspecies in cultivated rice (Oryza sativa L.). Of the 1,495 hybrid rice varieties, the majority (1,439 varieties) was from indica–indica crosses (Fig. 1a and Supplementary Fig. 3). The rest belonged to either indica–japonica (n=18) or japonica–japonica crosses (n=38). Indica–japonica and japonica–japonica hybrids have not been fully applied in agriculture up to now, probably due to the partial reproductive isolation in indica–japonica crosses and the limited diversity of japonica subspecies for japonica–japonica hybrids. According to the neighbour-joining (NJ) tree generated from whole-genome SNPs, there were no obvious population differentiations within the indica hybrids (Fig. 1b). We observed that the relationships depicted in the NJ tree agreed well with our pedigree-based expectations—the hybrid varieties derived from the same maternal lines (male sterile lines) tended to be clustered in the same clades in the NJ trees (Supplementary Fig. 4). We further analysed the heterozygosity of the hybrid rice varieties. For each hybrid rice variety, the level of heterozygosity was measured as the frequency of the heterozygous genotype at the 1.65 million SNP sites. As expected, the group of indica–japonica crosses has the highest level of heterozygosity (45.1% on average, with the top one reached 56.6%), followed by indica–indica crosses (21.8% on average) and japonica–japonica crosses (15.8% on average; Fig. 1c). We detected the genomic regions with unusual heterozygosity based on the Hardy–Weinberg equilibrium (HWE). HWE is an ideal situation, while selection or nonrandom mating can disequilibrate HWE at least for some local genomic loci. In the population of 1,439 indica hybrids, we compared the observed heterozygosity (Ho) with the expected heterozygosity (He, calculated by the HWE) at each SNP site and observed that many SNPs tended to have higher heterozygosity (Supplementary Fig. 5). We used ‘Ho−He’ to estimate the excess of homozygotes above the Hardy–Weinberg expectations in the population, and detected 3,520 highly heterozygous SNPs (Ho−He>0.4, 0.2% of the total SNPs) that were distributed randomly throughout the whole genome and eight genomic loci with extremely low-heterozygous SNPs (Ho−He 0.6), which resulted from allele differentiation between the maternal and paternal lines (Supplementary Table 14). The associations include two loci underlying plant height (Hd3a and Hd1)39 40, two loci underlying heading date (Hd1 and Ehd1, Fig. 3d)40 41 and OsC1 controlling the green or purple leaf sheath42 that help to distinguish the maternal and paternal lines in the field production. Trait–trait dynamics and genotype–environment interaction We investigated the genetic effects of the superior alleles of yield-related traits on yield per plant, and found that only a small fraction of the associated loci for the yield-related traits had direct influences on yield per plant (Supplementary Fig. 24). Among the associations with effects on yield, the superior alleles of heading date, plant height and grain number generally had positive effects on yield per plant, while the superior alleles of grain weight generally had negative effects on yield per plant. The flowering time genes (for example, OsSOC1, Ghd8 and Ghd7)35 43 44 have the largest effects on grain yield (longer growth stage for higher yield), followed with the genes regulating spikelets (for example, OsSPL14 and Gn1)34 36 37. For the genes underlying grain size and weight30 45 46, the alleles of GS3 and GW2 that increase grain weight represented significantly lower grain yield, while the alleles of qSW5 that increase both the grain weight and yield brought higher degree of chalkiness in rice grains. We further analysed the effects of grain quality-associated loci on the yield per plant. As expected, the superior alleles of grain quality generally had negative effects on the yield per plant. For the GWAS loci underlying chalky grain rate, lower chalky grain rate was generally accompanied with lower grain yield (Supplementary Table 15). Therefore, it needs to make a tradeoff between grain yield and other factors (for example, grain quality, growth duration and disease resistance) in breeding. Grain yield is a function of various components and factors under different complex agro-ecosystems. We found phenotypic changes between the traits in Sanya and those in Hangzhou (Supplementary Table 16). Among them, plant height and grain number fell over 20% in Sanya as compared with those in Hangzhou. In the hybrid rice panel, we observed that many associations (even with very strong signals) in one location showed weak or no associations in the other location. GWAS for heading date can serve as a typical example (Fig. 3a and Supplementary Fig. 16a). Through comparison of GWAS results between Sanya and Hangzhou, OsSOC1 and Ghd7 was significantly associated with heading date variation in both Sanya and Hangzhou. The two loci Ehd1 and Hd1 that were identified in Sanya showed relatively weak association signals in Hangzhou, while Hd3a and Ghd8 (controlling plant height as well) with weak associations in Sanya had strong associations in Hangzhou. Similar phenomenon was observed for plant height and grain number. The results suggest that these genes and loci were under the strong regulation of photoperiod and temperature conditions. GWAS result of grain weight was an exception to those findings, for which the major loci (for example, GS3, GW2 and qSW5) played their roles in both Sanya and Hangzhou consistently. Genomic landscapes of heterosis in hybrid rice population The exploitation of hybrid vigour in rice has been focusing on increasing grain yield. This might be the reason why most of the available rice hybrids do not possess better grain quality and disease resistance than the inbred lines. It should be noted that, to collect all the parental varieties of the 1,495 hybrid varieties is impractical, and many maternal lines of the hybrid varieties are of low fertility, which can obstruct phenotyping works. Nevertheless, there were parental homozygous genotypes for most loci in the hybrids themselves, which enabled the analyses of the heterotic loci and the dominance/overdominance effects. We evaluated the effect of heterozygous loci for the GWAS peaks above the suggestive P value (P 0.03 in 1,439 indica hybrids) for GWAS and heterosis analysis. Analysis of parental line haplotypes We designed a computational method to carry out whole-genome haplotype construction for the hybrids, with the aid of the genome sequences of 90 commonly used inbred parents. We proposed that the pedigree relationship of the hybrids can facilitate the inference of the parental lines’ haplotypes. For example, the hybrid Teyou86 (one rice hybrid variety in our collection) was from a cross between Minghui86 (the male parental line, one of the sequenced 90 inbred lines) and LongtefuA (the female parental line, neither the seed nor its DNA were sampled in our collection). Since the genome data of both the F1 (Teyou86) and the male parent (Minghui86) was already available, the haplotype of LongtefuA could be inferred directly and accurately. Furthermore, LongtefuA was the female parental line of 51 hybrids in our hybrid rice panel. Hence, the male parental lines of the remaining 50 hybrids could be inferred subsequently. Considering that, we developed a multiple-iteration approach for whole-genome haplotype reconstruction. We calculated the kinship value between the full-imputed genotypes of each hybrid and the genotypes of each inbred parental line (n=90 in the initial running round). For each hybrid, the inbred line with the highest kinship value was selected as ‘the candidate parent’, unless none of the inbred parental lines in the collection were found to show a close kinship with the hybrid’s genotypes. In haplotype reconstruction, the phased genotype of the hybrid and the haplotype of ‘the candidate parent’ were compared to infer the haplotype of the other parent. For the SNP sites where the genotype of ‘the candidate parent’ was missing (or it was contradictory with that in the hybrid genome), we used the phasing information of the nearest SNPs for local haplotyping. After the first running round, the haplotypes of all the inferred parents were collected together to generate a new panel of inbred parental line. The second running round was continued by comparing the 1,495 hybrids with the new inbred parental line panel (n=225 in the second running round). A total of four iterations were carried out and finally the parent’s information of 1,361 hybrids (91% of our collection) was retrieved. The deduced parental genotypes were estimated to be in agreement with the real genotypes at >98.0% of the total SNP sites. The computational analysis for the haplotype reconstruction was only based on genome data, where no pedigree records were used. Population genetics analysis The individual ancestries were estimated from whole-genome SNPs using the software ADMIXTURE (version 1.23)53. The matrix of pair-wise genetic distance derived from simple SNP matching coefficients was used to construct phylogenetic trees using the software PHYLIP51 (version 3.66)54. The software MEGA5 was used for visualizing the phylogenetic trees55. Principal-component analysis of the SNPs was performed using the software EIGENSOFT (version 5.0.1)56. The sequence diversity statistics (π) were computed in each 100-kb window of the rice genome. In the analysis of selections in modern breeding, the value of π was calculated for modern hybrids (π hybrid) and traditional landraces (π landrace) in indica and japonica, respectively, and the ratio of π in the population of landraces to that in the population of hybrids (π landrace/π hybrid) was used to detect the genetic-improvement sweeps. The genomic regions where both the hybrids and the landraces showed a low level of genetic diversity or the regions that had too many missing data and repetitive sequences were excluded for further analysis. The frequencies of ‘the homozygous genotypes of both the reference alleles’ (rr), ‘the homozygous genotypes of both the alternative alleles’ (aa) and the ‘heterozygous genotypes’ (ra) in the 1,439 indica hybrids were counted at each SNP site, from which we computed the values of the Ho and minor allele frequency. The He was calculated by the HWE at each SNP site. HWE tests were performed by the ‘genetics’ package in R language. The heterozygosity of each hybrid rice variety was measured as the proportion of the heterozygous genotypes at all the SNP sites. Phenotyping and GWAS Approximately 18 seeds for each variety from the collection of hybrids were germinated and planted in the experimental fields in Hangzhou, China (at N 30.32°, E 120.12°) in summers of 2012 and 2013, and in Sanya, China (at N 18.65°, E 109.80°) in winter of 2012. All the 1,495 accessions were grown in the consecutive farmland with well-distributed soil status and uniform condition. Lands tilling and raking were conducted as even as possible to make equal growing conditions for each accession. The phenotyping for this work involved a wide range of agronomic traits for grain yield, grain quality and disease resistance. The field traits including heading date, plant height, flag leaf length, flag leaf width, panicle number and panicle length, were measured directly in the field. Heading date was recorded daily as the number of days from sowing to the observation of first inflorescences that emerged above the flag leaf sheath. In total, 75 hybrid rice varieties showed extremely late maturing in Hangzhou, which were removed in GWAS for heading date in Hangzhou. Plant height, flag leaf length, flag leaf width and panicle length were measured for at least three samples of each accession. The grain-related traits, including grain number per panicle, grain length, grain width and grain weight per 1,000 grains, were measured in the laboratory following harvest. Grain weight was initially obtained by weighing ~600 fully filled grains, which was then converted to 1,000-grain weight value. For measuring grain quality traits, the fully filled grains from the hybrids planted in Sanya (from winter of 2012 to spring of 2013) were used. Harvested rice grains were air-dried and stored at room temperature for at least 3 months until grain moisture content fell to be <13.5%. The grain quality traits were then phenotyped according to the Chinese national standard (NY/T 2334-2013 ‘Determination of head rice yield, grain shape, chalky rice percentage, degree of chalkness and translucency—an image analysis method’). Amylose content and alkali digestion value of the grains were phenotyped and scored using milled rice grains57 58. To evaluate the disease-resistance traits of the hybrids, plants were germinated in Hangzhou in the summer of 2013, and incubate by the spraying method in low light and at room temperature to insure sporulation and subsequent reinfection of susceptible plants. The spraying method was carried out for the disease-resistance assay59. In brief, rice seedlings were inoculated by spraying fresh preparation of conidial suspension. The inoculated rice varieties were then planted in air-conditioned greenhouses, followed by the phenotyping of disease-resistance traits. The disease reactions were measured about 7 days after inoculation and evaluated and scored by the disease leaf area. Association analysis was conducted using the EMMAX software package. The matrix of pair-wise genetic distance derived from simple matching coefficients of SNPs was used to model the variance–covariance matrix of the random effect. Permutation tests were used to help define the genome-wide significant P value threshold60. We picked 10 traits, reshuffled the original phenotype data, and then performed association analysis using EMMAX with the same parameters. There ought to be no real associations between the SNPs and the ‘simulated’ phenotypes, so all the SNPs passing the threshold should be false positives. A total of 100 permutation analyses were performed, which detected three ‘association signals’ passing the whole-genome significant cutoff 10−6. GWAS on 38 real phenotypes identified a total of 130 association signals passing the threshold 10−6, which suggested a feasible FDR level of <0.01. Moreover, FDR value was calculated to be 0.2 for the suggestive P value of 10−4 according to the permutation tests. Additive, dominant and recessive models were also tested for the traits. The EMMAX software package simply follows the encoding scheme of the genotype data in the additive model by default. For dominant and recessive models, heterozygous genotypes (ra) were changed to be ‘homozygous genotypes of both the reference alleles’ (rr) or ‘homozygous genotypes of both the alternative alleles’ (aa) for all the SNPs. The newly made genotype data was then inputted into the EMMAX software package for GWAS. Heterosis analysis in rice hybrids For the parent–child trios, phenotypic performances (heading date, plant height, grain number per panicle and yield per plant) of both the F1 hybrids and their parental lines were used to evaluate the amount of heterosis (middle parent heterosis and over parent heterosis). In each parent–child trio, the middle parent heterosis index was calculated using the phenotypic measurements of the F1 hybrid and both the parents. Moreover, the parent–child trios could be divided into three types: positive over parent heterosis, negative over parent heterosis and F1 ranging between the parents. The proportions of the three types were counted as well to evaluate the amount of heterosis in the hybrid rice population. The degree of dominance ‘d/a’ was calculated using the peak SNPs of the associated loci, where ‘d’ and ‘a’ referred to the dominant effect and the additive effect, respectively. The effects of heterozygous alleles were analysed for the GWAS peaks above the suggestive P value (P<10−4, from the linear mixed model) underlying the yield per plant, panicle number, grain number and plant height. The SNP sites in which heterozygous genotypes or homozygous genotypes of both the minor alleles had a frequency of ≤15 in number were excluded in the calculation of ‘d/a’. The effects of heterozygous and homozygous genotypes were calculated for each peak SNP of the associated locus, where the average phenotypic measurements of the heterozygous genotypes and homozygous genotypes were calculated, respectively. Associated loci were screened in each 500-kb window of the rice genome, for the traits heading date, plant height and grain number per panicle in Sanya. The lowest GWAS P value (from the linear mixed model) of the SNPs in each 500-kb genomic region was recorded as the association signal of the loci. The peak SNPs at the top 100 associated loci (ranked in association signals) were used in the analysis. For each associated locus, the allele with better yield performance (for example, more grain numbers, higher plant and longer heading date) was defined as the superior gene allele. The computational simulations under the two scenarios were performed using the sim.map and sim.cross functions in the R/Bioconductor package, with recombination under the real genetic map in rice. In the first scenario, we generated in silico genotype data of 5,000 recombinant inbred lines. In the second scenario, we generated in silico genotype data of 500 BC5F3 lines (Supplementary Fig. 37). Author contributions B.H. conceived the project and its components. X.H. and B.H. designed the studies and contributed to the original concept of the project. S.Y. and J.G. contributed to the collection of hybrid rice. J.G., Qilin Z., B.C., J.X., N.C., Z.H. and S.Y. contributed in the phenotyping of the hybrid rice. W.L., Y.L., C.Z., D.F., Q.W. and Q.F. performed the genome sequencing. X.H., Y.Z., H.G. K.L., C.Z., T.H., L.Z. and Q.Z. performed the genome data analysis. X.H., Y.Z. and H.G performed GWAS, population genetics and statistical analyses. J.L. and Z.-X.W. contributed to the functional analyses. X.H. and B.H. analysed the whole data and wrote the paper. Additional information Accession codes: DNA sequencing data are deposited in the European Nucleotide Archive ( http://www.ebi.ac.uk/ena/) under accession numbers ERP005527. How to cite this article: Huang, X. et al. Genomic analysis of hybrid rice varieties reveals numerous superior alleles that contribute to heterosis. Nat. Commun. 6:6258 doi: 10.1038/ncomms7258 (2015). Supplementary Material Supplementary Information Supplementary Figures 1-37 and Supplementary Tables 1-17. Supplementary Dataset 1 The list of 1495 hybrid rice varieties sampled in the study Supplementary Dataset 2 Phenotype data measured in Sanya Supplementary Dataset 3 Phenotype data measured in Hangzhou
Author and article information
Contributors
Xiaoquan Qi:
ORCID: http://orcid.org/0000-0001-7175-115X
xqi@ibcas.ac.cn
Journal
Journal ID (nlm-ta): Nat Commun
Journal ID (iso-abbrev): Nat Commun
Title:
Nature Communications
Publisher:
Nature Publishing Group UK
(London
)
ISSN
(Electronic):
2041-1723
Publication date
(Electronic):
9
February
2018
Publication date PMC-release: 9
February
2018
Publication date Collection: 2018
Volume: 9
Electronic Location Identifier: 604
Affiliations
[1
]ISNI 0000000119573309, GRID grid.9227.e, Key Laboratory of Plant Molecular Physiology, Institute of Botany, , Chinese Academy of Sciences, ; Nanxincun 20, Fragrant Hill, Beijing 100093 China
[2
]ISNI 0000 0004 1797 8419, GRID grid.410726.6, University of Chinese Academy of Sciences, ; Yuquan Road 19, Beijing, 100049 China
Author information
Xiaoquan Qi
http://orcid.org/0000-0001-7175-115X
Article
Publisher ID: 3048DOI: 10.1038/s41467-018-03048-8
PMC ID: 5807508
PubMed ID: 29426861
SO-VID: 5d4c801c-5e95-4d27-9e8a-54245672aff3
Copyright © © The Author(s) 2018
License:
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
History
Date
received
: 21
December
2016
Date
accepted
: 16
January
2018
Categories
Subject:
Article
Custom metadata
issue-copyright-statement © The Author(s) 2018
ScienceOpen disciplines: Uncategorized
Data availability:
ScienceOpen disciplines: Uncategorized