as proposed by J.J Hill. so far we still can do such a thing:
library(Momocs)
data(bot)
bot.p <- PCA(efourier(bot, 12))
x <- PCcontrib(bot.p)
# all reconstructed shapes are in the ggplot2's $data
x$data %>%
# we dont want all columns
select(x=x_c, y=y_c, shp1, nax) %>%
# we split for each pair of nax x shp1
split(f=list(.$shp1, .$nax)) %>%
# you can stop here or continue to a panel
Out() %>% panel()
## Triple check the order of shapes what you obtain !
as proposed by J.J Hill. so far we still can do such a thing: