Add pvalue bracket with stat_pvalue_manual on geom_bar

I wouldl like to use stat_pvalue_manual in order to add pvalue on my geom barplot. I have these following dataset and stat.table (created to reproduce my data). I can add the pvalue, but the bracket does not appear as it should. Can you help me to see what is wrong in my code.

 # data
######################### 

Data_test=structure(list(PE = c(87, 32, 15, 87, 43, 65, 76, 25), 
                         BU = c("A1", "A2", "A3", "A4", "A1", "A2", "A3", "A4"), 
                         PH = c("NAF", "NAF","NAF", "NAF", "AF", "AF", "AF", "AF")), 
                   class = c("grouped_df", "tbl_df", "tbl", "data.frame"), 
                   row.names = c(NA, -8L), 
                   groups = structure(list(BU = c("A1","A1", "A2", "A2", "A3", "A3","A4", "A4"), 
                   PH = c("AF","NAF", "AF", "NAF", "AF", "NAF","AF", "NAF"), .rows = structure(list(5L, 1L, 6L, 2L, 7L, 3L, 8L, 4L), ptype = integer(0), class = c("vctrs_list_of", "vctrs_vctr", "list"))), class = c("tbl_df", "tbl", "data.frame"
                                                                                                                                                                                                                                                                                                                               ), row.names = c(NA, -8L), .drop = TRUE))

 # stat table
######################### 

Stat_test= structure(list(group1 = c("NAF", "NAF", "NAF", 
                          "NAF"), group2 = c("AF", "AF", "AF", 
                                                      "AF"), PE = c(65, 87, 87, 
                                                                            43), 
                                                                                                      
                                                                                                                                                         p = c(0.7654, 
                                                                                                                                                                                                                       0.456, 0.789, 0.003),  BU = c("A1", "A2", "A3", "A4")), class = "data.frame", row.names = c(NA, 
                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                  -4L))

# Graph
#########################  

library(ggpubr)
library(ggplot2)

ggplot(Data_test, aes(x=BU, y=PE)) +
  geom_bar(aes(fill=PH), stat="identity", position=position_dodge(), color="black") +
  stat_pvalue_manual(Stat_test, y.position = 100, size=4,
                     label = "p", inherit.aes = FALSE, remove.bracket = FALSE,  x="BU", 
                     tip.length = 0.02, bracket.size = 4, ) +
  ylim(c(0,100))

> Data_test
# A tibble: 8 × 3
# Groups:   BU, PH [8]
     PE BU    PH   
  <dbl> <chr> <chr>
1    87 A1    NAF  
2    32 A2    NAF  
3    15 A3    NAF  
4    87 A4    NAF  
5    43 A1    AF   
6    65 A2    AF   
7    76 A3    AF   
8    25 A4    AF

> Stat_test
  group1 group2 PE      p BU ypos
1    NAF     AF 65 0.7654 A1  100
2    NAF     AF 87 0.4560 A2  100
3    NAF     AF 87 0.7890 A3  100
4    NAF     AF 43 0.0030 A4  100

There may be an easier way, but adding xmin and xmax values to your Stat_test data frame will do the trick:

ggplot(Data_test, aes(x = BU, y = PE)) +
  geom_col(aes(fill = PH), position = position_dodge(), color = "black") +
  stat_pvalue_manual(Stat_test %>% 
                       mutate(xmin = 1:4 -0.2, xmax = 1:4 + 0.2), 
                     y.position = 100, size = 4, label = "p", 
                     tip.length = 0.02, bracket.size = 0.8) +
  ylim(c(0, 100)