如何让ggplot2图例为同一colour美学的不同类别显示对应专属geom

我用point、pointrange和line三种geom绘制不同类型数据，通过颜色区分数据源。现在图例同时显示所有geom的样式，需要调整成：

• "acs 1-year"仅显示pointrange样式
• "admin v1"和"admin v2"仅显示line样式
• "decennial census"仅显示point样式
  同时要保留基于colour美学的单个统一图例。

复现数据 #

df <- structure(list(year = c(1990, 1991, 1992, 1993, 1994, 1995, 1996, 
1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 
2008, 2009, 2010, 2011, 2012, 2013, 2014, 2015, 2016, 2017, 2018, 
2019, 2020, 2021, 2022), admin1 = c(105939L, 105568L, 105207L, 
105252L, 104878L, 104498L, 103960L, 104071L, 103689L, 103474L, 
104039L, 104874L, 104949L, 105767L, 106559L, 107339L, 106348L, 
105580L, 104455L, 102815L, 101121L, 99244L, 96823L, 95091L, 94082L, 
92769L, 91811L, 91680L, 91350L, 82767L, 89030L, 89764L, 89942L
), admin2 = c(105944L, 105573L, 105209L, 105254L, 104880L, 104500L, 
103961L, 104072L, 103691L, 103475L, 104039L, 104874L, 104949L, 
105767L, 106559L, 107339L, 106348L, 106021L, 104874L, 103208L, 
101580L, 99732L, 97343L, 95648L, 94685L, 93390L, 92467L, 92375L, 
92137L, 92703L, 93417L, 94222L, 95200L), decennial_count = c(107682, 
NA, NA, NA, NA, NA, NA, NA, NA, NA, 105235, NA, NA, NA, NA, NA, 
NA, NA, NA, NA, 100296, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, 
NA, NA), acs1 = c(NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, 
NA, NA, NA, NA, 113122, 109275, 105005, 105918, 104330, 102660, 
98506, 95432, 97196, 97841, 94046, 93184, 92218, 96609, 92982, 
NA, 96044, NA), acs1_moe = c(NA, NA, NA, NA, NA, NA, NA, NA, 
NA, NA, NA, NA, NA, NA, NA, 4433, 3735, 3484, 3227, 3379, 3527, 
3171, 3259, 3105, 3800, 3711, 3232, 3507, 3816, 3261, NA, 3817, 
NA)), row.names = c(NA, -33L), class = c("tbl_df", "tbl", "data.frame"
))

原绘图代码 #

ggplot(df) +
  geom_line(aes(x = year, y = admin1, color = "admin v1")) +
  geom_line(aes(x = year, y = admin2, color = "admin v2")) +
  geom_pointrange(aes(x = year, y = acs1, ymin = (acs1 - acs1_moe),
                      ymax = (acs1 + acs1_moe), color = "acs 1-year")) +
  geom_point(aes(year, decennial_count, color = "decennial census"),
             size = 4, shape = "diamond")

解决方法 #

核心思路是通过guides()的override.aes参数，针对每个图例项单独设置显示的几何元素，隐藏不需要的样式，同时保留统一的颜色图例。

修改后代码 #

library(ggplot2)

ggplot(df) +
  geom_line(aes(x = year, y = admin1, color = "admin v1")) +
  geom_line(aes(x = year, y = admin2, color = "admin v2")) +
  geom_pointrange(aes(x = year, y = acs1, ymin = (acs1 - acs1_moe),
                      ymax = (acs1 + acs1_moe), color = "acs 1-year")) +
  geom_point(aes(year, decennial_count, color = "decennial census"),
             size = 4, shape = "diamond") +
  guides(color = guide_legend(
    override.aes = list(
      # 按图例顺序设置：admin v1, admin v2, decennial census, acs 1-year
      shape = c(NA, NA, 18, NA),    # 仅decennial census显示菱形点，其余隐藏
      linetype = c(1, 1, 0, 1),     # 仅decennial census隐藏线，其余显示实线
      size = c(0.5, 0.5, 4, 0.5),   # 匹配原图的点/线大小
      ymin = c(NA, NA, NA, mean(df$acs1, na.rm = TRUE)), # 仅acs 1-year显示误差线下限
      ymax = c(NA, NA, NA, mean(df$acs1, na.rm = TRUE))  # 仅acs 1-year显示误差线上限
    )
  )) +
  # 固定图例顺序，避免自动排序混乱
  scale_color_discrete(breaks = c("admin v1", "admin v2", "decennial census", "acs 1-year"))

关键说明 #

• override.aes内的参数顺序必须和图例项的顺序完全对应（可通过scale_color_discrete(breaks)手动指定顺序）
• 用NA隐藏不需要显示的几何元素：比如给不需要点的类别设shape=NA，不需要线的类别设linetype=0
• 误差线的ymin和ymax仅给acs 1-year设置有效值，其余设为NA，确保只有它显示误差线

内容的提问来源于stack exchange，提问作者John J.