R语言：对两个ggplotGrob对象调用rbind函数时出现‘x和y必须具有相同列数’错误的问题排查

I’ve run into this exact issue before! The problem isn’t just about the number of columns looking the same—it’s about the structure and properties of those columns matching between your two gtable objects. Let me break it down and show you how to fix it.

What’s Causing the Error? #

When you add a 0-width column to g1 with gtable_add_cols(g1, unit(0,"mm")), you’re making the column counts match on paper, but gtable’s rbind() checks more than just column numbers. It verifies that each corresponding column has compatible width definitions and layout metadata. Your 0-width empty column doesn’t match the structure of the legend column in g2, so the function throws an error.

Solution 1: Fix the gtable Column Alignment Properly #

Instead of adding a generic 0-width column, we’ll make g1 match g2’s column structure exactly by reusing the width properties from g2’s extra columns. Here’s the adjusted code:

library(ggplot2)
library(gtable)
library(grid)

x <- seq(1992, 2002, by=2)
d1 <- data.frame(x=x, y=rnorm(length(x)))
xy <- expand.grid(x=x, y=x)
d2 <- data.frame(x=xy$x, y=xy$y, z= jitter(xy$x + xy$y))

p1 <- ggplot(data = d1, mapping = aes(x = x, y = y)) + geom_line(stat = "identity")
p2 <- ggplot(data = d2, mapping = aes(x=x, y=y, fill=z)) + geom_tile()

g1 <- ggplotGrob(p1)
g2 <- ggplotGrob(p2)

# Add columns to g1 that match the width properties of g2's extra columns
cols_to_add <- ncol(g2) - ncol(g1)
if (cols_to_add > 0) {
  g1 <- gtable_add_cols(g1, g2$widths[(ncol(g1)+1):ncol(g2)])
}

# Now stack the plots without error
g <- rbind(g1, g2, size="first")
g$widths <- unit.pmax(g1$widths, g2$widths)

# Center the legend vertically across both plots
g$layout[grepl("guide", g$layout$name), c("t", "b")] <- c(1, nrow(g))

grid.newpage()
grid.draw(g)

This works because we’re not just adding empty columns—we’re copying the exact width attributes from g2’s legend column into g1, so the two gtables are structurally compatible for rbind().

Solution 2: Use Patchwork (Easier Alternative) #

If you don’t want to mess with low-level gtable code, the patchwork package was built specifically for this kind of plot alignment. It handles all the column matching and legend alignment automatically:

library(ggplot2)
library(patchwork)

# Your plot code remains the same
x <- seq(1992, 2002, by=2)
d1 <- data.frame(x=x, y=rnorm(length(x)))
xy <- expand.grid(x=x, y=x)
d2 <- data.frame(x=xy$x, y=xy$y, z= jitter(xy$x + xy$y))

p1 <- ggplot(data = d1, mapping = aes(x = x, y = y)) + geom_line(stat = "identity")
p2 <- ggplot(data = d2, mapping = aes(x=x, y=y, fill=z)) + geom_tile()

# Stack plots and collect legends to center them
p1 / p2 + plot_layout(guides = "collect")

This is way more concise and avoids the gtable column mismatch entirely.

内容的提问来源于stack exchange，提问作者Wolff