diff --git a/r-package/DESCRIPTION b/r-package/DESCRIPTION
index ed557b54..f8a66e07 100644
--- a/r-package/DESCRIPTION
+++ b/r-package/DESCRIPTION
@@ -32,6 +32,7 @@ Depends:
     R (>= 3.5.0)
 Imports:
       curl,
+      dplyr (>= 0.8-3),
       data.table,
       httr (>= 1.4.1),
       methods,
@@ -41,7 +42,6 @@ Suggests:
     arrow (>= 15.0.1),
     censobr (>= 0.3.2),
     covr,
-    dplyr (>= 0.8-3),
     ggplot2 (>= 3.3.1),
     knitr,
     rmarkdown,
diff --git a/r-package/NEWS.md b/r-package/NEWS.md
index f0b74c8c..49f3082a 100644
--- a/r-package/NEWS.md
+++ b/r-package/NEWS.md
@@ -6,9 +6,13 @@
 - Function `read_urban_area()` now has a new parameter `code_state`, which will allow users to filter selected states. Closes #338
 - Function `read_metro_area()` now has a new parameter `code_state`, which will allow users to filter selected states.  Closes #338
 
+**Bug fix**
+- Using `data.table::rbindlist()` to rind data was throwing errors when some observations were of class `POLYGONS` and others were `MULTIPOLYGONS`. This has now been replaced with `dplyr::bind_rows()` at a very small performance penalty. Closes #346.
+
 **New data**
 - schools for 2023
 - health facilities for 202303
+- census tracts for 2020 and 2022
 
 
 # geobr v1.8.2 
diff --git a/r-package/R/utils.R b/r-package/R/utils.R
index 4aea6144..6e05a65d 100644
--- a/r-package/R/utils.R
+++ b/r-package/R/utils.R
@@ -283,7 +283,8 @@ load_gpkg <- function(temps=NULL){
 
     # read files and pile them up
     files <- lapply(X=temps, FUN= sf::st_read, quiet=TRUE)
-    temp_sf <- sf::st_as_sf(data.table::rbindlist(files, fill = TRUE)) # do.call('rbind', files)
+    # temp_sf <- sf::st_as_sf(data.table::rbindlist(files, fill = TRUE)) # do.call('rbind', files)
+    temp_sf <- dplyr::bind_rows(files)
 
     # closes issue 284
     col1 <- names(temp_sf)[1]