diff --git a/ChildProject/pipelines/zooniverse.py b/ChildProject/pipelines/zooniverse.py
index e509cdfc..cd10ebff 100644
--- a/ChildProject/pipelines/zooniverse.py
+++ b/ChildProject/pipelines/zooniverse.py
@@ -342,10 +342,10 @@ def extract_chunks(
                         "%Y-%m-%d %H:%M:%S"
                     ),
                     "uploaded": False,
-                    "project_id": "",
+                    "project_id": pd.NA,
                     "subject_set": "",
                     "subject_set_id": pd.NA,
-                    "zooniverse_id": 0,
+                    "zooniverse_id": pd.NA,
                     "keyword": keyword,
                     "dataset": self.project.experiment
                 }
diff --git a/tests/test_zooniverse.py b/tests/test_zooniverse.py
index 282269e5..455411c2 100644
--- a/tests/test_zooniverse.py
+++ b/tests/test_zooniverse.py
@@ -8,6 +8,7 @@
 from ChildProject.pipelines.samplers import PeriodicSampler
 from ChildProject.pipelines.zooniverse import ZooniversePipeline, pad_interval
 from ChildProject.pipelines.fake_panoptes import LOCATION_FAIL
+from ChildProject.pipelines.zooniverse import CHUNKS_DTYPES
 
 
 def test_padding():
@@ -44,7 +45,7 @@ def test_extraction():
         spectrogram=True,
     )
 
-    chunks = pd.read_csv(chunks)
+    chunks = pd.read_csv(chunks, dtype=CHUNKS_DTYPES)
 
     assert len(chunks) == 2 * len(segments)
     assert all(
@@ -85,7 +86,7 @@ def test_uploading(location_fail, amount, ignore_errors, record_orphan, result):
     if amount is None: amount = 1000
     
     if not location_fail: df = df[df['mp3'] != LOCATION_FAIL]
-    
+
     df.to_csv(new_path,index=False)
     
     zooniverse.upload_chunks(new_path,
@@ -99,10 +100,10 @@ def test_uploading(location_fail, amount, ignore_errors, record_orphan, result):
                              test_endpoint=True if result != 'max_subjects_continue.csv' else 2,
                              )
     
-    truth = pd.read_csv(os.path.join('tests','truth','zoochunks',result))
-    #shutil.copy(new_path, os.path.join('tests','truth','zoochunks',result))
+    truth = pd.read_csv(os.path.join('tests','truth','zoochunks',result), dtype=CHUNKS_DTYPES)
+    # shutil.copy(new_path, os.path.join('tests','truth','zoochunks',result))
     
-    pd.testing.assert_frame_equal(truth, pd.read_csv(new_path), check_like=True)
+    pd.testing.assert_frame_equal(truth, pd.read_csv(new_path, dtype=CHUNKS_DTYPES), check_like=True)
    
 #might benefit from a test using invalid csv or/and a test with a csv having no orphan chunk     
 BASE_ORPHAN_CHUNKS = os.path.join('tests', 'data', 'chunks_test_orphan.csv')   
@@ -126,8 +127,9 @@ def test_link_orphan(ignore_errors, result):
                                     ignore_errors,
                                     test_endpoint=True,
                                     )
-   # shutil.copy(new_path, os.path.join('tests','truth','zoochunks',result))
-    pd.testing.assert_frame_equal(pd.read_csv(os.path.join('tests','truth','zoochunks',result)), pd.read_csv(new_path))
+    # shutil.copy(new_path, os.path.join('tests','truth','zoochunks',result))
+    pd.testing.assert_frame_equal(pd.read_csv(os.path.join('tests','truth','zoochunks',result), dtype=CHUNKS_DTYPES),
+                                  pd.read_csv(new_path, dtype=CHUNKS_DTYPES))
     
 @pytest.mark.parametrize("result", 
                          [('reset_orphan.csv'),
@@ -142,8 +144,9 @@ def test_reset_orphan(result):
     
     zooniverse.reset_orphan_subjects(new_path)
     
-    #shutil.copy(new_path, os.path.join('tests','truth','zoochunks',result))
-    pd.testing.assert_frame_equal(pd.read_csv(os.path.join('tests','truth','zoochunks',result)), pd.read_csv(new_path))
+    # shutil.copy(new_path, os.path.join('tests','truth','zoochunks',result))
+    pd.testing.assert_frame_equal(pd.read_csv(os.path.join('tests','truth','zoochunks',result), dtype=CHUNKS_DTYPES),
+                                  pd.read_csv(new_path, dtype=CHUNKS_DTYPES))
     
 def test_classification():
     pass
\ No newline at end of file