Data and code from: The adaptive significance of polyandry: A meta-analysis
Data files
Feb 19, 2026 version files 264.34 KB
-
cleaning_data.R
1.61 KB
-
fecund_analysis.R
23.58 KB
-
fecund_data.csv
42.04 KB
-
fecund_tree.Rdata
3.19 KB
-
global_model_results.csv
101 B
-
long_data.csv
27.31 KB
-
long_tree.Rdata
2.36 KB
-
longevity_analysis.R
15.76 KB
-
phylo_cor_fecund.Rdata
3.87 KB
-
phylo_cor_long.Rdata
2.59 KB
-
phylogenetic_analyses.R
8.92 KB
-
polyandry_meta_full_dataset.csv
50.83 KB
-
potentially_eligible_papers.xlsx
75.23 KB
-
README.md
6.94 KB
Abstract
Polyandry is prevalent, but the optimal patterns of mating for females remain poorly understood despite their importance for our understanding of sexual selection. We therefore performed a meta-analysis on the female fitness consequences of mating with multiple males. Across 166 studies spanning 127 arthropod species, we found that mating with more males generally enhanced female fecundity and reduced female lifespan. The net fitness effects of polyandry, however, were small. Moreover, fecundity benefits were not clearly detectable when females mated with more than two males. Additionally, we found first that studies assessing partial as opposed to lifetime fitness reported greater benefits of polyandry. Second, protocols involving selection bias, where females were afforded some control over mating rates, reported lower costs of polyandry compared to studies without selection bias. Third, polyandry was reported as less beneficial in experiments that involved continuous housing of females with males. Finally, polyandry was more beneficial in species that transfer nuptial gifts. We thus suggest that future polyandry studies aim to measure lifetime fitness while also mitigating selection bias and exposure of females to excessive harassment. Doing so will help us understand how sexual selection operates in both sexes.
The R files contain the code used to produce the figures and analyses
- cleaning_data.R - separates the full dataset into fecundity and longevity data. It also contains code that generated the global model results figure
- fecund_analysis.R - contains code used to produce the fecundity analyses and figures
- longevity_analysis.R - contains code used to produce the fecundity analyses and figures
- phylogenetic_analyses.R - contains code used to produce the phylogenetic trees
R objects
- fecund_tree.Rdata - Stores the phylogenetic tree containing all species included in the fecundity analyses
- phylo_cor_fecund.Rdata - Stores the phylogenetic correlation matrix used in the fecundity analyses
- long_tree.Rdata - Stores the phylogenetic tree containing all species included in the longevity analyses
- phylo_cor_long.Rdata - Stores the phylogenetic correlation matrix used in the longevity analyses
Data files
- polyandry_meta_full_dataset.csv - the full dataset of effect sizes included in the meta-analyses
- fecund_data.csv - the subset of the full dataset that includes data on fecundity (eggs or offspring produced); this file was used for all fecundity analyses
- long_data.csv - the subset of the full dataset that includes data on longevity; this file was used for all longevity analyses
- global_model_results.csv - file that stores the results from the overall models looking at global effect sizes for fecundity and longevity; this data was used to generate Figure 1B|
- potentially_eligible_papers.xlsx - spreadsheet used to keep track of the papers we screened in detail for inclusion; the full list of included papers, along with reasons for exclusions, can be found here
Variables explained
Note: The fecund.csv and long_data.csv are subsets of the polyandry_meta_full_dataset.csv file. Therefore, the variables defined below mean the same thing across the three datasets. In these files, NAs mean "not available", which indicate when specific data were not reported by a study.
- source - the study that an effect size was extracted from
- species - species used to generate effect size
- order - taxonomic order of the species
- harass? - whether the effect size was derived from an experiment that involved continuous housing of males with females; binary categorical variable where y = yes and n = no
- nup_gift? - whether the species used to generate the effect size transfers nuptial gifts; binary categorical variable where y = yes and n = no
- lifetime? - whether the study assessed lifetime reproductive success (as opposed to partial lifetime reproductive success); binary categorical variable where y = yes and n = no
- bias? - whether the study involved selection bias; binary categorical variable where y = yes and n = no
- treatment - the degree of polyandry experienced by females being compared; the lower mating rate is always listed first, followed by an underscore and then the higher mating rate
- experiment - used to distinguish between multiple experiments within the same published study
- study - a number assigned to each distinct published study included in the meta-analysis
- con_N - sample size for females in the control group (the treatment that mated with fewer males)
- exp_N - sample size for females in the experimental group (the treatment that mated with more males)
- adjusted_con_N - adjusted control sample size to take into account shared control non-independence (when multiple effect sizes used the same control group); this value comes from dividing the original sample size with the number of times this control group was used as a reference
- con_long - lifespan of females in the control group (typically in days)
- con_long_SD - standard deviation for lifespan of females in the control group
- exp_long - lifespan of females in the experimental group (typically in days)
- exp_long_SD - standard deviation for lifespan of females in the experimental group
- con_fecund - fecundity (number of eggs laid) of females in the control group
- con_fecund_SD - standard deviation for fecundity of females in the control group
- exp_fecund - fecundity (number of eggs laid) of females in the experimental group
- exp_fecund_SD - standard deviation for fecundity of females in the experimental group
- con_fert - fertility (number of offspring produced) of females in the control group
- con_fert_SD - standard deviation for fertility of females in the control group
- exp_fert - fertility (number of offspring produced) of females in the experimental group
- exp_fert_SD - standard deviation for fertility of females in the experimental group
Variables in the global_model_results.csv file. In general, this file stores the results from our global effect size model which can be found in the R files. We created this csv file only for the purpose of data visualization.
- fitness_measure: categorical variable noting whether the mean effect size is for fecundity or longevity
- estimate: the estimated average effect size (log response ratio) for the effect of mating with a relatively greater number of males
- ci.lb: Lower bound of the 95% confidence interval for each estimated average effect size
- ci.ub: Upper bound of the 95% confidence interval for each estimated average effect size
potentially_eligible_papers.xlsx file explained:
Scan 1 (first sheet): Includes the list of 424 potentially eligible papers that were read in depth during stage two of our screening process (except for the two known studies from our research group: Yan et al. 2024 and Yan et al. 2025). Red-highlighted rows represent excluded studies, while green-highlighted rows represent included papers.
- Source - Denotes which database identified the study (Google Scholar, Web of Science (WOS), or the prior polyandry meta-analyses)
- Authors - Authors of the published study
- Screened by - Denotes which of this current study's author screened the study for inclusion
- Full citation - Full reference for the study
Scan 2 (excluded papers) (second sheet): List of the 256 potentially eligible papers that were excluded from the meta-analysis. The "Reference" column denotes the study, while the "Reason for exclusion" column provides a short explanation for why the study was not included.
Final paper list (third sheet): Includes the full list of the 166 studies that were eventually included in the meta-analysis. The studies are listed in alphabetical order.
Code/software
All of the analyses were conducted using R. The packages used and what each package was used for can be found at the beginning of each R script.