Portal de Periódicos da CAPES

B‐chrom: a database on B‐chromosomes of plants, animals and fungi

2017; Wiley; Volume: 216; Issue: 3 Linguagem: Inglês

10.1111/nph.14723

1469-8137

Ugo D’Ambrosio, M. Pilar Alonso‐Lifante, Karina Barros, Aleš Kovařı́k, Gemma Mas de Xaxars, Sonia García,

Plant Pathogens and Fungal Diseases

We present here B-chrom (www.bchrom.csic.es), an online database with comprehensive information on B chromosomes (Bs) for plants, animals and fungi. Data have been extracted from 3041 sources published between 1907 and July 2016. There are 5760 entries corresponding to 2828 species. Besides presence, number or range of B chromosomes, the database provides information on chromosome number, ploidy level and genome size when available. After an extensive publication search strategy and data mining, the content of the database has been analysed statistically and significant positive correlations, though faint, have been found between the average number of Bs, chromosome number and ploidy level. Some of the species with the highest number of Bs in plants reproduce asexually, which may be related to an accumulation of these selfish genomic elements. Despite the increased interest in B chromosome research in recent years (20% of the data in B-chrom was published in the last decade) there is still limited knowledge with respect to global biodiversity. The database is the first step to systematize information on Bs, and we expect that it will be used by scientists interested in cytogenetics for data-mining or comparative studies involving B chromosomes. B chromosomes, also known as supernumerary or accessory chromosomes, are additional and non-essential constituents of karyotypes. Their most significant traits are that: (1) they may be present in some, but not all, individuals of a population, or in certain cells of the same individual but not in all; and (2) they fail to recombine with A chromosomes during meiosis (Jones et al., 2007; Houben et al., 2013). B chromosomes are usually, but not always, smaller than A chromosomes and they are a sometimes overlooked source of intraspecific genome size variation (e.g. natural rye plants can have from zero to four Bs, which can clearly affect the constancy of its C-value; Jones, 1976). They are usually considered useless, but both favourable (e.g. antibiotic resistance in the fungus Nectria haematococca, Coleman et al., 2009; selective environmental advantages in Allium schoenoprasum, Holmes & Bougourd, 1989) and adverse (e.g. lower vigour and impaired fertility in Dactylis glomerata hybrids, Williams & Barclay, 1968; parasite-like behaviour of Bs in Eyprepocnemis plorans, Camacho et al., 2003) effects of Bs have been described. The presence of Bs is not associated with any phenotype in most cases (Valente et al., 2017), particularly when the number of Bs is low. The maximum number of Bs tolerated is variable across species and it is counterbalanced by their potential negative consequences, particularly on fertility and vigour (Houben, 2017). In genera displaying apomixis, the presence of Bs is well documented, such as in Boechera where these may be involved in the genetic control of apomixis (Kantama et al., 2007; Mandáková et al., 2015). There are also interesting relationships between Bs and sex chromosomes: Sharbel et al. (1998) proposed a sex chromosome as the ancestor of the B chromosome in a frog species; conversely, in some cichlid fish species it was hypothesized that a portion of sex chromosomes is derived from Bs (Yoshida et al., 2011). B chromosomes were first observed in insects from the genus Metapodius (now Acantocephala) (Wilson, 1907). In plants, they were discovered in crops from the genus Secale (Gotoh, 1924). Since then, thousands of reports have steadily increased our knowledge of the distribution and features of Bs across life on Earth. B chromosomes have been classically considered nonfunctional. However, it is not until recently that active genes have been found in Bs. For example, ribosomal RNA (rRNA) genes may play a role in the evolution of Bs, as these have been detected on Bs of many plant and animal species. In Plantago lagopus, a new B chromosome arose from the extensive amplification of 5S rRNA genes (Dhar et al., 2002). Also, transcription of rRNA genes was the first molecular evidence of gene activity in Bs, again both in plants and animals (Leach et al., 2005; van Vugt et al., 2005). Later, genes from other multigene families such as H1, H3 and H4 histones, as well as U2 snRNA, transposable elements and satellite DNA, have also been documented as components of Bs (Valente et al., 2017). There is no general ubiquitous mechanism for the evolutionary origin of Bs. Most likely, there are several possible origins; the most widely accepted is that they are derived from the A chromosome complement (Houben et al., 2013). After unbalanced or asymmetric translocation, small centric fragments can also become Bs (Jones & Rees, 1982). Some evidence also suggests that Bs can act as diploidizing agents after a polyploidization process (Jones & Houben, 2003). The study of Bs started with classical karyological methods and now even -omics approaches are being used for this research (Valente et al., 2017). Thousands of reports on Bs have been produced during the last two centuries, yet an overview of its distribution across the tree of life is still missing, as well as a clear understanding of their role. As more and more organisms have been found to harbour accessory genetic materials over the last decades, the number of literature reviews on Bs has increased substantially, including overarching examinations (Borisov, 2014), revisions based on taxonomic groups (e.g. Jones et al. (2007) and Datta et al. (2016) for plants, Palestis et al. (2010) for orthopterans, Makunin et al. (2014) for mammals, or Galazka & Freitag (2014) and Stukenbrock & Croll (2014) for fungi), as well as reviews analysing their putative role in genetics and evolution (Camacho et al., 2000; Banaei-Moghaddam et al., 2015). The phylogenetic diversity within and between taxa harbouring Bs suggests that this polyphyletic phenomenon is a rather common one, with their complex characteristics and dynamics still poorly understood from a systemic perspective (Valente et al., 2017). There are also periodical international conferences devoted to Bs (i.e. B-chromosome Conference, the last one held in Gatersleben, Germany, in 2014). Jones & Díez (2004) compiled a comprehensive database which included any report on the presence of Bs published between 1907 and 1994. However, the database has not been updated and many reports on Bs have been released since 1994. Currently, there is a wealth of information available but sometimes the access is difficult, as this can be published in a variety of local or national journals. Taking advantage of the current and powerful literature search engines, the main purpose of this work was constructing a new resource, which included extensive data on presence and numbers of Bs in plants, animals and fungi, and making it available online. The number of publications which include information on Bs is remarkable, and interest remains high (Fig. 1). To our knowledge, there are no online initiatives offering such information and embracing the three largest biological kingdoms. We expect that this comprehensive and updated catalogue of species presenting Bs will contribute to the understanding of these 'ultimate genome parasites' (Jones et al., 2007) allowing the analysis of their distribution across the tree of life. In order to obtain the data, a search strategy was created to retrieve scientific documents which included reports on Bs. The online bibliographic databases used were Scopus (https://www.scopus.com/), Web of Science (WOS, https://apps.webofknowledge.com/), SciELO (http://www.scielo.org/), Directory of Open Access Journals (DOAJ, https://doaj.org/) and Google Scholar (https://scholar.google.es/). Searches were limited to a specific time period (1995–July 2016) since the information provided by the 'B chromosome database' (Jones & Díez, 2004) comprising years 1907–1994 would be incorporated in B-chrom. Database queries were conducted from January to July 2016. The search strategy was adjusted to the different interfaces of each bibliographic database. In particular, 'B chromosome' was searched in the fields Title, Abstract and Keywords both in Scopus and WOS. However, in Google Scholar, searches can only be made using two fields: In the title of the article or Anywhere in the article. We selected the former due to the amount of noise generated by the latter. With regard to SciELO and DOAJ, the search was performed in all fields given the considerably lower number of results obtained with the previous search strategy. Scopus was our choice for preparing the initial corpus of publications due to its wider coverage of documents as compared with WOS, which indexes fewer publications (Mongeon & Paul-Hus, 2016). When the relevance of certain documents was not clear, we looked for the presence of other related keywords such as 'supernumerary chromosome', 'accessory chromosome' or 'selfish chromosome'. Apart from the presence of these keywords, we also evaluated where they appeared (title, abstract, keywords, etc.) to assess their relevance. As searches were performed, results were downloaded in CSV format, except in the case of DOAJ, where this option was not available. Subsequently, we used Google Sheets for analysing the results of our searches. We discarded duplicates and not relevant documents and we obtained the complete text of all the documents we had access to. Papers without at least a summary in English were excluded. From this preliminary corpus of documents (2837 publications), we created a bibliographic database using Zotero (https://www.zotero.org). The references from 'B chromosome database' (our starting point database) were manually added to that one. During this process, errors were corrected (spelling mistakes in author names, wrong publication years, etc.) and DOI identifiers were included (or URLs when DOIs were not available) to complete the bibliographic references. In its final form, B-chrom includes data from 3041 references (mostly journal articles, but also books and meeting proceedings): 2410 coming from Jones & Díez (2004) and 631 obtained from the five bibliographic databases analysed (meaning that 22% of the newly retrieved documents had relevant information). Note that the initial work by Jones & Díez (2004) covered 1906 to 1994, while our searches encompassed only from 1995 to July 2016. The information was manually extracted from each source publication and when available (in most cases), the presence and number of Bs were visually checked in the figures. Data were introduced in a Google Sheet filling the following fields for each entry: (1) kingdom, (2) phylum or (sub)division, (3) popular name of the containing group, (4) class, (5) order, (6) family, (7) genus, (8) specific epithet, (9) complete species name, (10) ploidy level, (11) somatic chromosome number (2n), (12) B-chromosomes (see later), (13) complete citation reference in APA format, and (14) DOI or URL where the source publication is available. For angiosperm plants, there was an additional category, (15) monocots or eudicots. Each entry and each publication had unique identification numbers. The information available from the work by Jones & Díez (2004) until 1994 and the IPCN compilations from years 1994–2006 (Goldblatt & Johnson, 1998, 2000, 2003, 2006, 2010) were imported to the spreadsheet and formatted according to the earlier mentioned structure. Data on Bs are displayed as either: presence, with the letters 'Bs', and a specific number (e.g. 1, 2, 8) or as a range (e.g. 2–6), depending on the information provided by the source publication. When the number of Bs referred to the gametic cells it was indicated with the letter 'G' after the number of Bs. Also, when Bs were found either in male or in female individuals it was shown by the letters 'm' and 'f', respectively. When the number of Bs was found at the haploid level we used the letter 'n' (the case of some liverworts). Finally, the letters 'PRS' indicate Paternal Sex Ratio, types of Bs occurring in certain arthropods (Werren & Stouthamer, 2003). Release 1.0 of B-chrom was launched in March 2017. Access to the information is made easy through a search box in which queries per genus or per species can be inserted. Additionally, data can be browsed by the largest groups present in the database (eudicots, monocots, gymnosperms, fungi, fish, insects and mammals) directly from the home page or from the 'Browse' tab. Data are returned in customizable tables in which users can select to display their desired options. Search results are also downloadable as CSV files from the institutional repository Digital CSIC. The default settings of a regular search include family, genus, species name, ploidy level, chromosome number, information on Bs (presence, number or range) and complete citation reference linked to its DOI or URL. Additionally, the tab 'Publications' offers the complete reference list of the publications retrieved for data mining, while the tab 'Links' provides directions to other databases with cytogenetic data, as well as other useful web resources. Finally, the tab 'Contact' is intended to foster communication with researchers interested in providing new data or in making any comments or corrections to the database. The database structure was created in the MySQL server and hosted in www.bchrom.csic.es. The initial Google Sheet in which the data were compiled was imported to a CSV file. The website uses Laravel v.5.3 (https://laravel.com/; for PHP 5 developments) and Bootstrap v.3.3.7 (with HTML, CSS and JS; http://getbootstrap.com/) open source frameworks. The Taxonomic Name Resolution Service v.4.0 (accessed 15 September 2016) (http://tnrs.iplantcollaborative.org) was used for correcting and standardizing plant names in order to avoid ambiguous, superfluous or incorrect names resulting in mismatched or unwitting duplication of records (Boyle et al., 2013) followed an approach applied previously (Garcia et al., 2017). When ploidy levels and/or chromosome numbers were not indicated in the source publication, data have been extracted from the Chromosome Counts Database (CCDB) (accessed 15–25 November 2016) (http://ccdb.tau.ac.il) (Rice et al., 2015). Genome size data have been obtained through the Plant DNA C-values database (accessed 15–25 November 2016) (http://data.kew.org/cvalues), the GSAD database (accessed 15–25 November 2016) (http://www.asteraceaegenomesize.com) and the Animal Genome Size database (accessed 15–25 November 2016) (http://www.genomesize.com). If a species had a different number of Bs, chromosome number or ploidy level that differed between or within publications, we treated each difference as a separate entry. Statistical analyses were performed with RStudio, v.0.98.1078, a user interface for R (http://www.rstudio.com). Duplicates were removed from the dataset before all analyses. Since datasets were not normally distributed, we performed the non-parametric Spearman rank correlation for three analyses: (1) number of Bs vs chromosome number (2n); (2) number of Bs vs ploidy level; (3) number of Bs vs genome size (2C). These analyses have been performed at different taxonomic levels (see Supporting Information Tables S1–S4). Some assumptions were made: when there was a range of Bs for a given entry we have used the average value, and when the number of Bs was not specified, we have assumed it was one, for calculations. It is difficult to estimate in how many species Bs may be present because the representation in the dataset is highly biased for the reasons explained earlier. Besides, most estimates refer only to plants. Darlington & Wylie (1955) listed chromosome numbers of over 17 000 species of flowering plants of which 0.8% had Bs while Fedorov (1969) estimated 1.1% of plant species had Bs. More recently, Levin et al. (2005) reported Bs in 8% of monocots and 3% of eudicots (c. 4% of angiosperms). In the CCDB (Rice et al., 2015), one of the most recent resources providing chromosome numbers for plants, data are available for 77 958 species. Considering that we have assembled information on Bs for 2087 plant species, we can estimate that 2.68% of species with known chromosome numbers have Bs. In animals it is more difficult to make this calculation since chromosome counts are more scattered in this kingdom and there are no available lists as in plants, to our knowledge. In the same line, it is impossible to know if Bs are more frequent in animals, plants or fungi, since the high frequency of Bs in certain groups better reflects the intensity and technical ease with which each group has been studied (Camacho et al., 2000). In fact, the effect of study intensity of certain groups on the presence of Bs can be huge, as pointed out by Palestis et al. (2004), who found that species less studied had 12-fold lower Bs frequency than the most studied in the database of mammalian karyotypes. Levin et al. (2005) also reported a significant correlation between the proportion of Bs across angiosperms and the study effort. More factors certainly bias the assessment of species with Bs; that is most chromosome counts are based on a single or few individuals, most species have not been assessed cytologically, and the phylogenetic relationships between species should also be taken into account to evaluate the frequency of Bs. Current assessments of biodiversity estimate that there may be 7000 000 species of animals, 400 000 of plants and 1500 000 of fungi (based on Chapman, 2009) therefore the correct figure of species with Bs is certainly much larger. Yet it is astonishing that some groups lack a single representation in the database, besides considerable cytogenetic knowledge, as is the case of birds (e.g. Tegelström & Ryttman, 1981; Ellegren, 2010), with > 18 000 species estimated (Barrowclough et al., 2016). Interestingly, most birds present small DNA contents and a narrow range of its variation, indicating that a tight control of genome size could explain the absence of Bs. Vujošević & Blagojević (2004) also related the absence of Bs in birds with their small genome sizes. Another group of animals in which Bs have not been detected to date is phylum Cnidaria (9000 estimated species), the group containing corals, sea anemones and jellyfish, although their chromosome counts are very scarce too. In plants, algae have no reports on Bs, despite the size of the group (c. 12 000 species described) and karyological knowledge (e.g. Austin, 1956; Kapraun, 1993; Lewis, 1996; Muravenko et al., 2001). Ferns and mosses (with 15 000 estimated species each) are poorly represented, with only seven and 10 species with data on Bs, respectively, despite being well-known karyologically, with reports on 45 pteridophyte families and 120 bryophyte families in the CCDB (Rice et al., 2015). Regarding flowering plants, it is surprising that some well-known families with available chromosome counts do not present any reports on Bs, such as Apocynaceae (1500 species, with 805 counts in the CCDB), Berberidaceae (700 species, with 156 counts in the CCDB), Cactaceae (1700 species, with 698 counts in the CCDB), or Ericaceae (4250 species, with 700 counts in the CCDB). Data have been extracted from 3041 papers published between 1907 and 2016. Fig. 1 shows the data and publication distribution divided into periods of 10 yr. The latter period (2007–2016) is the most productive in terms of individual B-chromosome assessments (1157 entries, or 20.10% of the whole dataset) and the second in which more papers on Bs were published (553, after the period 1967–1976 with 704 publications). These figures illustrate that the interest in cytogenetics remains high amongst the scientific community, as found for other recently updated or released cytogenetic databases like the Plant rDNA database (Garcia et al., 2014) or the Animal rDNA database (J. Sochorová, S. Garcia, F. Gálvez, R. Symonova & A. Kovařík, unpublished), in which recent years also tend to be the most productive. Table 1 lists the journals which have published more documents containing information on Bs. The 5760 entries available in the database correspond to 2828 eukaryotic species which have been reported to present Bs in their genomes, of which 73.56% (2087 species) were plants (53.20% monocots and 46.80% eudicots), 25.95% (736 species) animals and only 0.49% (14 species) fungi, excluding duplicates. With respect to the previous database on Bs (Jones & Díez, 2004), our database represents an increase of 65% in the number of entries and of 61.5% in the number of species (3484 entries and 1757 species in the 'B chromosome database'). In total, data are available for 311 families (119 of plants, 185 of animals and seven of fungi), 1095 genera (635 of plants, 450 of animals and 10 of fungi). In animals, the sample is mostly composed of insects (52.28%), mammals (17.83%) and ray-finned fish (Actinopterygii) (16.28%). This is not surprising given the relative abundance, potential interest or suitability for cytogenetic studies of these groups. Indeed, insects account for the most diverse and abundant group of animals, mammals have been subject to deep cytogenetic studies, particularly in some model species like mice, and ray-finned fish are the dominant class of vertebrates, accounting for most (> 95%) extant fish species. In plants, the representation is again biased by economic interest (crops) or relative abundance, with Poales (21.76%), Asterales (20.75%) and Asparagales (15.72%) the most represented orders, which is in accordance with previous findings (Houben et al., 2013). As for fungi, data are only available for 14 species from several orders from the phylum Ascomycota, which is its largest phylum. Model organisms such as maize, rye and the grasshopper Eyprepocnemis plorans are the species which have been studied most times individually (with 309, 210 and 81 independent reports, respectively). Organisms harbouring Bs in our database present ploidy levels ranging from 1 (some fungi) to 22, although the range is narrower in animals (2–6) than in plants (1–22). The most common ploidy level is the diploid, accounting for 66% of the database, followed by the tetraploid (11%) (Fig. 2). Chromosome numbers range from 4 to 720, with animals having again a narrower range (6–150). In 13.30% of entries only presence (Bs) is recorded, and in 25.51% only one B-chromosome is found (the modal value), the remaining ranging from 1 to 50. The average number of Bs is 2.4, being one to two Bs the most common reports (44.31% of the database entries), followed by those that state only 'presence' of Bs (17.98%). The highest B chromosome numbers were detected in the plant species Pachyphytum fittkaui (Crassulaceae) with 2n = 120 + 50B, followed by Albuca bracteata (synonym Ornithogalum caudatum) (Asparagaceae) with 2n = 18 + 36B and Zea mays (Poaceae) with 2n = 20 + 34B. In the three species, vegetative reproduction is well known (for Crassulaceae, Guo et al., 2015; for Asparagaceae, Byers et al., 2014; for maize, Wolff, 1971). Also, in other taxa with a high number of Bs, asexual reproduction has been observed, as in Fritillaria japonica (Liliaceae) (2n = 22 + 26B) or Centaurea scabiosa (Asteraceae) (2n = 22 + 20B). Perhaps species that reproduce vegetatively can better tolerate the presence of Bs or genome size accumulation because of the absence of meiosis as a controlling mechanism of additional genomic load, although more data are needed to substantiate this hypothesis. Among animals, the highest B chromosome number is found in the rodent Apodemus peninsulae (Muridae) (2n = 48 + 30B), followed by the spider Clubiona japonicola (Clubionidae) (2n = 24 + 28B). [Correction added after first publication 25 July 2017: the preceding sentence has been amended.] Although extreme reports of Bs are found more frequently in plants than in animals, there are more animal than plant species harbouring ≥ 5 Bs (23.1% vs 17.3%) or ≥ 10 Bs (6.1% vs 3.7%), bearing in mind that our knowledge on Bs in animals is more limited than in plants. Usually, Bs do not exceed the A chromosome complement, however, this happens in 28 species and it can reach up to 5.5-fold of the A chromosome complement as in the case of Brachycome lineariloba (Asteraceae) (2n = 4 + 22B) (Smith-White & Carter, 1970). Certainly, the presence of Bs can interfere with the normal functioning of cells, including processes of mitosis and meiosis and this may be why an overload of Bs is rarely tolerated, while, one B chromosome is the most common situation. We have analysed the possible relationship between the average number of Bs and polyploidy, chromosome number and genome size at different taxonomic levels (Tables S1–S4) in plants and animals (the small number of available species of fungi does not allow proper statistical analyses). There is a very faint trend to higher number of Bs with higher ploidy levels, chromosome numbers and genome sizes in both plants and animals although the relationship is only significant in eudicots and with chromosome number (rho = 0.082, P = 0.003) and ploidy level (rho = 0.080, P = 0.005), and in animals but only with chromosome number (rho = 0.127, P = 0.0005). Palestis et al. (2004) found a positive correlation of Bs with genome size, assessing B-chromosome frequency across angiosperms. In the same line, Trivers et al. (2004) found a strong positive correlation between Bs and genome size in British angiosperms, and Levin et al. (2005) hypothesized that species with small genomes would have a lower incidence of Bs, as larger genomes would better tolerate additional genetic material. However, to our knowledge the positive (though faint) relationship found between ploidy level and B presence had not been detected in previous works (Jones & Rees, 1982; Trivers et al., 2004; Levin et al., 2005). Nevertheless, the large and biased sample prevents general conclusions and these relationships are better studied at lower taxonomic levels such as family or genera. In these cases, certain groups behave completely differently from others: for example while in genera Artemisia, Bromus, Oryzopsis, Lolium, Diabrotica and Ophris the number of Bs is positively and significantly correlated with chromosome number, the contrary is true for genera Poa, Fritillaria, Crotalaria, Brachycome, Cytisus and Listera (see Table S3). It has been hypothesized that there are several mechanisms of origin of Bs and also different selection/environmental pressures may shape the destiny of B-chromosome behaviour depending on the group. Besides, in most groups there is no relationship with Bs, highlighting the independent, perhaps parasitic, nature of these unpredictable genomic components. We have also studied the possible influence of polyploidy on the presence of Bs in certain genera in which enough data were available (Table S5) to allow a proper analysis. In all three genera (Allium, Artemisia and Festuca, representatives of the most commonly studied families regarding Bs) we have found that species which are only present at diploid level have lower proportions of counts with Bs than species which are present at different ploidy levels (Fig. 3). We can hypothesize that the mechanism(s) promoting polyploidy and/or chromosome number variability may be related to the ways in which Bs might have arisen. De novo origin of Bs has been detected in the complex of Prospero autumnale (previously Scilla autumnalis) in which these may be by-products of its large-scale chromosomal rearrangements (Jang et al., 2016). The release of B-chrom has meant a considerable assembling effort, but it is still an initial step on the systematization of data on Bs. The database is envisaged as a long-term project, and in future updates we aim to include other relevant information such as Bs morphology (e.g. visibility and position of centromere), relative size with respect to the A-chromosome complement (e.g. micro- or macro-Bs), known gene content (e.g. rDNA), etc. Furthermore, in future releases, we would like to incorporate data from other information sources which have not been explored e.g. PhD theses or reference lists from the most important books and articles dealing with this topic. The analyses here presented are a reflection of the seemingly unpredictable nature of these 'passengers in the genome' (Jones et al., 2007). Yet, as pointed out previously, the database embraces a small fraction of eukaryotic diversity, highlighting some relevant gaps in knowledge. In particular, research should be triggered in unexplored groups such as birds, algae or fungi. The authors would like to thank Paula Bonaventura and María Luisa Gutiérrez (IBB-CSIC-ICUB, Barcelona, Spain) for their help in data mining and Teresa Garnatje (IBB-CSIC-ICUB) for carefully reading the manuscript. The authors acknowledge Sergi Garcia for the creation of the database web environment. The Dirección General de Investigación Científica y Técnica from the Government of Spain (CGL2016-75694-P), the Czech Science Foundation (P506/16/02149J) and the Government of Catalonia ('Ajuts a grups de recerca consolidats', 2014SGR514) are acknowledged for funding. S.G. benefits from a 'Ramón y Cajal' contract from the Government of Spain (RYC-2014-16608). S.G. planned and designed the research. U.D'A., M.P.A-F. and K.B. performed publication search and data collection. G.M.d.X., S.G. and A.K. analysed data. U.D'A., M.P.A-F., K.B. and S.G. wrote the manuscript with significant contributions from A.K. and G.M.d.X. Please note: Wiley Blackwell are not responsible for the content or functionality of any Supporting Information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office. Table S1 Results of the statistical analyses comparing plants and animals, on the one hand, and monocots and eudicots, on the other hand Table S2 Results of the statistical analyses at the family level Table S3 Results of the statistical analyses at the genus level Table S4 Results of the statistical analyses at the order level Table S5 Data used for the analysis performed in genera Allium, Artemisia and Festuca Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.