diff --git a/doc/schema_template.md b/doc/schema_template.md
index 679d839d3f92..d08c7a4a0d06 100644
--- a/doc/schema_template.md
+++ b/doc/schema_template.md
@@ -82,7 +82,7 @@ $tag:tool|outputs|data|actions|action://complexType[@name='Action']
 $tag:tool|outputs|data|actions|action|option://complexType[@name='ActionsOption']
 $tag:tool|outputs|data|discover_datasets://complexType[@name='OutputDiscoverDatasets']
 $tag:tool|outputs|collection://complexType[@name='OutputCollection']
-$tag:tool|outputs|collection|data://complexType[@name='OutputCollectionDataElement']
+$tag:tool|outputs|collection|data://complexType[@name='OutputCollectionData']
 $tag:tool|outputs|collection|filter://complexType[@name='OutputFilter']
 $tag:tool|outputs|collection|discover_datasets://complexType[@name='OutputCollectionDiscoverDatasets']
 $tag:tool|tests://complexType[@name='Tests']
diff --git a/lib/galaxy/tool_util/xsd/galaxy.xsd b/lib/galaxy/tool_util/xsd/galaxy.xsd
index 73558250251f..7c17312c4f0b 100644
--- a/lib/galaxy/tool_util/xsd/galaxy.xsd
+++ b/lib/galaxy/tool_util/xsd/galaxy.xsd
@@ -5827,15 +5827,24 @@ on Human (hg18)``.
     <xs:attributeGroup ref="OutputDataAttributes"/>
   </xs:complexType>
 
+  <!-- Allowed tags included in collection-data -->
+  <xs:group name="OutputCollectionDataElement">
+    <xs:choice>
+      <xs:element name="actions" type="Actions" />
+      <xs:element name="change_format" type="ChangeFormat" />
+    </xs:choice>
+  </xs:group>
+
+  <!-- Allowed tags included in collection -->
   <xs:group name="OutputCollectionElement">
     <xs:choice>
-      <xs:element name="data" type="OutputCollectionDataElement" />
+      <xs:element name="data" type="OutputCollectionData" />
       <xs:element name="discover_datasets" type="OutputCollectionDiscoverDatasets" />
       <xs:element name="filter" type="OutputFilter" />
     </xs:choice>
   </xs:group>
 
-  <xs:complexType name="OutputCollectionDataElement">
+  <xs:complexType name="OutputCollectionData">
     <xs:annotation>
       <xs:documentation xml:lang="en"><![CDATA[
 
@@ -5844,7 +5853,7 @@ define the elements of a collection statically. See also [Planemo's documentatio
 ]]></xs:documentation>
     </xs:annotation>
     <xs:sequence>
-      <xs:group ref="OutputCollectionElement" minOccurs="0" maxOccurs="unbounded" />
+      <xs:group ref="OutputCollectionDataElement" minOccurs="0" maxOccurs="unbounded" />
     </xs:sequence>
     <xs:attributeGroup ref="OutputCommon" />
     <xs:attributeGroup ref="OutputDataAttributes" />
@@ -6002,23 +6011,7 @@ Therefore a filter for such a variable looks like the following example.
     </xs:simpleContent>
   </xs:complexType>
 
-  <xs:complexType name="OutputDiscoverDatasets">
-     <xs:annotation>
-      <xs:documentation xml:lang="en"><![CDATA[
-
-Describe datasets to dynamically collect after the job complete.
-
-There are many simple tools with examples of this element distributed with
-Galaxy, including:
-
-* [multi_output.xml](https://github.com/galaxyproject/galaxy/blob/dev/test/functional/tools/multi_output.xml)
-* [multi_output_assign_primary.xml](https://github.com/galaxyproject/galaxy/blob/dev/test/functional/tools/multi_output_assign_primary.xml)
-* [multi_output_configured.xml](https://github.com/galaxyproject/galaxy/blob/dev/test/functional/tools/multi_output_configured.xml)
-
-More information can be found on Planemo's documentation for
-[multiple output files](https://planemo.readthedocs.io/en/latest/writing_advanced.html#multiple-output-files).
-]]></xs:documentation>
-    </xs:annotation>
+  <xs:attributeGroup name="OutputDiscoverDatasetsCommon">
     <xs:attribute name="from_provided_metadata" type="xs:boolean" use="optional">
       <xs:annotation>
         <xs:documentation xml:lang="en">Indicate that dataset filenames should simply be read from the provided metadata file (e.g. galaxy.json). If this is set - pattern and sort must not be set.</xs:documentation>
@@ -6069,6 +6062,26 @@ More information can be found on Planemo's documentation for
         <xs:documentation xml:lang="en">Indication if this dataset is visible in output history. This defaults to ``false``, but probably shouldn't - be sure to set to ``true`` if that is your intention.</xs:documentation>
       </xs:annotation>
     </xs:attribute>
+  </xs:attributeGroup>
+
+  <xs:complexType name="OutputDiscoverDatasets">
+     <xs:annotation>
+      <xs:documentation xml:lang="en"><![CDATA[
+
+Describe datasets to dynamically collect after the job complete.
+
+There are many simple tools with examples of this element distributed with
+Galaxy, including:
+
+* [multi_output.xml](https://github.com/galaxyproject/galaxy/blob/dev/test/functional/tools/multi_output.xml)
+* [multi_output_assign_primary.xml](https://github.com/galaxyproject/galaxy/blob/dev/test/functional/tools/multi_output_assign_primary.xml)
+* [multi_output_configured.xml](https://github.com/galaxyproject/galaxy/blob/dev/test/functional/tools/multi_output_configured.xml)
+
+More information can be found on Planemo's documentation for
+[multiple output files](https://planemo.readthedocs.io/en/latest/writing_advanced.html#multiple-output-files).
+]]></xs:documentation>
+    </xs:annotation>
+    <xs:attributeGroup ref="OutputDiscoverDatasetsCommon"/>
     <xs:attribute name="assign_primary_output" type="xs:boolean" use="optional">
       <xs:annotation>
         <xs:documentation xml:lang="en">Replace the primary dataset described by the parameter ``data`` parameter with the first output discovered.</xs:documentation>
@@ -6092,51 +6105,7 @@ Galaxy, including:
 
 ]]></xs:documentation>
     </xs:annotation>
-    <xs:attribute name="from_provided_metadata" type="xs:boolean" use="optional">
-      <xs:annotation>
-        <xs:documentation xml:lang="en">Indicate that dataset filenames should simply be read from the provided metadata file (e.g. galaxy.json). If this is set - pattern and sort_by must not be set.</xs:documentation>
-      </xs:annotation>
-    </xs:attribute>
-    <xs:attribute name="pattern" type="xs:string" use="optional">
-      <xs:annotation>
-        <xs:documentation xml:lang="en">Regular expression used to find filenames and parse dynamic properties.</xs:documentation>
-      </xs:annotation>
-    </xs:attribute>
-    <xs:attribute name="directory" type="xs:string" use="optional">
-      <xs:annotation>
-        <xs:documentation xml:lang="en">Directory (relative to working directory) to search for files.</xs:documentation>
-      </xs:annotation>
-    </xs:attribute>
-    <xs:attribute name="recurse" type="xs:boolean" use="optional">
-      <xs:annotation>
-        <xs:documentation xml:lang="en">Indicates that the specified directory should be searched recursively for matching files.</xs:documentation>
-      </xs:annotation>
-    </xs:attribute>
-    <xs:attribute name="match_relative_path" type="xs:boolean" use="optional">
-      <xs:annotation>
-        <xs:documentation xml:lang="en">Indicates that the entire path of the discovered dataset relative to the specified directory should be available for matching patterns.</xs:documentation>
-      </xs:annotation>
-    </xs:attribute>
-    <xs:attribute name="format" type="xs:string" use="optional">
-      <xs:annotation>
-        <xs:documentation xml:lang="en">Format (or datatype) of discovered datasets (an alias with ``ext``).</xs:documentation>
-      </xs:annotation>
-    </xs:attribute>
-    <xs:attribute name="ext" type="xs:string" use="optional">
-      <xs:annotation>
-        <xs:documentation xml:lang="en">Format (or datatype) of discovered datasets (an alias with ``format``).</xs:documentation>
-      </xs:annotation>
-    </xs:attribute>
-    <xs:attribute name="sort_by" type="xs:string" use="optional">
-      <xs:annotation>
-        <xs:documentation xml:lang="en">A string `[reverse_][SORT_COMP_]SORTBY` describing the desired sort order of the collection elements. `SORTBY` can be `filename`, `name`, `designation`, `dbkey` and the optional `SORT_COMP` can be either `lexical` or `numeric`. Default is lexical sorting by filename.</xs:documentation>
-      </xs:annotation>
-    </xs:attribute>
-    <xs:attribute name="visible" type="xs:boolean" use="optional">
-      <xs:annotation>
-        <xs:documentation xml:lang="en">Indication if this dataset is visible in the history. This defaults to ``false``, but probably shouldn't - be sure to set to ``true`` if that is your intention.</xs:documentation>
-      </xs:annotation>
-    </xs:attribute>
+    <xs:attributeGroup ref="OutputDiscoverDatasetsCommon"/>
   </xs:complexType>
   <xs:complexType name="Actions">
     <xs:annotation>
diff --git a/test/functional/tools/collection_creates_pair_format.xml b/test/functional/tools/collection_creates_pair_format.xml
index 549f49cd7b84..f3dea456fd2d 100644
--- a/test/functional/tools/collection_creates_pair_format.xml
+++ b/test/functional/tools/collection_creates_pair_format.xml
@@ -7,16 +7,28 @@
   </command>
   <inputs>
     <param name="input1" type="data" label="Input" help="Input to be split." />
+    <param name="out_format" type="select" label="Output data type">
+      <option value="">do not change</option>
+      <option value="interval">interval</option>
+  </param>
   </inputs>
   <outputs>
     <!-- format is specified separately for the elements, i.e. default format (fasta in this case) is ignored -->
     <collection name="paired_output" format="fasta" type="paired" label="Split Pair">
-      <data name="forward" format="txt" />
+      <data name="forward" format="txt">
+        <change_format>
+          <when input="out_format" value="interval" format="interval" />
+        </change_format>
+      </data>
       <data name="reverse" format_source="input1" from_work_dir="reverse.txt" />
     </collection>
     <!-- no format is specified separately for the elements, i.e. default format (fasta in this case) is used -->
     <collection name="paired_output_default_format" format="fasta" type="paired" label="Split Pair">
-      <data name="forward" />
+      <data name="forward">
+        <change_format>
+          <when input="out_format" value="interval" format="interval" />
+        </change_format>
+      </data>
       <data name="reverse" from_work_dir="reverse.txt" />
     </collection>
   </outputs>
@@ -48,5 +60,33 @@
         </element>
       </output_collection>
     </test>
+    <test>
+      <param name="input1" ftype="bed" value="simple_lines_interleaved.txt" />
+      <param name="out_format" value="interval"/>
+      <output_collection name="paired_output" type="paired">
+        <element name="forward" ftype="interval">
+          <assert_contents>
+            <has_text_matching expression="^This is a line of text.\nThis is a line of text.\n$" />
+          </assert_contents>
+        </element>
+        <element name="reverse" ftype="bed">
+          <assert_contents>
+            <has_text_matching expression="^This is a different line of text.\nThis is a different line of text.\n$" />
+          </assert_contents>
+        </element>
+      </output_collection>
+      <output_collection name="paired_output_default_format" type="paired">
+        <element name="forward" ftype="interval">
+          <assert_contents>
+            <has_text_matching expression="^This is a line of text.\nThis is a line of text.\n$" />
+          </assert_contents>
+        </element>
+        <element name="reverse" ftype="fasta">
+          <assert_contents>
+            <has_text_matching expression="^This is a different line of text.\nThis is a different line of text.\n$" />
+          </assert_contents>
+        </element>
+      </output_collection>
+    </test>
   </tests>
 </tool>
diff --git a/test/functional/tools/metadata_column_names.xml b/test/functional/tools/metadata_column_names.xml
index c7edd8f32634..918042f52426 100644
--- a/test/functional/tools/metadata_column_names.xml
+++ b/test/functional/tools/metadata_column_names.xml
@@ -1,7 +1,9 @@
 <tool id="metadata_columns" name="metadata_columns" version="1.0.0">
     <description>Tests whether metadata is being set correctly.</description>
     <command><![CDATA[
-        cp '$input' '$output'
+        cp '$input' '$output';
+        cp '$input' '$paired_output.forward' ;
+        cp '$input' '$paired_output.reverse' ;
     ]]></command>
     <inputs>
         <param name="input" type="data" multiple="false" />
@@ -12,6 +14,18 @@
                 <action name="column_names" type="metadata" default="First,${input.name}" />
             </actions>
         </data>
+        <collection name="paired_output" format="tabular" type="paired" label="Split Pair">
+            <data name="forward">
+                <actions>
+                    <action name="dbkey" type="metadata" default="hg38" />
+                </actions>
+            </data>
+            <data name="reverse">
+                <actions>
+                    <action name="column_names" type="metadata" default="A,B,C" />
+                </actions>
+            </data>
+    </collection>
     </outputs>
     <tests>
         <test>
@@ -19,6 +33,14 @@
             <output name="output">
                 <metadata name="column_names" value="First,2.tabular"/>
             </output>
+            <output_collection name="paired_output" type="paired">
+                <element name="forward" ftype="tabular" value="2.tabular">
+                    <metadata name="dbkey" value="hg38"/>
+                </element>
+                <element name="reverse" ftype="tabular" value="2.tabular">
+                    <metadata name="column_names" value="A,B,C"/>
+                </element>
+              </output_collection>
         </test>
     </tests>
 </tool>