Polyploid clades tend to contain more species than their non-polyploid sister clades
Clades with polyploid taxa that are inferred to have undergone genome duplications at or close to their originations contain, on average, significantly more species than sister clades in which such duplication events did not occur. This striking difference was found across nearly all major clades and in nearly every major subgroup of organisms tested. This strongly implies that polyploidy is an important evolutionary cause of diversification18. There have been a small number of studies investigating the association between ploidy level and diversity in specific plant clades 30,58,59. Although these investigations all identified an association between polyploidy and higher taxonomic diversity, it has been unclear whether this results from enhanced evolvability or higher speciation rates in plant clades or whether the effect is driven by reproductive isolation in the immediate aftermath of the WGD events themselves. The taxonomic generality and strength of the effect observed here suggests that the role of WGD events as a cause of diversity may have been underestimated 19. Rather than being prone to rapid extinction 60,61, or leading to evolutionary dead ends 62,63, polyploids appear to propagate persistent lineages that subsequently diversify more readily than non-polyploid close relatives.
Polyploidy is a widespread correlate of species richness in plants
Polyploidy is extremely common in living plants, particularly angiosperms 26 and ferns 27, with many groups showing successive rounds of genome duplication in their history 28. Perhaps 15% of angiosperm and 31% of fern speciation is directly linked to a ploidy increase 29–31. Within the monocotyledons, grasses are highly polyploid 32 and have informed studies of model polyploid genomes 33,34. Polyploidy is also common in ‘true’ dicotyledon groups such as the Brassicaceae 35, Fabaceae 36, Violaceae 37 and Orchidaceae 38.
Despite a long history of polyploidy research in plants 64–66, the prevalence of diploid taxa 67,68 meant that polyploidy was not widely considered as a possible cause of diversity. This assumption can now be challenged 69 and studies of polyploidy have received renewed interest 70. The composition of our dataset suggests that polyploidy is particularly common in derived angiosperms (particularly eudicots) and ferns but much rarer in gymnosperms and more ‘basal’ angiosperms (ANA clade plus magnoliids and Chloranthales). Prior studies have stressed the link between polyploidy and speciation in angiosperms 71, with most major angiosperm clades inferred to have polyploid ancestry 72. While genomic changes such as WGDs have been found to correlate with the origins of major gymnosperm clades as well as the root of gymnosperms itself, more recent shifts in gymnosperm diversification seem to be decoupled from putative WGDs73. We were able to identify few documented instances of natural polyploids in basal angiosperms and gymnosperms, either because polyploidy is genuinely rare 74, or because any gene duplications were sufficiently ancient to obfuscate its signal 21. Polyploidy appears to be extremely common in Magnolia 75 but has not been investigated in related magnoliids and Chloranthales. Hence, while polyploid magnoliids are more species rich than their non-polyploid sister taxon, the sample size (9 clade pairs) is small and the difference is not significant. Despite being common, polyploidy in ferns has only recently received systematic attention 76,77, and all of our polyploid fern clades were genera. The absence of clear patterns in ploidy level at higher taxonomic levels could be one reason why fern polyploidy has received relatively little study.
Polyploidy is also a widespread correlate of species richness in animals
Although thought to be rare in most animals 30, polyploidy is being identified in an increasing number of amphibians 39,40 and fishes 41,42. It is well documented in teleosts, especially salmonids 43, catfish 44 and carp 45. Invertebrate examples include insects 46, crustaceans 47 and molluscs 48.
The evolutionary significance of polyploidy in animals is contentious and discussion has largely focused on ancient Whole Genome Duplication events (WGDs). For example, two WGDs have been proposed close to the origin of vertebrates 24 and another at the origin of bony fishes 78 which, it has been suggested, helped facilitate the evolution of novel morphologies and more complex phenotypes 79. The inclusion of fossil data reduces support for this close temporal association in favour of more protracted evolutionary consequences 25. There are very few living basal vertebrates and gnathostomes and these have very long evolutionary branches separating them from other vertebrate groups and from their last common ancestors with other clades, making it difficult to assess the impact of polyploidy in these cases. However, at least two orders of ray-finned fish (Acipenseriformes and Salmoniformes) appear to be entirely polyploid, suggesting that more recent genome duplications have also played an important role in the evolution of teleosts 42. Polyploid families are also common within Siluriformes (catfish) and Cypriniformes (carp and minnows), while polyploid genera are found within most teleost groups. Paralogs of developmentally relevant genes often appear to be conserved in teleosts 80 and at least some of them are associated with physiological traits unique to this group, including pigmentation types and colour patterns 81. Polyploidy seems to be extremely rare in chondrichthyans, possibly because inferred substitution rates are much lower in cartilaginous fish than in other vertebrates 82, and particularly in comparison with the fast rates in teleosts 83.
The large majority of other polyploid vertebrate clades are amphibians, with some polyploids also documented in reptiles. Ploidy levels in these groups appear less evolutionarily conserved than in teleosts, with polyploid and non-polyploid taxa often being closely related. As a result, all but one of the clade pairs analysed were genera, albeit spread broadly across the major groups of lissamphibians and squamates. Although polyploidy has been recognised as common in particular groups of anurans, including the clawed frogs 84, it has only been linked to speciation in a few cases 85,86. There are also documented cases of polyploidy in birds and mammals, although these are far more tentative and controversial. ZZW triploidy is known to occur in the domestic chicken, although embryo mortality is extremely high 87, as well as in the Blue-and-Yellow Macaw 88. Within mammals, the mountain vizcacha rat has been characterised as an allotetraploid with a hybrid origin 89 although this characterisation is questionable 90.
Polyploidy in vertebrates has been studied in greater detail than in other animals 91. Despite this, we were able to find numerous documented cases of polyploid invertebrates, although just over half of our documented clades were insects (32 out of 62). Despite containing well over 5.5 million species1, the number of polyploid insect clades is thought to be less than 100 46. We suspect that this discrepancy may result from a disproportionate focus on insects in the literature, which are appealing for three reasons. Firstly, the segmented bodyplan of arthropods make them the ideal model for studying the role of Hox gene changes and duplications in determining the expression of morphological traits such as limb identity 92,93. Secondly, the fruit fly Drosophila is the most widely used model for studies of genetic evolution and expression in animals. Thirdly, insects have extraordinarily high diversity and are major components of nearly all terrestrial ecosystems despite their relatively conserved bodyplan. The evolution of novel Hox gene regions is thought to have facilitated morphological diversification in arthropods generally and insects in particular 94,95. While it is possible that polyploidy is associated with diversification of gene regulatory networks, such as Hox genes in insects, we cannot rule out the possibility that polyploidy is simply more poorly documented in other invertebrate groups.
While the majority of animal groups in our sample showed significantly higher diversity in polyploid clades, reptiles, and annelid worms did not. However, sample sizes for these two groups were small, and in each case polyploid clades still contained more species on average. We think it likely that our inability to detect differences in these clades reflect their under-representation in the literature documenting polyploidy. Studies of oligochaete annelids have suggested little correlation between genome size and life history traits, with the possible exception of parthenogenesis in highly polyploid earthworms 96. Polyploidy may be rare in annelids with the exception of some highly polyploid genera 97 but there are no comprehensive assessments of polyploidy in the phylum. Polyploidy in reptiles appears entirely restricted to the squamates, although highly labile and variable within that group. Viable squamate polyploids appear to be exclusively triploids 91, likely arising through the mating of diploid hybrids with sexual lineages 98. There is little evidence for strong ecological 99, or phenotypic 100 differences between parthenogenetic triploids and sexual diploids in the few squamate taxa where polyploidy has been studied in detail.
Polyploid clades at multiple taxonomic levels are more speciose than their non-polyploid sisters
Clades inferred to have undergone genome duplications close to their origins show significantly higher species richness at multiple taxonomic levels. Separate analyses of genera and higher clades both reveal a similar pattern, as do analyses of families considered alone. This makes it unlikely that our results are strongly influenced by the scale at which clades are sampled, as well as by any non-independence of nested clades.
Documented cases of polyploid groups are far more common at the level of genera (321 out of 356 clade pairs) than at higher taxonomic levels. As closely related taxa are, typically, more similar genetically, it may be unsurprising to observe that polyploidy occurs convergently numerous times within the same clade. Ultimately, polyploidy first arises within small populations. Large clades diversifying after genome duplications will be harder to detect as the signal of the ploidy event is overwritten by the subsequent loss of genes and mutation in the retained genes 101. While genome duplications are therefore inevitably most common at lower taxonomic levels, the effect may be exaggerated due to our limited ability to detect such events in large clades.
Our study provides empirical evidence that genome duplications correlate with greater diversification across the tree of life. Most previous attempts to investigate the distribution and variation in polyploidy have documented its occurrence at low taxonomic levels in specific groups of organisms 102,103, while consideration of its evolutionary significance at the macroevolutionary scale has focused on ancient whole genome duplications 20,41. While this study presents compelling evidence of a more general relationship between polyploidy and diversity, species richness is a simplistic measure of diversity that fails to capture many aspects of ecological and morphological variation. Defining clades with ancestral genome duplication events using the literature on extant polyploids lends itself well to studying a broad taxonomic sample. However, it ignores the potentially complex history of genetic changes in different lineages within clades. Further work relating changes in the genome with rates and amounts of morphological evolution may help to explain the ubiquitous pattern seen in this study.
Polyploidy is often implicated in the diversification and evolution of key traits and is associated with a number of physiological effects 104. Polyploidy can alter patterns of gene expression which lead to increases in cell size and hence tissue growth 105 as well as altering cell architecture and regulatory mechanisms 106. Direct, simple changes in traits like body size are rare 107, as in most cases developmental mechanisms regulate, eliminate or modify these effects 108. Instead polyploidy often affects gene expression in complex ways to produce a broad range of ‘hybrid vigour’ effects 109, including downregulation as well as the upregulation of genes 110. While polyploidy can introduce errors into cell division that increase the likelihood of sterility 111 and other deleterious consequences 112,113, certain groups (e.g. plants) have genetic and developmental architecture that mitigate these effects 114. In plants, sterility is often countered by a greatly increased capacity for selfing 115, allowing many polyploids to reproduce asexually by parthenogenesis 116 which can also result in instant reproductive isolation and hence speciation 117. However, polyploidy may also cause a failure of self-incompatibility mechanisms which allow traits such as dioecy to evolve118
The developmental trajectories of many groups have become more complex through time 119-122. This is because the addition of new developmental stages is constrained by existing developmental pathways 123,124 and also because genes, developmental pathways, tissues and organs often acquire a greater number of increasingly integrated functions, and thus pleiotropy, over time 125. The accumulation of pleiotropic effects makes it increasingly likely that mutations will deleteriously affect at least one process or pathway as genes acquire multiple functions 126. Given increasing complexity, one might therefore expect the evolutionary flexibility of organisms to decline over time as bodyplan development becomes more canalised 127,128. Canalisation may help to explain how closely related organisms sharing the same genetic architecture evolved the same phenotypic traits in parallel 129-131 and may be relatively common even at the more fundamental genetic level, constraining change along ‘lines of least resistance’ 132. Genome duplications may offer a mechanism by which lineages can circumvent the worst effects of pleiotropy133, freeing up copies of genes that can – albeit very occasionally – acquire new functions and thereby increase evolvability134-136.