Hybridization can instantaneously generate novel genetic variation, which can promote ecological speciation by creating novel adaptive phenotypes. However, it remains unclear how hybridization, creating novel mating phenotypes (e.g., mating season, genitalia shapes, sexual displays, mate preferences), affects speciation especially when the phenotypes do not confer adaptive advantages. Here, based on individual-based evolutionary simulations, we propose that transgressive segregation of mating traits can drive incipient hybrid speciation. Simulations demonstrated that incipient hybrid speciation occurred most frequently when the hybrid population received moderate continued immigration from parental lineages causing recurrent episodes of hybridization. Recurrent hybridization constantly generated genetic variation, which promoted the rapid stochastic evolution of mating phenotypes in a hybrid population. The stochastic evolution continued until a novel mating phenotype came to dominate the hybrid population, which reproductively isolates the hybrid population from parental lineages. However, too frequent hybridization rather hindered the evolution of reproductive isolation by inflating the variation of mating phenotypes to produce phenotypes allowing mating with parental lineages. Simulations also revealed conditions for long-term persistence of hybrid species after their incipient emergence. Our results suggest that recurrent transgressive segregation of mating phenotypes can offer a plausible explanation for hybrid speciation and radiations that involved little adaptive ecological divergence.

Java programs for individual-based computer simulations, R codes to analyze and visualize simulation results, and final output files of the computer simulations.

Codes and data for “Transgressive segregation in mating traits drives hybrid speciation” by Kotaro Kagawa, Gaku Takimoto & Ole Seehausen.

===========================================

Reference

Kagawa K, Takimoto G, & Seehausen O. (2023). “Transgressive segregation in mating traits drives hybrid speciation”, Evolution, (https://doi.org/10.1093/evolut/qpad072)

==========================================

Code Overview

There are four main directories: “Java_IBM_Matching”, “Java_IBM_PreferenceTrait”, “R_Scripts”, and “Appendix_configs_Figures”. Directories “Java_IBM_Matching” and “Java_IBM_PreferenceTrait” contain Java source cords for our individual-based evolutionary simulations with the matching model and preference/trait model, respectively. The directory “R_Scripts” contains R scripts for analyzing and visualizing simulation results. The directory “Appendix_configs_Figures” contains 2 subdirectories including 48 sub-subdirectories, each of which contains four input files for running simulations shown in figures of the paper and its SI. 

Codes_Kagawa_et_al_2023_Evolution.zip

├── Java_IBM_Matching

│   ├── configs # four input files for simulation.

│   │   ├── basal_parameter_set.txt # list of the default parameter values.

│   │   ├── parameters_AH.txt # list of parameter values to be explored (for hybridization).

│   │   ├── parameters_BH.txt # list of parameter values to be explored (for allopatric phase).

│   │   └── simulation_mode.txt # the scenario to be simulated.

│   └── src # java source codes for the individual-based simulation of the matching model.

│       ├── MersenneTwisterFast.java # random number generator*.

│       ├── Run_AfterHybrid.java # code to run simulations of evolution by hybridization.

│       ├── Run_BeforeHybrid.java # code to run simulations of the allopatric period.

│       └── Simulation.java # core program of the individual-based model (matching model).

├── Java_IBM_PreferenceTrait

│   ├── configs # four input files for simulation.

│   │   ├── basal_parameter_set.txt # list of the default parameter values.

│   │   ├── parameters_AH.txt # list of parameter values to be explored (for hybridization).

│   │   ├── parameters_BH.txt # list of parameter values to be explored (for allopatric phase).

│   │   └── simulation_mode.txt # the scenario to be simulated.

│   └── src # java source codes for the individual-based simulation of the preference/trait model.

│       ├── MersenneTwisterFast.java # random number generator (*).

│       ├── Run_AfterHybrid.java # code to run simulations of evolution by hybridization.

│       ├── Run_BeforeHybrid.java # code to run simulations of the allopatric period.

│       └── Simulation.java # core program of the individual-based model (preference/trait model).

├── R_Scripts # scripts for analyzing raw simulation outputs and for visualizing results.

│   ├── Analysis_DynamicsReplicates.R # to analyze simulations under the same condition.

│   ├── Analysis_ParameterDependence.R # to analyze simulations under various conditions.

│   ├── Plot_DynamicsExample.R # to visualize evolutionary dynamics in a simulation.

│   ├── Plot_DynamicsReplicates.R # to visualize evolutionary dynamics in many simulations.

│   ├── Plot_ParameterDependence.R # to visualize effects of parameters on evolutionary outcomes.

│   ├── Plot_FigS2.R # script to generate Fig. S2

│   ├── Plot_FigS5.R # script to generate Fig. S5

│   └── Plot_FigS17.R # script to generate Fig. S17

└── Appendix_configs_Figures # input files (the directory “config”) for running simulation for each figure.

    ├── MatchingModel # input files for the Matching model (“Java_IBM_Matching”)

    │   ├── configs_Fig3a_One-timeHybridization # input files for simulations for a part of Fig. 3a

    │   ├── configs_Fig3a_S4a # input files for simulations for the main part of Figs. 3a, S4a

    │   ├── configs_Fig4a_5_a_c # input files for simulations for Figs. 4a, 5a, c

    │   ├── configs_Fig4c # input files for simulations for Fig. 4c

    │   ├── configs_FigS10a_c # input files for simulations for Fig. S10a, c

    │   ├── configs_FigS10b # input files for simulations for Fig. S10b

    │   ├── configs_FigS11a # input files for simulations for Fig. S11a

    │   ├── configs_FigS12a # input files for simulations for Fig. S12a

    │   ├── configs_FigS13a # input files for simulations for Fig. S13a

    │   ├── configs_FigS13b # input files for simulations for Fig. S13b

    │   ├── configs_FigS14a # input files for simulations for Fig. S14a

    │   ├── configs_FigS14c # input files for simulations for Fig. S14c

    │   ├── configs_FigS15 # input files for simulations for Fig. S15

    │   ├── configs_FigS16a # input files for simulations for Fig. S16a

    │   ├── configs_FigS16b # input files for simulations for Fig. S16b

    │   ├── configs_FigS16c # input files for simulations for Fig. S16c

    │   ├── configs_FigS17a_b # input files for simulations for Fig. S17a, b

    │   ├── configs_FigS18a # input files for simulations for Fig. S18a

    │   ├── configs_FigS3a_c_d # input files for simulations for Fig. S3a, c, d

    │   ├── configs_FigS3b # input files for simulations for Fig. S3b

    │   ├── configs_FigS7a # input files for simulations for Fig. S7a

    │   ├── configs_FigS8a # input files for simulations for the main part of Fig. S8a

    │   ├── configs_FigS8a_One-timeHybridization # input files for simulations for a part of Fig. S8a

    │   ├── configs_FigS9a # input files for simulations for Figs. 10b

    │   └── configs_FigS9b # input files for simulations for Figs. 10b

    └── PreferenceTraitModel # input files for the Matching model (“Java_IBM_PreferenceTrait”)

        ├── configs_Fig3b_One-timeHybridization # input files for simulations for a part of Fig. 3b

        ├── configs_Fig3b_S4c # input files for simulations for the main part of Figs. 3b, S4c

        ├── configs_Fig4b_5b_d # input files for simulations for Figs. 4b, 5b, d

        ├── configs_Fig4d # input files for simulations for Fig. 4d

        ├── configs_FigS10d_f # input files for simulations for Fig. S10d, f

        ├── configs_FigS10e # input files for simulations for Fig. S10e

        ├── configs_FigS11b # input files for simulations for Fig. S11b

        ├── configs_FigS12b # input files for simulations for Fig. S12b

        ├── configs_FigS13c # input files for simulations for Fig. S13c

        ├── configs_FigS13d # input files for simulations for Fig. S13d

        ├── configs_FigS14b # input files for simulations for Fig. S14b

        ├── configs_FigS14d # input files for simulations for Fig. S14d

        ├── configs_FigS16d # input files for simulations for Fig. S16d

        ├── configs_FigS17c_d # input files for simulations for Fig. S17c. d

        ├── configs_FigS18b # input files for simulations for Fig. S18b

        ├── configs_FigS2c # input files for simulations for Fig. S2c

        ├── configs_FigS3e_g_h # input files for simulations for Fig. S3e, g, h

        ├── configs_FigS3f # input files for simulations for Fig. S3f

        ├── configs_FigS7b # input files for simulations for Fig. S7b

        ├── configs_FigS8b # input files for simulations for the main part of Fig. S8b

        ├── configs_FigS8b_One-timeHybridization # input files for simulations for a part of Fig. S8b

        ├── configs_FigS9c # input files for simulations for Fig. S9c

        └── configs_FigS9d # input files for simulations for Fig. S9d

* “MersenneTwisterFast.java” is the java source code of the Mersenne-Twister-random-number-generator implemented by Dr. Sean Luke (2004), which is distributed at the following website: https://cs.gmu.edu/~sean/research/mersenne/MersenneTwisterFast.java

Data Overview

This dataset contains two main directories: “RawSimulationOutputExamples” and “SimulationResults_For_Figures”. The directory “RawSimulationOutputExamples” contains raw output files (6140 files in total) from simulations under a small subset of parameter conditions that we explored. These files are intermediate files to be processed by R codes and are not necessary for reproducing our study, but they are useful for checking how our Java programs and R scripts work. The directory “SimulationResults_For_Figures” contains source data for figures, which were generated by analyzing raw output files of simulations.

 

Data_Kagawa_et_al_2023_Evolution.zip

├── RawSimulationOutputExamples # examples of raw simulation outputs for testing codes.

│   ├── Example_raw_output_MatchingModel # results with the matching model.

│   │   ├── Allopatric_Matching (cont.545 files) # results from simulations of allopatric phase.

│   │   └── SecondaryContact_Matching (cont.2524 files) # results from simulations of hybridization.

│   └── Example_raw_output_PreferenceTraitModel # results with the preference/trait model.

│       ├── Allopatric_PrefTrait (cont.545 files) # results from simulations of allopatric phase.

│       └── SecondaryContact_PrefTrait (cont.2524 files) # results from simulations of hybridization.

└── SimulationResults_For_Figures

    └── Simulation_Final_Results # source data for figures (input files for R-scripts starting with “Plot_”)

├── DynamicsExamples (cont.10files) # data for Figs. 2, 6, 18

├── DynamicsReplicates (cont.8files) # data for Figs. 4, 5

        ├── ParameterDependency (cont.40 files) # data for Figs. 3, S3, S7-S16

        ├── FigS2 (cont.4 files) # data for Fig. S2

        ├── FigS5 (cont.2 files) # data for Fig. S5

        └── FigS17 (cont.6 files) # data for Fig. S17

 

==========================================

Instructions for running Java programs for individual-based simulation

Step 1: Compilation of java source codes

Our individual-based simulation models are implemented with Java language. Before running simulations, all java source codes files (“Simulation.java”, “Run_AfterHybrid.java”, “Run_BeforeHybrid.java”, and “MersenneTwisterFast.java”) have to be compiled with the Open Java Development Kit (Open JDK) or Java Development Kit (JDK) with the following command: “javac *.java”. 

Step 2: Running simulations of the allopatric phase with the default condition

After compilation, simulations of the allopatric evolution of two parental lineages before they go into a secondary contact (i.e., the “allopatric phase”) can be run with Open JDK by the following command: “java Run_BeforeHybrid”. Raw output files of simulations will be stored in directories “Allopatric_Matching” (or “Allopatric_PrefTrait” in the case of preference/trait model), which will be automatically generated by the program. 

Step 3: Running simulations of the evolution following hybridization with the default condition

Simulation of the evolutionary dynamics following a secondary contact and hybridization between parental lineages can be run with the following command: “java Run_AfterHybrid”. This simulation requires raw output files from simulations of the allopatric phase; the simulation program automatically extract data from the directory “Allopatric_Matching” (“Allopatric_PrefTrait”) to construct the initial condition for simulating a secondary contact between two allopatrically diverged lineages. The directory “RawSimulationOutputExamples” contains some examples of raw output files from simulations of the allopatric phase. Raw output files from simulations of the evolution following hybridization will be stored in directories “SecondaryContact_Matching” or “SecondaryContact_PrefTrait”, which will be automatically generated by the program. 

Step 4: Running simulations with various conditions

Conditions for the simulation including parameter values, the presence/absence of continuing parental immigration to the hybrid population, and the shape of fitness landscape are controlled by text files in the directory “configs” (for details of parameters and simulation scenarios, please see the main text and SI of the paper). Simulation programs automatically extract data from input files in this directory. Therefore, simulations with varied conditions can be run by modifying files in the directory “configs”. Parameter values for simulation of the allopatric phase are in the file “parameter_table_BH.txt”. When Run_BeforeHybrid.java is executed, the program run simulations with all combinations of parameter values written in this text file (up to two parameters can be varied at once). Parameter values for simulation of the evolution following hybridization are in the file “parameter_table_AH.txt”. When “Run_AfterHybrid.java” is executed, the program run simulations with all combinations of parameter values written in this file for all pairs of parental populations that have been generated with “Run_BeforeHybrid.java” under parameter sets in “parameter_table_BH.txt”. The scenario to be simulated is designated by a number written in the file simulation_mode.txt. Simulations for each figure can be reproduced by using the input files in “configs_Figures_MatchingModel” and “configs_Figures_PreferenceTraitModel”, which are subdirectories of the directory “SimulationResults_For_Figures”. The correspondence relationship between parameters in the paper and those in the Java source codes and R scripts are summarized in the Table 1.

==========================================

Instructions for running R scripts.

Step 1: Visualization of evolutionary dynamics in a single simulation (e.g., Fig. 2)

Evolutionary dynamics in a simulation can be visualized by using the script “Plot_Dynamics.R”. Input file of this script is a raw simulation output file in which properties of all individuals are recorded with a constant time interval (the default name of this file is “population_x_.txt” (x = 0, 1, 2, …)). 

Step 2: Analysis of raw outputs of many simulations with the same parameter condition (e.g., Fig. 4, 5)

The script “Analysis_DynamicsReplicates.R” processes raw output files from a set of many simulation replications under the same parameter condition. For each simulation run, the script calculates the mean and variance of mating trait phenotypes and the strength of reproductive isolation at several timepoints of the simulation. Then, the script creates a file summarizing evolutionary dynamics of mean and variance of phenotypes as well as the strength of reproductive isolation (the default name of the file is “Stats_Integrated_RI.txt”). 

Step 3: Visualization of dynamics in simulations with the same parameter condition (e.g., Fig. 4, 5)

The script “Plot_DynamicsReplicates.R” visualizes evolutionary dynamics in many simulations with the same parameter condition using the final output files of the step 2 as input files.

Step 4: Analysis of raw outputs of many simulations with varied parameter conditions (e.g., Fig. 3)

The script “Analysis_ParameterDependence.R” analyzes raw output files from a set of many simulations with systematically varied parameter values. For each simulation run, the script categorizes evolutionary outcome, quantifies the strength of reproductive isolation, and calculates the degree of genetic differentiation between parental lineages using methods described in the supplementary text S4. Then, the script creates some intermediate files and a final output file, in which parameter values and evolutionary outcome of each simulation are summarized (the default name of the file is “Results_integrated.txt”).

Step 5: Visualization of evolutionary results of simulations with varied parameter conditions (e.g., Fig. 3)

Using final output files of the step 4, the script “Plot_ParameterDependence.R” visualizes evolutionary outcomes of many simulations with systematically varied parameter conditions.  

Step 6: Scripts to generate other supplementary figures

l   Plot_FigS2.R: Using a raw output file of a simulation of the preference/trait model, this script calculates and visualizes the genetic variance and covariance of male and female mating traits (Fig S2).

l   Plot_FigS5.R: This script visualizes the time series of population size, average survival fitness, and average reproductive fitness in a single simulation by extracting data from a raw output file of the simulation.

l   Plot_FigS17.R: This script calculates and visualizes the reproductive fitness landscape in a simulation run (Supplementary Text S6 for details). 

==========================================

Format of raw output files of individual-based simulation with Java programs

The simulation program runs multiple simulations at once. The x-th simulation generates a set of the following raw output files:

(1)   “alleles_x_y_.txt”: The list of all derived alleles that were existing in the system at the generation y. Each row stores the following six properties of each derived allele: the ID of the allele (“mutation_id”), the position of the derived allele in the genome (“position”), the name of population in which the derived allele originated (“origin”), ID of trait that the allele affects (“trait_id”), ID of the genetic locus that the allele belongs to (“locus_id”), the phenotypic effect value of the allele (“pheno_effect”).

(2)   “loci_x.txt”: This files stores positions of all genetic loci that could potentially affect mating trait phenotypes in this simulation run. All simulations starting from the same common ancestor share the same list of positions of loci.

(3)   “individuals_x_y_.txt”: This file stores the list of all haploid genotypes from all individuals at the generation y. Each row stores the information of a haploid genotype as a list of IDs of derived alleles. The first five columns store the following properties of the haploid genome and its carrier (a diploid individual): generation, which should be equal to y, name of the population that the individual belonged to (“population_id”; 0 or 1 -> parental population, 2 -> hybrid zone), sex of the individual (“female”; 0 -> male, 1 -> female), ID of the individual (“id_no”), and the ID of the haploid genome (“chromo_no”; 0 or 1). Following columns list all derived alleles involved in the haploid genome.

(4)   “para_x_.txt”: This file stores all parameter values that have been used for running the simulation.

(5)   “population_x_.txt” This is the main file of raw simulation output, in which properties of all individuals in the system are recorded at a constant interval. Table meaning of each row is shown in the Table 2. Following files follow the same format: all files in directories “DynamicsExamples”, “FigS5”, and “FigS17”; “FigS2a.txt” and “FigS2c.txt” in the directory “FigS2”.

(6)   “population_x_last100_.txt” In this file, properties of all individuals in the system are recorded every generation of the last 100 generations of the simulation. Format of the file is same as “population_x_.txt”.

(7)   “statistics_x_.txt”: In this file, the following six statistics are recorded at a constant time interval: population size at the juvenile stage (“pop_size”), population size at the adult stage (“survivor_size”), number of genetic sites with allele polymorphism (num_snp), number of derived alleles that have fixed in the entire system after starting the simulation (“num_fixed”), the mean value of the mating trait value (“mean_x0”. In the preference/trait model, the mean values of male mating trait and female mate preference are recorded as “mean_x0” and “mean_y0”, respectively), the standard deviation of the mating trait value (“sd_x0”. In the preference/trait model, the mean values of male mating trait and female mate preference are recorded as “sd_x0” and “sd_y0”, respectively). In simulations of the allopatric phase, these statistics are calculated for the parental population. In simulations of the evolution following hybridization, on the other hand, these statistics are calculated for the hybrid population.

(8)   Four files starting from “#” (“#basal_parameter_set.txt”, “#parameters_AH.txt”, “parameters_BH.txt”, and “simulation_mode.txt”) are copies of the four input files in the directory “config” that were used to run the set of simulations (see the “step 4” of the “Instructions for running Java programs for individual-based simulation” section).

==========================================

Format of files summarizing results of many simulations

Our R scripts summarize results from a set of many simulations into single files, which follow either one of the following tree formats:

(1)   “Results_integrated.txt”: Each row of this file summarizes information of single simulation run. The first 23 columns are parameter values used for the simulation (Table 1). The remaining columns are statistics that summarize simulation results. Meanings of these statistics are described in the Table 3. All files in the directory “ParameterDependency” follow this format.

(2)   “Stats_Integrated_RI.txt”: This file summarizes evolutionary dynamics in many simulations of the evolution following hybridization. Statistics that summarize evolutionary state of the hybrid population are provided for each timepoint (the column “generation”) in each simulation (the column “trial_id”). Meaning of each statistic is described in the Table 3. Files “Fig5a_c.txt” and “Fig5b_d.txt” in the directory “DynamicsReplicates” follow this format.

(3)   “mean_trait_dynamics.txt”: This file integrates the column “mean_x0” the file “statistics_x_.txt” of many simulations. Each column of this file except for the first column showing generation shows the dynamics of mean trait value in a single simulation run. In the case of preference/trait model, the dynamics of mean male and female mating trait values are separately stored in “mean_trait_dynamics.txt” and “mean_pref_dynamics.txt”, respectively. The following files follow this format: “Fig4a.txt”, “Fig4b.txt”, “Fig4c.txt”, and “Fig4d.txt” in the directory “DynamicsReplicates” and “FigS2b_mean_trait_dynamics.txt” and “FigS2b_mean_pref_dynamics.txt” in the directory “FigS2”.

Table 1. Correspondence between symbols for parameters and their names in the Java simulation programs, R scripts, and simulation output files.

*For the mating trait in the Matching model and the male trait in the Preference/Trait mode

**For the female trait in the Preference/Trait model

Table 2. Contents of “population_x_.txt” (the main file of raw simulation outputs).

Table 3. Results shown in “Results_integrated.txt” and “Stats_Integrated_RI.txt” (files for generating figures that summarize results of many simulations)

*Supplementary Text S4 for details.