如何在双Y轴（不同刻度）的ggplot图中绘制清晰误差棒？

我尝试使用ggplot2绘制带有双Y轴的图表，但两个轴的刻度差异极大，其中EVI变量的误差棒因标准误过小难以在图中辨识。以下是我的代码：

date<-as.Date(c("2008-10-01","2009-09-01","2010-11-01","2011-09-01",
              "2012-10-01","2013-09-01","2014-09-01","2015-09-01",
              "2016-09-01","2017-09-01"))
year2<-c(2008,2009,2010,2011,2012,2013,2014,2015,2016,2017)
evi.total<-c(0.074,0.064,0.066,0.060,0.073,NA,0.064,0.073,0.076,0.076)
densidad.total<-c(22.78,17.19,19.39,16.46,32.34,NA,13.62,26.41,26.05,34.77)
se.dens.tot<-c(4.50, 4.76,4.13,3.93,7.75,NA,3.51,5.63,4.86,7.02)
sd.evi.general<-c(0.011,0.010,0.011,0.010,0.011,NA,0.012,0.011,0.012,0.014)
n<-c(5018,5018,5018,5018,5018,NA,5018,5018,5018,5018)
se.evi<-sd.evi.general/(sqrt(n))

data<-data.frame(date,year2,evi.total,densidad.total,se.dens.tot,sd.evi.general,
                 se.evi)

library(ggplot2)

p <- ggplot(data=data,aes(x = date))
p<-p + scale_x_date(date_breaks="3 month", date_labels = "%b-%y", 
                  )                            
p <- p + geom_line(aes(y =densidad.total, colour = "Guanaco's density"),size=.5)+
         geom_linerange(aes(y=densidad.total,ymin=densidad.total-se.dens.tot, 
                            ymax=densidad.total+se.dens.tot), 
                            color="tomato1", size=1,lwd=.07)

p <- p + geom_line (aes(y =evi.total*1000, colour = "EVI"),size=.5) +  
  labs(color=NULL) + geom_linerange(aes(y=evi.total*1000,
                                    ymin=evi.total*1000-se.evi*1000, 
                                    ymax=evi.total*1000+se.evi*1000),
                                     color="aquamarine4",size=1)
                             

p <- p + scale_y_continuous(sec.axis = sec_axis(~./1000, name = " EVI"))

p <- p + scale_colour_manual(values = c("aquamarine4", "tomato1"))
p <- p +theme_classic()

p <- p + labs(y = expression ("Guanaco's density "(guanaco/~km^2), x = "",
              colour = "")) 
p <- p + theme(legend.position = "bottom")+
  theme(text = element_text(size = 20,hjust = 1),  
        axis.line.x = element_line(color="black", size = 1),
        axis.line.y = element_line(color="black", size = 1),
        axis.text.x=element_text(colour="black",angle=60, size=10),
        axis.text.y=element_text(colour="black"))


p

我曾参考相关解决方案，但仍未解决误差棒可视性问题。

解决方案 #

1. 放大EVI误差棒显示比例 #

由于EVI的标准误极小，按原比例缩放后误差棒长度几乎不可见，可以手动放大误差棒的范围，同时在图表中明确标注该处理，避免误导。

修改EVI的geom_linerange代码：

geom_linerange(aes(y=evi.total*1000,
                   ymin=evi.total*1000 - se.evi*1000*100, 
                   ymax=evi.total*1000 + se.evi*1000*100),
               color="aquamarine4", size=1)

同时在图表标题或图例中添加说明：EVI误差棒已放大100倍。

2. 拆分为独立子图（推荐） #

双Y轴容易造成视觉误导，将两个指标拆分为上下独立的子图，各自使用适配的刻度，能彻底解决误差棒可视性问题，也更符合数据可视化规范。可以借助patchwork包实现：

# 安装并加载patchwork
install.packages("patchwork")
library(patchwork)

# 绘制骆马密度图表
p1 <- ggplot(data=data,aes(x = date)) +
  scale_x_date(date_breaks="3 month", date_labels = "%b-%y") +
  geom_line(aes(y =densidad.total, colour = "Guanaco's density"),size=.5)+
  geom_linerange(aes(y=densidad.total,ymin=densidad.total-se.dens.tot, 
                     ymax=densidad.total+se.dens.tot), 
                 color="tomato1", size=1,lwd=.07) +
  scale_colour_manual(values = "tomato1") +
  theme_classic() +
  labs(y = expression ("Guanaco's density "(guanaco/~km^2)), x = "") +
  theme(legend.position = "bottom",
        text = element_text(size = 20,hjust = 1),  
        axis.line.x = element_line(color="black", size = 1),
        axis.line.y = element_line(color="black", size = 1),
        axis.text.x=element_text(colour="black",angle=60, size=10),
        axis.text.y=element_text(colour="black"))

# 绘制EVI图表
p2 <- ggplot(data=data,aes(x = date)) +
  scale_x_date(date_breaks="3 month", date_labels = "%b-%y") +
  geom_line(aes(y =evi.total, colour = "EVI"),size=.5) +
  geom_linerange(aes(y=evi.total,
                     ymin=evi.total-se.evi, 
                     ymax=evi.total+se.evi),
                 color="aquamarine4",size=1) +
  scale_colour_manual(values = "aquamarine4") +
  theme_classic() +
  labs(y = "EVI", x = "") +
  theme(legend.position = "bottom",
        text = element_text(size = 20,hjust = 1),  
        axis.line.x = element_line(color="black", size = 1),
        axis.line.y = element_line(color="black", size = 1),
        axis.text.x=element_text(colour="black",angle=60, size=10),
        axis.text.y=element_text(colour="black"))

# 合并上下子图
p1 / p2

3. 调整EVI缩放系数 #

通过增大EVI的缩放倍数，让其数值范围更接近密度的刻度范围，从而间接增加误差棒长度。例如将原缩放系数1000改为5000：

# 修改EVI线条和误差棒的缩放
p <- p + geom_line (aes(y =evi.total*5000, colour = "EVI"),size=.5) +  
  labs(color=NULL) + geom_linerange(aes(y=evi.total*5000,
                                        ymin=evi.total*5000-se.evi*5000, 
                                        ymax=evi.total*5000+se.evi*5000),
                                     color="aquamarine4",size=1)

# 调整右侧Y轴的转换公式
p <- p + scale_y_continuous(sec.axis = sec_axis(~./5000, name = " EVI"))

这种方式能提升误差棒可视性，但仍存在双Y轴的视觉歧义，需谨慎使用。

内容的提问来源于stack exchange，提问作者user228095