在ggplot2分组叠加图中设置不同形状类型的实现方法

当然可以实现你的需求，核心是绕过全局形状映射的限制，给不同图层单独指定形状规则。以下是两种可行方案：

方案一：无需额外包，手动指定形状数值 #

直接在每个图层的aes()中根据分组变量分配空心/实心形状，再调整图例统一显示：

library(ggplot2)

# 获取seeding.rate的实际水平（替换为你的数据真实值）
seed_levels <- levels(econ4$seeding.rate)

p <- ggplot() +  
  # 抖动观测点：空心形状分组
  geom_jitter(data = econ4, 
              position = position_jitter(0.2),
              aes(x = location, 
                  y = profit, 
                  shape = ifelse(seeding.rate == seed_levels[1], 1, 2),  # 1=空心圆，2=空心三角
                  colour = urea.cost.USD),
              alpha = 0.05) +  # alpha放在aes外，避免多余图例
  theme_bw() + 
  geom_hline(yintercept = 0) + 
  labs(y = "profit per ha (USD)", shape = "seeding.rate") +
  # 均值点：实心形状分组
  geom_point(data = econ4, 
             aes(x = location, 
                 y = mean.profit, 
                 colour = urea.cost.USD, 
                 shape = ifelse(seeding.rate == seed_levels[1], 16, 17)),  #16=实心圆，17=实心三角
             size = 2) +  # 增大均值点，强化视觉区分
  geom_linerange(data = econ4, 
                 aes(x = location, 
                     y = mean.profit, 
                     ymin = mean.profit - se.profit, 
                     ymax = mean.profit + se.profit, 
                     colour = urea.cost.USD)) +
  scale_colour_manual(values = c("#0072B2","#009E73", "#CC79A7")) +
  coord_flip() +
  # 调整图例：合并空心/实心对应的同一分组
  scale_shape_manual(values = c(1, 2, 16, 17),
                     labels = c(seed_levels[1], seed_levels[2], seed_levels[1], seed_levels[2]),
                     guide = guide_legend(
                         override.aes = list(shape = c(1, 16)),
                         label = seed_levels,
                         ncol = 1
                     ))
p

方案二：使用ggnewscale包（更简洁） #

ggnewscale允许为同一美学属性设置多个映射规则，代码结构更清晰：

library(ggplot2)
library(ggnewscale)  # 需要先安装：install.packages("ggnewscale")

p <- ggplot() +  
  # 抖动点：空心形状映射
  geom_jitter(data = econ4, 
              position = position_jitter(0.2),
              aes(x = location, y = profit, shape = seeding.rate, colour = urea.cost.USD),
              alpha = 0.05) +
  scale_shape_manual(name = "seeding.rate", values = c(1, 2)) +  # 空心形状
  scale_colour_manual(values = c("#0072B2","#009E73", "#CC79A7")) +
  
  new_scale("shape") +  # 开启新的shape映射规则
  
  # 均值点：实心形状映射
  geom_point(data = econ4, 
             aes(x = location, y = mean.profit, shape = seeding.rate, colour = urea.cost.USD),
             size = 2) +
  scale_shape_manual(name = "seeding.rate", values = c(16, 17)) +  # 实心形状
  
  # 其他图层
  geom_linerange(data = econ4, 
                 aes(x = location, y = mean.profit, ymin = mean.profit-se.profit, ymax = mean.profit+se.profit, colour = urea.cost.USD)) +
  theme_bw() + 
  geom_hline(yintercept = 0) + 
  labs(y = "profit per ha (USD)") +
  coord_flip()
p

关键注意事项 #

• 不要把alpha=0.05放在aes()内部，否则会生成多余的透明度图例。
• 给均值点设置size=2可以和抖动点形成明显视觉区分，提升图表可读性。

内容的提问来源于stack exchange，提问作者LKK