ggplot2中为单独添加的元素添加图例的技术求助

我来帮你搞定这个图例问题！你之前的思路方向是对的，但没抓住ggplot2图例的核心——它是基于**美学映射（aesthetic mappings）**生成的，而不是直接手动指定颜色后硬加图例。下面是一步到位的解决方案：

第一步：把宽数据整理为长格式 #

ggplot2最擅长处理“整洁数据”（每一行对应一个观测），所以我们先把你的宽数据（RR、MI、PE各列为独立变量）转成三列结构：Month（月份）、Variable（记录是RR/MI/PE）、Value（对应数值）。用tidyr::pivot_longer就能轻松完成转换：

library(ggplot2)
library(tidyr) # 加载转换数据的包

# 你的模拟数据
Month <- c(1, 1, 1, 2, 2, 2, 3, 3, 3, 4, 4, 4, 5, 5, 5, 6, 6, 6, 7, 7, 7, 8, 8, 8, 9, 9, 9, 10, 10, 10, 11, 11, 11, 12, 12, 12)
RR <- c(81.30271, 56.97511, 88.95428, 78.43363, 51.39398, 83.53967, 95.21302, 63.74089, 94.27137, 92.36272, 69.30449, 100.34571, 128.52172, 89.84833, 133.61527, 128.23847, 91.21498, 145.45016, 124.72499, 85.96523, 153.82113, 123.22878, 97.14116, 154.57004, 111.59289, 83.26763, 105.47873, 95.55557, 88.77395, 87.31212, 93.03579, 65.35055, 93.22413, 103.43140, 70.40292, 91.58487)
MI <- c(66.729379, 41.891775, 52.137614, 59.376967, 30.717318, 49.339675, 62.469691, 25.667561, 60.211374, 48.902722, 18.764486, 65.565712, 69.985054, 27.418330, 89.231939, 55.685138, 11.484980, 90.666826, 44.101654, -8.448102, 87.637798, 55.978782, 15.431156, 92.310042, 69.596228, 34.897628, 56.505393, 69.008904, 61.627285, 36.935451, 76.392457, 47.493886, 53.750796, 88.204738, 54.806257, 54.358201)
PE <- c(14.57333, 15.08333, 36.81667, 19.05667, 20.67667, 34.20000, 32.74333, 38.07333, 34.06000, 43.46000, 50.54000, 34.78000, 58.53667, 62.43000, 44.38333, 72.55333, 79.73000, 54.78333, 80.62333, 94.41333, 66.18333, 67.25000, 81.71000, 62.26000, 41.99667, 48.37000, 48.97333, 26.54667, 27.14667, 50.37667, 16.64333, 17.85667, 39.47333, 15.22667, 15.59667, 37.22667)
tt <- data.frame(Month, RR, MI, PE)

# 转换为长格式数据
tt_long <- pivot_longer(tt, cols = c(RR, MI, PE), names_to = "Variable", values_to = "Value")

第二步：用美学映射自动生成图例 #

现在只需要一次调用geom_boxplot和stat_summary，通过aes把fill、colour映射到Variable列，ggplot会自动识别并生成对应的图例。最后用scale_fill_manual和scale_colour_manual设置你想要的颜色即可：

ggplot(data = tt_long, aes(x = factor(Month), y = Value)) + 
  geom_boxplot(aes(fill = Variable, colour = Variable), alpha = 0.6) + 
  stat_summary(aes(colour = Variable), fun = mean, geom = "smooth", se = FALSE) + # se=FALSE去掉置信区间，和你原代码效果一致
  labs(x = "Months", y = "Flux, mm", fill = "", colour = "") + # 去掉图例标题
  scale_fill_manual(values = c("RR" = "dodgerblue4", "MI" = "dimgray", "PE" = "firebrick")) +
  scale_colour_manual(values = c("RR" = "dodgerblue4", "MI" = "dimgray", "PE" = "firebrick")) +
  theme_bw(base_size = 18)

为什么你的原代码没成功？ #

• 你直接给每个geom_boxplot和stat_summary指定fill/colour（而非在aes里做映射），ggplot不会把这些元素和图例关联——因为它们没有绑定到数据的某个变量。
• 你用scale_fill_manual时，breaks和labels用的是颜色值，这不符合逻辑：breaks应该对应你要映射的变量名（RR/MI/PE），而非颜色代码。

这样修改后，不仅能正确生成图例，代码也更简洁易维护，后续要添加新变量时，只需要调整数据转换部分即可，不用重复编写多个geom图层。

内容的提问来源于stack exchange，提问作者user3262756