Differences in pathogen resistance between diploid and polyploid plants: a systematic review and meta-analysis
Data files
Aug 15, 2023 version files 70.72 KB
-
family_level_angiosperm_tree.txt
-
HagenandMason2023_data.xlsx
-
README.md
Abstract
Polyploidy, the state of having more than two full sets of chromosomes, has been hypothesized to provide several evolutionary advantages to flowering plants, including increased ability to resist pathogens and parasites. However, studies comparing pathogen resistance in conspecific and congeneric diploids and polyploids have produced mixed results. While the supposed relationship between polyploidy and pathogen resistance has been commented on in several narrative reviews, it has never been subjected to a systematic meta-analysis. We examined the effect of polyploidy on pathogen resistance by synthesizing 214 effect sizes from 128 studies. We find that, overall, there is no consistent effect of polyploidy on pathogen resistance. Subgroup analyses suggest that polyploids perform significantly better than diploids only in resisting hemibiotrophic pathogens, and autopolyploids tend show greater resistance than allopolyploids. This is surprising given the fact that polyploids possess extra allele copies of R-gene alleles that provide resistance to biotrophic pathogens, and this pattern may indicate that signaling cascades needed to elicit hypersensitive responses are disrupted by polyploidy. Disruption is supported by the observation that, across all pathogens, autopolyploids show significantly greater resistance compared to diploids, whereas allopolyploids do not. This is corroborated by the observation that synthetic autopolyploids perform significantly better than their allopolyploid and established counterparts. Regarding pathogen type, diploids show greater resistance than polyploids to pathogens that are fungi or nematodes. Analyses of publication bias indicate little to no bias, and analyses of heterogeneity indicate that phylogeny explains almost none of the observed heterogeneity. These results underscore the importance of not only systematic review but also the strong degree to which the effects of polyploidy depend on ecological context.
Methods
Literature search. We searched for relevant publications using Google Scholar, using individual searches for studies comparing diploids and various cytotypes. These searches returned a total of 1,602 publications, which, after pruning for papers that met our criteria, produced 73 studies able to be analyzed. These were combined with 55 papers found outside our systematic search, for a grand total of 128 publications with 214 individual effect sizes.
Data extraction. For each effect size, we recorded the following information: mean, standard deviation, sample size, and for both the diploid and polyploid groups in each effect size: authors, Linnean family, ploidy level of the polyploid, whether or not the diploids under study were hybrids, whether the polyploids under study were autopolyploids or allopolyploids, whether the species under study were wild or cultivated, whether the polyploids under study were synthetic or established, the type of pathogen with which plants were infected, and the effect direction (i.e., whether a higher value indicates greater or lesser pathogen resistance).
Analyses. Effect sizes were calculated using standardized mean difference with Hedges's g, employing the R package escalc in R versions 4.0.3 and 4.2.1. We analyzed the data using multi-level meta-analytic models constructed with the rma and rma.mv functions from the package metafor. We included four random effects in the rma.mv model: average infection time in days before disease incidence was calculated, the between-study effect, the within-study effect, and a phylo variable calculated using a family-level phylogeny of angiosperms. We examined the statistical heterogeneity explained by each random effect using the I2 statistic, and we tested for publication bias using funnel plot analysis, Egger's regression, and fail-safe numbers.
Usage notes
- R is required to run the code in Hagen_Mason_2023_code.R
- The data in HagenandMason2023_data.xlsx can be opened in Microsoft Excel or an open-source alternative like LibreOffice Calc
- The bibliography of papers included in the meta-analysis, included_papers_biblio.docx, can be opened in Microsoft Word or an open-source alternative like LibreOffice Writer