ggplot2中能否为geom_line的多个图例分别配置不同图例键？

要实现颜色图例用填充方块（便于区分数据集颜色）、线型图例用线条（展示变量线型差异）的效果，你可以在guides()中分别为每个美学映射指定对应的key_glyph参数，而非全局修改geom_line的键样式：

library(tidyverse)

data_1 <-
  tibble(
    dataset = "Dataset 1",
    x = seq(0, 10),
    y1 = 2 * x,
    y2 = 3 * x
  )

data_2 <-
  tibble(
    dataset = "Dataset 2",
    x = seq(0, 10),
    y1 = 2.2 * x,
    y2 = 3.3 * x
  )

data <- bind_rows(data_1, data_2) |> 
  pivot_longer(cols = c(y1, y2), names_to = "variable")

ggplot(
  data = data,
  mapping = aes(x = x, y = value, colour = dataset, linetype = variable)
) +
  geom_line() +
  guides(
    colour = guide_legend(
      order = 1,
      key_glyph = "rect"  # 为颜色图例设置填充方块键
    ),
    linetype = guide_legend(
      key_glyph = "path"  # 为线型图例保留默认线条键
    )
  ) +
  theme_bw()

关键说明 #

• key_glyph是guide_legend()的专属参数，并非override.aes的子参数，这也是你之前尝试override.aes = list(key_glyph = "rect")无效的原因。
• 使用"rect"作为颜色图例的键样式，能更直观地展示颜色差异；"path"是geom_line默认的键样式，适合呈现线型变化。

自定义图例键（可选） #

如果需要更个性化的图例键样式，可以自定义draw_key函数。例如创建带边框的颜色方块：

# 自定义带边框的方块键
draw_key_bordered_rect <- function(data, params, size) {
  grid::rectGrob(
    width = grid::unit(0.6, "npc"),
    height = grid::unit(0.6, "npc"),
    gp = grid::gpar(
      fill = data$colour,
      col = "black",  # 添加边框颜色
      lwd = 1         # 边框宽度
    )
  )
}

ggplot(
  data = data,
  mapping = aes(x = x, y = value, colour = dataset, linetype = variable)
) +
  geom_line() +
  guides(
    colour = guide_legend(
      order = 1,
      key_glyph = draw_key_bordered_rect
    ),
    linetype = guide_legend(key_glyph = "path")
  ) +
  theme_bw()

内容的提问来源于stack exchange，提问作者David