如何在ggplot箱线图误差棒上限精准添加显著性字母

我尝试为箱线图的误差棒上限添加显著性字母，但字母位置无法精准对应上限。希望将字母精准放置在每个箱线图的误差棒上限处，同时保留图表其他元素不变。

原使用代码 #

#ANOVA提取显著性字母
mod <- aov(gas ~ Genotype, data = data) 
tu <- TukeyHSD(mod)
group <- multcompLetters4(mod, tu)
group_letter <- data.frame(group$`Genotype`$Letters)
# 数据汇总并添加字母列
df = data %>% group_by(Genotype, Species) %>%
summarize(MaxVal = max(gas), sd=sd(gas),mean=mean(gas),
            median= median(gas),se = sd(gas/sqrt(gas)),
            Q50 = quantile(gas, probs = 0.50),
            Q25 = quantile(gas, probs = 0.25),
            Q75 = quantile(gas, probs = 0.75),
            WhiskUp = (Q75 * (Q75 - Q25)),.groups = "keep")
df$label <- group_letter$group.Genotype.Letters

bp <- boxplot(gas ~ Genotype, data = data, plot = FALSE)
bp
data %>% 
  transform(Genotype = factor(Genotype, levels = bp$names[order(-bp$stats[4,])])) %>%
  ggplot(aes(x=Genotype, y=Pn, fill=Species)) +
  stat_boxplot(geom = "errorbar") +
  scale_fill_manual(values=c('#e9002d', '#ffaa00', '#00b000')) +
  geom_boxplot(linetype = 3) +
  theme(axis.text.x = element_text(size = 10, colour = "black",vjust = 0.5, angle = 90))+ 
  geom_text(data = df, aes(y = MaxVal-WhiskUp, label = label), fontface="italic", 
            size=3.5, position = position_dodge(width = .75), 
            angle=90, hjust=0.4, vjust=0.3)

数据集 #

Genotype    Species gas
BGB001  sp1 0.214
BGB001  sp1 0.216
BGB001  sp1 0.109
BGB001  sp1 0.097
BGB003  sp1 0.182
BGB003  sp1 0.270
BGB003  sp1 0.181
BGB003  sp1 0.282
BGB008  sp1 0.349
BGB008  sp1 0.206
BGB008  sp1 0.388
BGB008  sp1 0.157
BGB009  sp1 0.170
BGB009  sp1 0.179
BGB009  sp1 0.197
BGB009  sp1 0.164
BGB011  sp1 0.266
BGB011  sp1 0.198
BGB011  sp1 0.410
BGB011  sp1 0.352
BGB027  sp1 0.348
BGB027  sp1 0.344
BGB027  sp1 0.353
BGB027  sp1 0.350
BGB045  sp1 0.311
BGB045  sp1 0.211
BGB045  sp1 0.277
BGB045  sp1 0.340
BGB048  sp2 0.486
BGB048  sp2 0.504
BGB048  sp2 0.187
BGB048  sp2 0.205
BGB055  sp2 0.257
BGB055  sp2 0.273
BGB055  sp2 0.328
BGB055  sp2 0.342
BGB068  sp2 0.238
BGB068  sp2 0.121
BGB068  sp2 0.260
BGB068  sp2 0.133
BGB077  sp2 0.261
BGB077  sp2 0.203
BGB077  sp2 0.275
BGB077  sp2 0.219
BGB083  sp2 0.498
BGB083  sp2 0.485
BGB083  sp2 0.515
BGB083  sp2 0.499
BGB086  sp2 0.471
BGB086  sp2 0.183
BGB086  sp2 0.441
BGB086  sp2 0.134
BGB088  sp3 0.196
BGB088  sp3 0.178
BGB088  sp3 0.174
BGB088  sp3 0.229
BGB089  sp3 0.196
BGB089  sp3 0.448
BGB089  sp3 0.213
BGB089  sp3 0.484
BGB091  sp3 0.143
BGB091  sp3 0.187
BGB091  sp3 0.192
BGB091  sp3 0.155
BGB093  sp3 0.297
BGB093  sp3 0.314
BGB093  sp3 0.497
BGB093  sp3 0.519
BGB094  sp2 0.426
BGB094  sp2 0.335
BGB094  sp2 0.307
BGB094  sp2 0.221
BGB095  sp2 0.128
BGB095  sp2 0.099
BGB095  sp2 0.190
BGB095  sp2 0.186
BGB096  sp2 0.357
BGB096  sp2 0.274
BGB096  sp2 0.350
BGB096  sp2 0.539
BGB097  sp2 0.172
BGB097  sp2 0.174
BGB097  sp2 0.208
BGB097  sp2 0.459
BGB098  sp2 0.379
BGB098  sp2 0.380
BGB098  sp2 0.543
BGB098  sp2 0.557
BGB099  sp2 0.361
BGB099  sp2 0.354
BGB099  sp2 0.296
BGB099  sp2 0.289
BGB100  sp2 0.357
BGB100  sp2 0.529
BGB100  sp2 0.323
BGB100  sp2 0.503
BGB101  sp2 0.231
BGB101  sp2 0.235
BGB101  sp2 0.233
BGB101  sp2 0.236
BGB102  sp2 0.323
BGB102  sp2 0.402
BGB102  sp2 0.389
BGB102  sp2 0.501
BGB103  sp2 0.315
BGB103  sp2 0.314
BGB103  sp2 0.312
BGB103  sp2 0.311
BGB104  sp2 0.423
BGB104  sp2 0.323
BGB104  sp2 0.280
BGB104  sp2 0.160
BGB105  sp2 0.151
BGB105  sp2 0.137
BGB105  sp2 0.170
BGB105  sp2 0.092
BGB106  sp2 0.274
BGB106  sp2 0.279
BGB106  sp2 0.241
BGB106  sp2 0.246
BGB107  sp2 0.137
BGB107  sp2 0.145
BGB107  sp2 0.148
BGB107  sp2 0.133
BGB108  sp2 0.505
BGB108  sp2 0.526
BGB108  sp2 0.351
BGB108  sp2 0.380
BGB109  sp2 0.307
BGB109  sp2 0.283
BGB109  sp2 0.306
BGB109  sp2 0.283
BGB110  sp2 0.148
BGB110  sp2 0.186
BGB110  sp2 0.089
BGB110  sp2 0.318
BGB111  sp2 0.162
BGB111  sp2 0.185
BGB111  sp2 0.125
BGB111  sp2 0.138
BGB113  sp2 0.673
BGB113  sp2 0.780
BGB113  sp2 0.482
BGB113  sp2 0.666
BGB444  sp2 0.384
BGB444  sp2 0.222
BGB444  sp2 0.374
BGB444  sp2 0.190
BGB451  sp2 0.569
BGB451  sp2 0.562
BGB451  sp2 0.641
BGB451  sp2 0.635
BGB453  sp2 0.136
BGB453  sp2 0.159
BGB453  sp2 0.173
BGB453  sp2 0.194
BGB460  sp1 0.127
BGB460  sp1 0.246
BGB460  sp1 0.189
BGB460  sp1 0.314
BGB467  sp2 0.270
BGB467  sp2 0.274
BGB467  sp2 0.273
BGB467  sp2 0.270
BGB472  sp2 0.451
BGB472  sp2 0.634
BGB472  sp2 0.445
BGB472  sp2 0.600

解决方案 #

核心问题分析 #

1. 上须计算错误：原代码中WhiskUp = Q75*(Q75-Q25)不符合箱线图上须的定义（正确为Q75 + 1.5*(Q75-Q25)），导致字母位置偏离实际误差棒。
2. 变量映射错误：ggplot中aes(y=Pn)是笔误，应改为y=gas才能正确关联数据。
3. 手动计算偏差：用max(gas)和手动计算的上须值无法和ggplot自动生成的箱线图误差棒完全匹配。

修正后代码 #

library(tidyverse)
library(multcompView)

# 1. 提取显著性字母
mod <- aov(gas ~ Genotype, data = data) 
tu <- TukeyHSD(mod)
group <- multcompLetters4(mod, tu)
group_letter <- data.frame(
  Genotype = rownames(group$Genotype), 
  label = group$Genotype$Letters
)

# 2. 获取箱线图的精确上须值
bp <- boxplot(gas ~ Genotype, data = data, plot = FALSE)
whisk_up_df <- data.frame(
  Genotype = bp$names, 
  whisk_up = bp$stats[5,]  # 直接取boxplot返回的上须值
)

# 3. 合并数据：基因型、物种、字母、上须值
df <- data %>% 
  group_by(Genotype, Species) %>%
  summarize(.groups = "keep") %>%
  left_join(group_letter, by = "Genotype") %>%
  left_join(whisk_up_df, by = "Genotype")

# 4. 绘制图表并添加字母
data %>% 
  transform(Genotype = factor(Genotype, levels = bp$names[order(-bp$stats[4,])])) %>%
  ggplot(aes(x=Genotype, y=gas, fill=Species)) +
  stat_boxplot(geom = "errorbar") +
  scale_fill_manual(values=c('#e9002d', '#ffaa00', '#00b000')) +
  geom_boxplot(linetype = 3) +
  theme(axis.text.x = element_text(size = 10, colour = "black", vjust = 0.5, angle = 90)) +
  geom_text(data = df, aes(y = whisk_up, label = label), 
            fontface="italic", size=3.5, 
            position = position_dodge(width = .75), 
            hjust = -0.1)  # 调整hjust让字母位于误差棒外侧

关键改进点 #

• 直接使用boxplot()返回的stats[5,]作为上须值，确保和ggplot绘制的误差棒完全对齐；
• 通过left_join整合所有必要数据，避免手动匹配出错；
• 调整geom_text的hjust参数，让字母刚好落在误差棒外侧，不遮挡图表元素；
• 修正了原代码中的变量笔误和统计量计算错误。

内容的提问来源于stack exchange，提问作者Ikram Bashir