In flowering plants, shifts from outcrossing to partial or complete self-fertilization have occurred independently thousands of times, yet the underlying adaptive processes are difficult to discern. Selfing’s ability to provide reproductive assurance when pollination is uncertain is an oft-cited ecological explanation for its evolution, but this benefit may be outweighed by genetic costs diminishing its selective advantage over outcrossing. We directly studied the fitness effects of a self-compatibility (SC) mutation that was backcrossed into a self-incompatible (SI) population of Leavenworthia alabamica, illuminating the direction and magnitude of selection on the mating-system modifier. In array experiments conducted in two years, SC plants produced 17-26% more seed, but this advantage was counteracted by extensive seed discounting -- the replacement of high-quality outcrossed seeds by selfed seeds. Using a simple model and simulations, we demonstrate that SC mutations with these attributes rarely spread to high frequency in natural populations, unless inbreeding depression falls below a threshold value (0.57 ≤ threshold ≤ 0.70) in SI populations. A combination of heavy seed discounting and moderate inbreeding depression likely explains why outcrossing adaptations such as self-incompatibility are maintained generally, despite persistent input of selfing mutations and frequent limits on outcross seed production in nature.
2014+2015 Field data
Data on maternal fitness taken from plants in experimental arrays in 2014 and 2015. Plants are coded by their family of origin and S-locus genotype (0=SC, 1=SI).
2014+2015 Greenhouse data
Floral trait data measured on plants in controlled glasshouses in 2014 and 2015. Plants are coded by family of origin and S-locus genotype. Forced self pollinations permitted measures of self-compatibility, and estimates of autonomous fruit set are also included.
Emasculation experiment
This file contains data on fruit and seed set of emasculated flowers visited by a single pollinator. This experiment was conducted in spring 2015 in site 1 (the Waco glade).
Pollinator Data 2014
This file contains information on pollinators and their sequence of visits for plants in experimental arrays (2014 only). Plants in the top row were given an index of 1-8 (reading from left to right). Plants in the second, third, and fourth rows were given an index of 9-16, 17-24, and 25-32, respectively.The first tab has information on the first day each plant flowered. Pollinator codes were:Ant = ant; Beetle = beetle; B = honeybee or carpenter bee; L = lepidoptera; F = fly (excluding bee-fly); BF = bee-fly; AH = andrenidae/halictidae; K = katydid
U = unknown.
Pollinator Data 2015
This file contains information on pollinators and their sequence of visits for plants in experimental arrays (2015 only). Plants in the top row were given an index of 1-8 (reading from left to right). Plants in the second, third, and fourth rows were given an index of 9-16, 17-24, and 25-32, respectively.The first tab has information on the first day each plant flowered. Pollinator codes were:Ant = ant; Beetle = beetle; B = honeybee or carpenter bee; L = lepidoptera; F = fly (excluding bee-fly); BF = bee-fly; AH = andrenidae/halictidae; K = katydid
U = unknown.
S-locus Genotyping 2014
Information on the S-locus genotype of plants, assayed with SSCP (single-strand conformation polymorphism). Within each family, we adopted a convention for scoring the 4 possible genotypes. A = Lal2-a2/Lal2-a2 homozygote (SC plant); D = Si/Sj heterozygote (SI plant). B and C denote plants that are heterozygous for the Lal2-a2 mutation (these plants were not used).
S-locus genotyping 2015
Information on the S-locus genotype of plants, assayed with SSCP (single-strand conformation polymorphism). Within each family, we adopted a convention for scoring the 4 possible genotypes. A = Lal2-a2/Lal2-a2 homozygote (SC plant); D = Si/Sj heterozygote (SI plant). B and C denote plants that are heterozygous for the Lal2-a2 mutation (these plants were not used). We also post here the total numbers of A,B,C, and D genotypes in each family in 2014 and 2015.
Fragment Analysis 2014
Progeny array data for plants in the 2014 experimental arrays, in MLTR format. Plants were genotyped at 6 hypervariable microsatellite markers. Maternal plants are denoted with (!). These data were also used for paternity inference, following conversion to the Cervus file format.
Fragment Analysis 2015
Progeny array data for plants in the 2015 experimental arrays, in MLTR format. Plants were genotyped at 6 hypervariable microsatellite markers. Maternal plants are denoted with (!). These data were also used for paternity inference, following conversion to the Cervus file format.
main.cpp
C++ code that simulates the S-locus and 1000 unlinked loci causing inbreeding depression. Self-incompatible populations reach a mutation-selection-drift equilibrium, whereupon a SC mutation invades. The final frequency of the SC mutation is then tabulated.
Individual.h
This file contains functions called by the C++ code, main.cpp.
Locus
This file contains functions called by the C++ code, main.cpp.