Self-sterility in flowering plants: preventing self-fertilization increases family diversification rates

Abstract

Background and scope: New data are presented on the distribution and frequency of self-sterility (SS) - predominantly pre-zygotic self-incompatibility (SI) systems - in flowering plants and the hypothesis is tested that families with self-sterile taxa have higher net diversification rates (DRs) than those with exclusively self-compatible taxa using both absolute and relative rate tests.

Key results: Three major forms of SI systems (where pollen is rejected at the stigmatic, stylar or ovarian interface) are found to occur in the oldest families of flowering plants, with times of divergence >100 million years before the present (mybp), while post-fertilization SS and heterostyly appear in families with crown ages of 81 and 87 mybp, respectively. It is also founnd that many (22) angiosperm families exhibit >1 SI phenotype and that the distribution of different types of SS does not show strong phylogenetic clustering, collectively suggesting that SS and SI systems have evolved repeatedly de novo in angiosperm history. Families bearing self-sterile taxa have higher absolute DRs using all available calibrations of the angiosperm tree, and this affect is caused mostly by the high DR of families with homomorphic SI systems (in particular stigmatic SI) or those in which multiple SS/SI phenotypes have been observed (polymorphic). Lastly, using sister comparisons, it is further demonstrated that in 29 of 38 sister pairs (including 95 families), the self-sterile sister group had higher species richness and DR than its self-compatible sister based on either the total number of taxa in the clade with SS or only the estimated fraction to harbour SS based on literature surveys.

Conclusions: Collectively, these analyses point to the importance of SS, particularly pre-zygotic SI in the evolution of flowering plants.

Fig. 1.

Types of self-incompatibility in the angiosperms. (A) Heteromorphic: self-incompatibility (SI) may be associated with heterostyly (morphological variation in anther and stigma heights), such that populations have two (dimorphism) or three (trimorphism) floral morphs that exhibit reciprocal anther–stigma heights. Variation in anther–stigma heights is controlled by a group of linked genes that may also be linked to a locus causing genetic SI in flowers of one of the types described in the right panel. (B–E) Homomorphic: sites of expression of self-incompatibility. SI may occur at (B) the stigmatic surface such that a pollen grain (pg) does not germinate, (C) in the stylar region such that a pollen grain does not reach the ovary (ov) or micropyle (mi), (D) in the ovary – such that the pollen tube cannot penetrate the micropyle (mi) or, if it does, fertilization does not occur (ovarian inhibition) or (E) post-fertilization – if following fertilization, the zygote (zy) and/or endosperm (en) fail to develop.

Fig. 2.

The number of families in each of the six major groups of angiosperms: ANITA grade, Magnoliids, monocots + Commelinids, eudicots + core eudicots–Asterids–Rosids, Asterids and Rosids, with different classifications of self-sterility. (A) Number of families in each phylogenetic group classified by their SS/SC status including unclassified families across all 445 families of angiosperms recognized by APG II (Stevens, 2010 onwards). (B) Number of families in each phylogenetic group classified by their SI phenotype for the 78 families for which the SI phenotype data are available. (C) Number of families across each phylogenetic group exhibiting different forms of genetic control of SI for the 27 families for which data are available.

Fig. 3.

Age of divergence (Wikström et al., 2001; MYBP = millions of years before present) of families categorized as having self-compatibility (SC), heteromorphic, or one of the following SS/SI phenotypes: stigmatic, stylar, or ovarian SI, post-fertilization SS, Heteromorphic + SI (Het + SI) and SS unclassified. When polymorphic expression of SS/SI (two or more sites of pollen tube inhibition have been noted in the same family), the age of divergence of the family was plotted multiple times to heuristically visualize the time of appearance of the SS/SI phenotype.

Fig. 4.

Ancestral state reconstruction of self-sterility among 435 angiosperm families of (A) ANITA grade + Magnoliids+ monocots + Commelinids, (B) Eudicots + core Eudicots + Asterids and (C) Rosids sections using the phylogeny and calibration given in Soltis et al. (2011). Families were classified as possessing self-compatibility (SC), stigmatic, stylar, or ovarian SI, post-fertilization SS, or as having heteromorphic and heteromorphic + SI, or as being polymorphic because multiple types of SS/SI have been noted in the family. Families considered to be self-sterile but for which the SI has not been confirmed are classified as possessing self-sterility (SS unclassified), while those for which no breeding information is available are described as unknown. Divided internal nodes with different colours represent equivocal reconstructions in different states optimized by maximum parsimony.

Fig. 4.

Ancestral state reconstruction of self-sterility among 435 angiosperm families of (A) ANITA grade + Magnoliids+ monocots + Commelinids, (B) Eudicots + core Eudicots + Asterids and (C) Rosids sections using the phylogeny and calibration given in Soltis et al. (2011). Families were classified as possessing self-compatibility (SC), stigmatic, stylar, or ovarian SI, post-fertilization SS, or as having heteromorphic and heteromorphic + SI, or as being polymorphic because multiple types of SS/SI have been noted in the family. Families considered to be self-sterile but for which the SI has not been confirmed are classified as possessing self-sterility (SS unclassified), while those for which no breeding information is available are described as unknown. Divided internal nodes with different colours represent equivocal reconstructions in different states optimized by maximum parsimony.

Fig. 4.

Ancestral state reconstruction of self-sterility among 435 angiosperm families of (A) ANITA grade + Magnoliids+ monocots + Commelinids, (B) Eudicots + core Eudicots + Asterids and (C) Rosids sections using the phylogeny and calibration given in Soltis et al. (2011). Families were classified as possessing self-compatibility (SC), stigmatic, stylar, or ovarian SI, post-fertilization SS, or as having heteromorphic and heteromorphic + SI, or as being polymorphic because multiple types of SS/SI have been noted in the family. Families considered to be self-sterile but for which the SI has not been confirmed are classified as possessing self-sterility (SS unclassified), while those for which no breeding information is available are described as unknown. Divided internal nodes with different colours represent equivocal reconstructions in different states optimized by maximum parsimony.

Fig. 5.

Family DR of the Kendall–Moran estimator (rˆ_KM) for 230 families identified as possessing self-compatibility (SC), possessing heteromorphic, or presenting self sterility unclassified (SS unclass) or one of six SI phenotypes: stigmatic, stylar, ovarian, post-fertilization SS, heteromorphic + SI (Het + SI) or as being polymorphic in the site of SI expression. MYBP = millions of years before present.

Fig. 6.

(A) Mean and standard error of the Kendall–Moran estimator of DR (rˆ_KM) in 164 and 66 families bearing SS/SI and SC, respectively, using the estimates of Wikström et al. (2001), Davies et al. (2004) and exponential and log-normal calibrations of Bell et al. (2010) for the divergence age of the families. Mean and standard error of Kendall–Moran estimator of DR (rˆ_KM) using the estimates of (B) Wikström et al. (2001), (C) Davies et al. (2004) and (D) exponential and (E) log-normal calibrations of Bell et al. (2010) for the divergence age of the 44 families bearing homomorphic SI, 19 with heteromorphy, 22 with polymorphic basis to SI, 74 families bearing SS unclassified species and for 66 families bearing SC. Means with different letters were found to be significantly different (Tukey HSD test, P < 0·05).

Fig. 7.

Proportion of the total number of species in the sister-group pair bearing SC (black columns) versus (A) the total number of species in the sister clade with SS/SI (grey columns) or (B) the expected proportion of taxa with SS in the clade with SS/SI (grey columns) for each one of the 38 sister group comparisons inferred from the Soltis et al. (2011) topology (those for other comparisons performed are presented in Supplementary Data).