Data and code for study entitled "Do environmental fluctuations during development affect trait variation? An experimental test with salinity"

We tested how developmental environments (freshwater, stable-saline, fluctuating-saline) influence variation in life-history traits (age at maturity, size at maturity, relative gut length, immunity) and reproductive traits in both sexes – females (egg number, egg size) and males (relative gonopodium length, sperm count, sperm velocity) at both young and old adult stages (before and after 12 weeks of mating).

We have attached the raw data spreadsheet (Raw data.xlsx). For young males, life-history and reproductive traits were measured in different individuals and are provided in separate sheets. For young females, old females, and old males, all traits were measured on the same individuals and are combined in a single sheet for each group. We analysed mean egg size in our models; individual egg size data for young females are also included in the raw data file.

The R code (Trait variability.R) was run using R v1.3.1093. Analyses used data from female.csv and male.csv.

----- Analytical approach -----

We used Bayesian multivariate mixed-effects models to estimate trait covariances and test whether developmental environment affected both trait means and variances, run separately for each age–sex group. Separate models were used for young males' life-history and reproductive traits. Environment (3 levels) was included as a fixed effect, with brood identity as a random effect. Traits including age at maturity, immunity, egg number (young females), and sperm count were power transformed to improve normality. Each model used four MCMC chains (5000 iterations, 1000 burn-in), with convergence confirmed (Rhat = 1) and effective sample sizes >1480.

To test environmental effects on trait variance, we modelled both trait means and standard deviations (SDs) on a log scale. We calculated:

lnVR = ln(SD1/SD2), to assess changes in raw variance

lnCVR = ln(CV1/CV2), to assess changes in relative variance (accounting for trait means)

We compared specific environment pairs to assess:

Effect of salinity: stable salinity vs freshwater

Effect of fluctuations: fluctuating salinity vs stable salinity

我们测试了不同发育环境（淡水、稳定盐度、波动盐度）对雌雄两性在成年早期和晚期（交配12周前后）的生活史性状（life-history traits，包括成熟年龄、成熟体型、相对肠道长度、免疫能力）及繁殖性状（reproductive traits）的变异影响——雌性的繁殖性状为产卵数、卵大小，雄性为相对交接器长度、精子数量、精子活力。

我们附上了原始数据表格（Raw data.xlsx）。对于成年早期雄性，生活史性状与繁殖性状在不同个体中测量，数据分别位于独立工作表中；对于成年早期雌性、成年晚期雌性及成年晚期雄性，所有性状均在同一批个体中测量，每组数据整合于单个工作表内。

R代码（Trait variability.R）使用R v1.3.1093运行，分析所用数据来自female.csv和male.csv文件。

----- 分析方法 -----

我们采用贝叶斯多变量混合效应模型（Bayesian multivariate mixed-effects models）估算性状协方差，并检验发育环境是否同时影响性状均值与方差，模型针对每个年龄-性别组单独运行。成年早期雄性的生活史性状与繁殖性状使用独立模型分析。环境（3个水平）作为固定效应纳入模型，窝别作为随机效应。成熟年龄、免疫能力、产卵数（成年早期雌性）及精子数量等性状经幂变换以提升正态性。每个模型使用4条马尔可夫链蒙特卡罗链（MCMC chains），迭代5000次（含1000次burn-in期），收敛性经确认（Rhat=1）且有效样本量>1480。

为检验环境对性状方差的影响，我们在对数尺度上对性状均值与标准差（SD）建模，并计算以下指标：
lnVR = ln(SD1/SD2)，用于评估原始方差变化
lnCVR = ln(CV1/CV2)，用于评估相对方差变化（已考虑性状均值影响）

我们通过比较特定环境组合以评估：
盐度效应：稳定盐度 vs 淡水
波动效应：波动盐度 vs 稳定盐度