如何在ggplot中为两个geom_point系列分别手动定义shape美学？

这个问题我太熟悉了！在ggplot2里，每个美学属性（比如shape）只能对应一个比例尺（scale），你重复调用scale_shape_manual()会直接覆盖掉之前的设置，这就是报错的原因，同时也导致箭头形状没法正常显示。下面给你两个可行的解决方案：

方法1：统一管理所有shape映射（无需额外包） #

我们可以把两组shape的取值都整合到同一个scale_shape_manual里，再通过guides()分别控制两个图例的显示逻辑：

1. 先确保你的形状映射向量包含所有需要的取值：

sh_pal_all <- c(
  "Up" = "\U2191", 
  "No Change" = "\U2194", 
  "Down" = "\U2193", 
  "A1" = "circle", 
  "B1" = "square", 
  "C1" = "diamond", 
  "D1" = "triangle", 
  "E1" = "circle"
)

2. 修改后的完整代码：

ggplot(data = meltdf2_all, aes(x = Month, y = variable, group = variable)) + 
  geom_line(linetype = "dashed", colour = "grey") + 
  # 第一个点层：映射住宿类型到shape
  geom_point(aes(colour = colour, shape = accommodation), size = 3) + 
  # 第二个点层：映射风险变化到shape
  geom_point(data = melt_change_all, aes(shape = value), size = 3) + 
  scale_colour_manual(values = pal_all, limits = names(pal_all), guide = guide_legend("Risk Level")) + 
  # 统一的shape比例尺，包含所有映射关系
  scale_shape_manual(values = sh_pal_all) + 
  scale_x_date(date_breaks = "1 month" , date_labels = "%b-%y") + 
  theme(
    axis.title.y=element_blank(), 
    axis.text.y=element_blank(), 
    axis.ticks.y=element_blank(), 
    axis.line.x = element_line(colour = "darkgrey"), 
    axis.title.x=element_blank(), 
    panel.background = element_rect(fill = "white"), 
    panel.grid.major = element_blank(), 
    panel.grid.minor = element_blank(), 
    legend.position = c(0.2,0.70), 
    legend.key.size = unit(1.5,"line"), 
    legend.box = "horizontal"
  ) + 
  guides(
    colour = guide_legend(override.aes = list(size = 3)),
    # 配置住宿类型的图例
    shape = guide_legend(
      override.aes = list(size = 5),
      breaks = names(sh_pal_all[4:8]),
      title = "Accommodation Type"
    ),
    # 配置风险变化的次级图例
    shape.sec = guide_legend(
      override.aes = list(size = 5),
      breaks = names(sh_pal_all[1:3]),
      title = "Latest Change"
    )
  ) + 
  scale_size(guide = "none")

注意：shape.sec是ggplot2 3.3.0及以上版本才支持的功能，如果你的版本较低，建议用下面的方法2。

方法2：使用ggnewscale添加独立的shape比例尺（推荐） #

ggnewscale包就是专门解决「同一个美学属性需要多个独立比例尺」的场景，步骤很简单：

1. 先安装并加载包：

install.packages("ggnewscale")
library(ggnewscale)

2. 修改后的完整代码：

sh_pal_all <- c(
  "Up" = "\U2191", 
  "No Change" = "\U2194", 
  "Down" = "\U2193", 
  "A1" = "circle", 
  "B1" = "square", 
  "C1" = "diamond", 
  "D1" = "triangle", 
  "E1" = "circle"
)

ggplot(data = meltdf2_all, aes(x = Month, y = variable, group = variable)) + 
  geom_line(linetype = "dashed", colour = "grey") + 
  # 第一个点层：住宿类型的shape映射
  geom_point(aes(colour = colour, shape = accommodation), size = 3) + 
  # 第一个shape比例尺，对应住宿类型
  scale_shape_manual(values = sh_pal_all[4:8], limits = names(sh_pal_all[4:8]), guide = guide_legend("Accommodation Type")) + 
  # 关键：新增一个shape比例尺的上下文，让后续图层使用独立的shape设置
  new_scale("shape") + 
  # 第二个点层：风险变化的shape映射
  geom_point(data = melt_change_all, aes(shape = value), size = 3) + 
  # 第二个shape比例尺，对应风险变化
  scale_shape_manual(values = sh_pal_all[1:3], limits = names(sh_pal_all[1:3]), guide = guide_legend("Latest Change")) + 
  # 颜色比例尺和其他设置
  scale_colour_manual(values = pal_all, limits = names(pal_all), guide = guide_legend("Risk Level")) + 
  scale_x_date(date_breaks = "1 month" , date_labels = "%b-%y") + 
  theme(
    axis.title.y=element_blank(), 
    axis.text.y=element_blank(), 
    axis.ticks.y=element_blank(), 
    axis.line.x = element_line(colour = "darkgrey"), 
    axis.title.x=element_blank(), 
    panel.background = element_rect(fill = "white"), 
    panel.grid.major = element_blank(), 
    panel.grid.minor = element_blank(), 
    legend.position = c(0.2,0.70), 
    legend.key.size = unit(1.5,"line"), 
    legend.box = "horizontal"
  ) + 
  guides(
    colour = guide_legend(override.aes = list(size = 3))
  ) + 
  scale_size(guide = "none")

这个方法兼容性更好，而且逻辑更清晰，两个shape比例尺完全独立，不会互相干扰，箭头形状也能正常显示。

内容的提问来源于stack exchange，提问作者Ant