ggplot绘图：解决误差线穿透空心散点的问题

使用ggplot绘制带散点与误差棒的图表，需求如下：

• 用颜色区分treatment变量
• 用实心/空心点区分trend变量（trend对应lambda是否大于1）
• 避免误差线穿透散点

尝试两种方案均未成功：

1. 手动设置fill和col刻度，但未达到预期效果：

plot1 <- ggplot(data = dat, aes(x = species, y = lambda, col = treatment, fill = trend)) +
  geom_abline(intercept = 1, slope = 0, linetype = "dashed") +
  geom_errorbar(aes(ymin = lower_BCPI_red, ymax = upper_BCPI_red), width = 0.2, position = position_dodge(width = 0.7)) +
  geom_point(position = position_dodge(width = 0.7), size = 4) +
  scale_color_manual(
    values = c("darkblue", "darkgreen"),
    labels = c("treat1" = "Treatment 1", "treat2" = "Treatment 2"),
    guide = guide_legend(order = 1)) +
  scale_fill_manual(
    values = c("white", "darkgreen"),
    labels = c("grow" = expression(paste(lambda, " > 1")), "shrink" = expression(paste(lambda, " < 1"))),
    guide = guide_legend(order = 2))

2. 用形状区分trend变量，但空心形状（21）会被误差线穿透；尝试叠加白色点遮挡，但因position_dodge未对齐失效：

plot2 <- ggplot(data = dat, aes(x = species, y = lambda, col = treatment, shape = trend)) +
  geom_abline(intercept = 1, slope = 0, linetype = "dashed") +
  geom_errorbar(aes(ymin = lower_BCPI_red, ymax = upper_BCPI_red), width = 0.2, position = position_dodge(width = 0.7)) +
  geom_point(position = position_dodge(width = 0.7), size = 4, col = "white", fill = "white") +
  geom_point(position = position_dodge(width = 0.7), size = 4) +
  scale_color_manual(
    values = c("darkblue", "darkgreen"),
    labels = c("treat1" = "Treatment 1", "treat2" = "Treatment 2"),
    guide = guide_legend(order = 1)) +
  scale_shape_manual(
    values = c(19, 21),
    labels = c("grow" = expression(paste(lambda, " > 1")), "shrink" = expression(paste(lambda, " < 1"))),
    guide = guide_legend(order = 2)) 

数据：

dat <- structure(list(species = c("spec2", "spec2", "spec1", "spec1"
), treatment = c("treat2", "treat1", "treat2", "treat1"), trend = c("grow", 
"grow", "grow", "shrink"), lambda = c(2.93041381166779, 2.02556855849611, 
1.26209237527084, 0.983136908926705), lower_BCPI_red = c(2.23364635527003, 
1.08955927540168, 1.02410220209615, 0.917503038357754), upper_BCPI_red = c(3.87343015758682, 
5.08404412887266, 1.40109042463976, 0.998656722552173)), row.names = c(NA, 
-4L), class = c("tbl_df", "tbl", "data.frame"))

解决方案 #

核心思路：

• 使用支持填充的形状：用实心形状（19）对应lambda>1（trend=grow），用可填充的空心形状（21）对应lambda<1（trend=shrink）
• 统一position_dodge参数，避免图层对齐偏差
• 为空心形状设置白色填充，遮挡下方的误差线

完整代码：

# 统一定义dodge位置，避免对齐问题
pos_dodge <- position_dodge(width = 0.7)

ggplot(dat, aes(x = species, y = lambda, color = treatment)) +
  # 添加lambda=1的参考线
  geom_abline(intercept = 1, slope = 0, linetype = "dashed") +
  # 绘制误差棒，使用统一的dodge位置
  geom_errorbar(aes(ymin = lower_BCPI_red, ymax = upper_BCPI_red), 
                width = 0.2, position = pos_dodge) +
  # 绘制散点：shape映射trend，fill设为白色（仅对空心形状生效）
  geom_point(aes(shape = trend), 
             size = 4, 
             position = pos_dodge,
             fill = "white") +
  # 设置treatment的颜色映射
  scale_color_manual(
    values = c("darkblue", "darkgreen"),
    labels = c("treat1" = "处理1", "treat2" = "处理2"),
    guide = guide_legend(order = 1)
  ) +
  # 设置trend的形状映射：grow用实心19，shrink用空心21
  scale_shape_manual(
    values = c(grow = 19, shrink = 21),
    labels = c(grow = expression(lambda > 1), shrink = expression(lambda < 1)),
    guide = guide_legend(order = 2)
  )

效果说明 #

• trend=grow（lambda>1）：实心点，颜色与对应treatment一致
• trend=shrink（lambda<1）：空心点，边框颜色与对应treatment一致，内部白色填充遮挡住下方的误差线
• 所有图层使用统一的pos_dodge，确保散点与误差棒完全对齐

内容的提问来源于stack exchange，提问作者Em Laskey