diff --git a/CHANGELOG.md b/CHANGELOG.md
index 0b05600b..186c58c7 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -23,6 +23,8 @@ any parts of the framework not mentioned in the documentation should generally b
 * Avoided that an empty attributes dict is rendered in case serializer does not
   provide any attribute fields.
 * Avoided shadowing of exception when rendering errors (regression since 4.3.0).
+* Ensured that sparse fields only applies when rendering, not when parsing.
+* Adjusted that sparse fields properly removes meta fields when not defined.
 
 ### Removed
 
diff --git a/example/tests/integration/test_non_paginated_responses.py b/example/tests/integration/test_non_paginated_responses.py
index 92d26de3..6434b7a7 100644
--- a/example/tests/integration/test_non_paginated_responses.py
+++ b/example/tests/integration/test_non_paginated_responses.py
@@ -14,7 +14,7 @@ def test_multiple_entries_no_pagination(multiple_entries, client):
     expected = {
         "data": [
             {
-                "type": "posts",
+                "type": "entries",
                 "id": "1",
                 "attributes": {
                     "headline": multiple_entries[0].headline,
@@ -70,7 +70,7 @@ def test_multiple_entries_no_pagination(multiple_entries, client):
                 },
             },
             {
-                "type": "posts",
+                "type": "entries",
                 "id": "2",
                 "attributes": {
                     "headline": multiple_entries[1].headline,
diff --git a/example/tests/integration/test_pagination.py b/example/tests/integration/test_pagination.py
index 1a4bd056..0f5ac17e 100644
--- a/example/tests/integration/test_pagination.py
+++ b/example/tests/integration/test_pagination.py
@@ -14,7 +14,7 @@ def test_pagination_with_single_entry(single_entry, client):
     expected = {
         "data": [
             {
-                "type": "posts",
+                "type": "entries",
                 "id": "1",
                 "attributes": {
                     "headline": single_entry.headline,
diff --git a/example/tests/integration/test_sparse_fieldsets.py b/example/tests/integration/test_sparse_fieldsets.py
index 605d218d..cf9cee20 100644
--- a/example/tests/integration/test_sparse_fieldsets.py
+++ b/example/tests/integration/test_sparse_fieldsets.py
@@ -15,6 +15,7 @@ def test_sparse_fieldset_valid_fields(client, entry):
     entry = data[0]
     assert entry["attributes"].keys() == {"headline"}
     assert entry["relationships"].keys() == {"blog"}
+    assert "meta" not in entry
 
 
 @pytest.mark.parametrize(
diff --git a/example/tests/test_filters.py b/example/tests/test_filters.py
index 87f9d059..8e45ded1 100644
--- a/example/tests/test_filters.py
+++ b/example/tests/test_filters.py
@@ -470,7 +470,7 @@ def test_search_keywords(self):
         expected_result = {
             "data": [
                 {
-                    "type": "posts",
+                    "type": "entries",
                     "id": "7",
                     "attributes": {
                         "headline": "ANTH3868X",
diff --git a/example/tests/unit/test_renderer_class_methods.py b/example/tests/unit/test_renderer_class_methods.py
index 6e7c9ea1..838f6819 100644
--- a/example/tests/unit/test_renderer_class_methods.py
+++ b/example/tests/unit/test_renderer_class_methods.py
@@ -102,8 +102,8 @@ def test_extract_attributes():
     assert sorted(JSONRenderer.extract_attributes(fields, resource)) == sorted(
         expected
     ), "Regular fields should be extracted"
-    assert sorted(JSONRenderer.extract_attributes(fields, {})) == sorted(
-        {"username": ""}
+    assert (
+        JSONRenderer.extract_attributes(fields, {}) == {}
     ), "Should not extract read_only fields on empty serializer"
 
 
diff --git a/example/views.py b/example/views.py
index 9c949684..da171698 100644
--- a/example/views.py
+++ b/example/views.py
@@ -112,7 +112,10 @@ class BlogCustomViewSet(JsonApiViewSet):
 
 class EntryViewSet(ModelViewSet):
     queryset = Entry.objects.all()
-    resource_name = "posts"
+    # TODO it should not be supported to overwrite resource name
+    # of endpoints with serializers as includes and sparse fields
+    # cannot be aware of it
+    # resource_name = "posts"
 
     def get_serializer_class(self):
         return EntrySerializer
diff --git a/rest_framework_json_api/renderers.py b/rest_framework_json_api/renderers.py
index 639f0b11..8c632f6a 100644
--- a/rest_framework_json_api/renderers.py
+++ b/rest_framework_json_api/renderers.py
@@ -17,13 +17,26 @@
 from rest_framework.settings import api_settings
 
 import rest_framework_json_api
-from rest_framework_json_api import utils
 from rest_framework_json_api.relations import (
     HyperlinkedMixin,
     ManySerializerMethodResourceRelatedField,
     ResourceRelatedField,
     SkipDataMixin,
 )
+from rest_framework_json_api.utils import (
+    format_errors,
+    format_field_name,
+    format_field_names,
+    get_included_resources,
+    get_related_resource_type,
+    get_relation_instance,
+    get_resource_id,
+    get_resource_name,
+    get_resource_type_from_instance,
+    get_resource_type_from_serializer,
+    get_serializer_fields,
+    is_relationship_field,
+)
 
 
 class JSONRenderer(renderers.JSONRenderer):
@@ -57,31 +70,20 @@ class JSONRenderer(renderers.JSONRenderer):
     def extract_attributes(cls, fields, resource):
         """
         Builds the `attributes` object of the JSON:API resource object.
-        """
-        data = {}
-        for field_name, field in iter(fields.items()):
-            # ID is always provided in the root of JSON:API so remove it from attributes
-            if field_name == "id":
-                continue
-            # don't output a key for write only fields
-            if fields[field_name].write_only:
-                continue
-            # Skip fields with relations
-            if utils.is_relationship_field(field):
-                continue
 
-            # Skip read_only attribute fields when `resource` is an empty
-            # serializer. Prevents the "Raw Data" form of the browsable API
-            # from rendering `"foo": null` for read only fields
-            try:
-                resource[field_name]
-            except KeyError:
-                if fields[field_name].read_only:
-                    continue
+        Ensures that ID which is always provided in a JSON:API resource object
+        and relationships are not returned.
+        """
 
-            data.update({field_name: resource.get(field_name)})
+        invalid_fields = {"id", api_settings.URL_FIELD_NAME}
 
-        return utils.format_field_names(data)
+        return {
+            format_field_name(field_name): value
+            for field_name, value in resource.items()
+            if field_name in fields
+            and field_name not in invalid_fields
+            and not is_relationship_field(fields[field_name])
+        }
 
     @classmethod
     def extract_relationships(cls, fields, resource, resource_instance):
@@ -107,14 +109,14 @@ def extract_relationships(cls, fields, resource, resource_instance):
                 continue
 
             # Skip fields without relations
-            if not utils.is_relationship_field(field):
+            if not is_relationship_field(field):
                 continue
 
             source = field.source
-            relation_type = utils.get_related_resource_type(field)
+            relation_type = get_related_resource_type(field)
 
             if isinstance(field, relations.HyperlinkedIdentityField):
-                resolved, relation_instance = utils.get_relation_instance(
+                resolved, relation_instance = get_relation_instance(
                     resource_instance, source, field.parent
                 )
                 if not resolved:
@@ -166,7 +168,7 @@ def extract_relationships(cls, fields, resource, resource_instance):
                 field,
                 (relations.PrimaryKeyRelatedField, relations.HyperlinkedRelatedField),
             ):
-                resolved, relation = utils.get_relation_instance(
+                resolved, relation = get_relation_instance(
                     resource_instance, f"{source}_id", field.parent
                 )
                 if not resolved:
@@ -189,7 +191,7 @@ def extract_relationships(cls, fields, resource, resource_instance):
                 continue
 
             if isinstance(field, relations.ManyRelatedField):
-                resolved, relation_instance = utils.get_relation_instance(
+                resolved, relation_instance = get_relation_instance(
                     resource_instance, source, field.parent
                 )
                 if not resolved:
@@ -222,9 +224,7 @@ def extract_relationships(cls, fields, resource, resource_instance):
                 for nested_resource_instance in relation_instance:
                     nested_resource_instance_type = (
                         relation_type
-                        or utils.get_resource_type_from_instance(
-                            nested_resource_instance
-                        )
+                        or get_resource_type_from_instance(nested_resource_instance)
                     )
 
                     relation_data.append(
@@ -243,7 +243,7 @@ def extract_relationships(cls, fields, resource, resource_instance):
                 )
                 continue
 
-        return utils.format_field_names(data)
+        return format_field_names(data)
 
     @classmethod
     def extract_relation_instance(cls, field, resource_instance):
@@ -289,7 +289,7 @@ def extract_included(
                 continue
 
             # Skip fields without relations
-            if not utils.is_relationship_field(field):
+            if not is_relationship_field(field):
                 continue
 
             try:
@@ -341,7 +341,7 @@ def extract_included(
 
             if isinstance(field, ListSerializer):
                 serializer = field.child
-                relation_type = utils.get_resource_type_from_serializer(serializer)
+                relation_type = get_resource_type_from_serializer(serializer)
                 relation_queryset = list(relation_instance)
 
                 if serializer_data:
@@ -350,11 +350,9 @@ def extract_included(
                         nested_resource_instance = relation_queryset[position]
                         resource_type = (
                             relation_type
-                            or utils.get_resource_type_from_instance(
-                                nested_resource_instance
-                            )
+                            or get_resource_type_from_instance(nested_resource_instance)
                         )
-                        serializer_fields = utils.get_serializer_fields(
+                        serializer_fields = get_serializer_fields(
                             serializer.__class__(
                                 nested_resource_instance, context=serializer.context
                             )
@@ -378,10 +376,10 @@ def extract_included(
                         )
 
             if isinstance(field, Serializer):
-                relation_type = utils.get_resource_type_from_serializer(field)
+                relation_type = get_resource_type_from_serializer(field)
 
                 # Get the serializer fields
-                serializer_fields = utils.get_serializer_fields(field)
+                serializer_fields = get_serializer_fields(field)
                 if serializer_data:
                     new_item = cls.build_json_resource_obj(
                         serializer_fields,
@@ -414,7 +412,8 @@ def extract_meta(cls, serializer, resource):
         meta_fields = getattr(meta, "meta_fields", [])
         data = {}
         for field_name in meta_fields:
-            data.update({field_name: resource.get(field_name)})
+            if field_name in resource:
+                data.update({field_name: resource[field_name]})
         return data
 
     @classmethod
@@ -434,6 +433,24 @@ def extract_root_meta(cls, serializer, resource):
             data.update(json_api_meta)
         return data
 
+    @classmethod
+    def _filter_sparse_fields(cls, serializer, fields, resource_name):
+        request = serializer.context.get("request")
+        if request:
+            sparse_fieldset_query_param = f"fields[{resource_name}]"
+            sparse_fieldset_value = request.query_params.get(
+                sparse_fieldset_query_param
+            )
+            if sparse_fieldset_value:
+                sparse_fields = sparse_fieldset_value.split(",")
+                return {
+                    field_name: field
+                    for field_name, field, in fields.items()
+                    if field_name in sparse_fields
+                }
+
+        return fields
+
     @classmethod
     def build_json_resource_obj(
         cls,
@@ -449,11 +466,15 @@ def build_json_resource_obj(
         """
         # Determine type from the instance if the underlying model is polymorphic
         if force_type_resolution:
-            resource_name = utils.get_resource_type_from_instance(resource_instance)
+            resource_name = get_resource_type_from_instance(resource_instance)
         resource_data = {
             "type": resource_name,
-            "id": utils.get_resource_id(resource_instance, resource),
+            "id": get_resource_id(resource_instance, resource),
         }
+
+        # TODO remove this filter by rewriting extract_relationships
+        # so it uses the serialized data as a basis
+        fields = cls._filter_sparse_fields(serializer, fields, resource_name)
         attributes = cls.extract_attributes(fields, resource)
         if attributes:
             resource_data["attributes"] = attributes
@@ -468,7 +489,7 @@ def build_json_resource_obj(
 
         meta = cls.extract_meta(serializer, resource)
         if meta:
-            resource_data["meta"] = utils.format_field_names(meta)
+            resource_data["meta"] = format_field_names(meta)
 
         return resource_data
 
@@ -485,7 +506,7 @@ def render_relationship_view(
 
     def render_errors(self, data, accepted_media_type=None, renderer_context=None):
         return super().render(
-            utils.format_errors(data), accepted_media_type, renderer_context
+            format_errors(data), accepted_media_type, renderer_context
         )
 
     def render(self, data, accepted_media_type=None, renderer_context=None):
@@ -495,7 +516,7 @@ def render(self, data, accepted_media_type=None, renderer_context=None):
         request = renderer_context.get("request", None)
 
         # Get the resource name.
-        resource_name = utils.get_resource_name(renderer_context)
+        resource_name = get_resource_name(renderer_context)
 
         # If this is an error response, skip the rest.
         if resource_name == "errors":
@@ -531,7 +552,7 @@ def render(self, data, accepted_media_type=None, renderer_context=None):
 
         serializer = getattr(serializer_data, "serializer", None)
 
-        included_resources = utils.get_included_resources(request, serializer)
+        included_resources = get_included_resources(request, serializer)
 
         if serializer is not None:
             # Extract root meta for any type of serializer
@@ -558,7 +579,7 @@ def render(self, data, accepted_media_type=None, renderer_context=None):
                     else:
                         resource_serializer_class = serializer.child
 
-                    fields = utils.get_serializer_fields(resource_serializer_class)
+                    fields = get_serializer_fields(resource_serializer_class)
                     force_type_resolution = getattr(
                         resource_serializer_class, "_poly_force_type_resolution", False
                     )
@@ -581,7 +602,7 @@ def render(self, data, accepted_media_type=None, renderer_context=None):
                         included_cache,
                     )
             else:
-                fields = utils.get_serializer_fields(serializer)
+                fields = get_serializer_fields(serializer)
                 force_type_resolution = getattr(
                     serializer, "_poly_force_type_resolution", False
                 )
@@ -640,7 +661,7 @@ def render(self, data, accepted_media_type=None, renderer_context=None):
                     )
 
         if json_api_meta:
-            render_data["meta"] = utils.format_field_names(json_api_meta)
+            render_data["meta"] = format_field_names(json_api_meta)
 
         return super().render(render_data, accepted_media_type, renderer_context)
 
@@ -690,7 +711,6 @@ def get_includes_form(self, view):
                 serializer_class = view.get_serializer_class()
         except AttributeError:
             return
-
         if not hasattr(serializer_class, "included_serializers"):
             return
 
diff --git a/rest_framework_json_api/serializers.py b/rest_framework_json_api/serializers.py
index c680f60a..66650caf 100644
--- a/rest_framework_json_api/serializers.py
+++ b/rest_framework_json_api/serializers.py
@@ -75,35 +75,32 @@ class SparseFieldsetsMixin:
     Specification: https://jsonapi.org/format/#fetching-sparse-fieldsets
     """
 
-    def __init__(self, *args, **kwargs):
-        super().__init__(*args, **kwargs)
-        context = kwargs.get("context")
-        request = context.get("request") if context else None
+    @property
+    def _readable_fields(self):
+        request = self.context.get("request") if self.context else None
+        readable_fields = super()._readable_fields
 
         if request:
-            sparse_fieldset_query_param = "fields[{}]".format(
-                get_resource_type_from_serializer(self)
-            )
             try:
-                param_name = next(
-                    key
-                    for key in request.query_params
-                    if sparse_fieldset_query_param == key
+                resource_type = get_resource_type_from_serializer(self)
+                sparse_fieldset_query_param = f"fields[{resource_type}]"
+
+                sparse_fieldset_value = request.query_params.get(
+                    sparse_fieldset_query_param
                 )
-            except StopIteration:
+                if sparse_fieldset_value:
+                    sparse_fields = sparse_fieldset_value.split(",")
+                    return (
+                        field
+                        for field in readable_fields
+                        if field.field_name in sparse_fields
+                        or field.field_name == api_settings.URL_FIELD_NAME
+                    )
+            except AttributeError:
+                # no type on serializer, must be used only as only nested
                 pass
-            else:
-                fieldset = request.query_params.get(param_name).split(",")
-                # iterate over a *copy* of self.fields' underlying dict, because we may
-                # modify the original during the iteration.
-                # self.fields is a `rest_framework.utils.serializer_helpers.BindingDict`
-                for field_name, _field in self.fields.fields.copy().items():
-                    if (
-                        field_name == api_settings.URL_FIELD_NAME
-                    ):  # leave self link there
-                        continue
-                    if field_name not in fieldset:
-                        self.fields.pop(field_name)
+
+        return readable_fields
 
 
 class IncludedResourcesValidationMixin:
diff --git a/tests/serializers.py b/tests/serializers.py
index 4159039d..ddf28f98 100644
--- a/tests/serializers.py
+++ b/tests/serializers.py
@@ -26,7 +26,10 @@ class ForeignKeySourceSerializer(serializers.ModelSerializer):
 
     class Meta:
         model = ForeignKeySource
-        fields = ("target",)
+        fields = (
+            "name",
+            "target",
+        )
 
 
 class ManyToManyTargetSerializer(serializers.ModelSerializer):
diff --git a/tests/test_relations.py b/tests/test_relations.py
index ad4bebcb..5f8883be1 100644
--- a/tests/test_relations.py
+++ b/tests/test_relations.py
@@ -39,7 +39,9 @@ def test_serialize(
         settings.JSON_API_FORMAT_TYPES = format_type
         settings.JSON_API_PLURALIZE_TYPES = pluralize_type
 
-        serializer = ForeignKeySourceSerializer(instance={"target": foreign_key_target})
+        serializer = ForeignKeySourceSerializer(
+            instance={"target": foreign_key_target, "name": "Test"}
+        )
         expected = {
             "type": resource_type,
             "id": str(foreign_key_target.pk),
@@ -85,7 +87,10 @@ def test_deserialize(
         settings.JSON_API_PLURALIZE_TYPES = pluralize_type
 
         serializer = ForeignKeySourceSerializer(
-            data={"target": {"type": resource_type, "id": str(foreign_key_target.pk)}}
+            data={
+                "target": {"type": resource_type, "id": str(foreign_key_target.pk)},
+                "name": "Test",
+            }
         )
 
         assert serializer.is_valid()
@@ -191,7 +196,9 @@ def test_deserialize_many_to_many_relation(
         ],
     )
     def test_invalid_resource_id_object(self, resource_identifier, error):
-        serializer = ForeignKeySourceSerializer(data={"target": resource_identifier})
+        serializer = ForeignKeySourceSerializer(
+            data={"target": resource_identifier, "name": "Test"}
+        )
         assert not serializer.is_valid()
         assert serializer.errors == {"target": [error]}
 
diff --git a/tests/test_views.py b/tests/test_views.py
index acba7e66..468c2cbd 100644
--- a/tests/test_views.py
+++ b/tests/test_views.py
@@ -237,6 +237,43 @@ def test_delete(self, client, model):
         assert BasicModel.objects.count() == 0
         assert len(response.rendered_content) == 0
 
+    @pytest.mark.urls(__name__)
+    def test_create_with_sparse_fields(self, client, foreign_key_target):
+        url = reverse("foreign-key-source-list")
+        data = {
+            "data": {
+                "id": None,
+                "type": "ForeignKeySource",
+                "attributes": {"name": "Test"},
+                "relationships": {
+                    "target": {
+                        "data": {
+                            "id": str(foreign_key_target.pk),
+                            "type": "ForeignKeyTarget",
+                        }
+                    }
+                },
+            }
+        }
+        response = client.post(f"{url}?fields[ForeignKeySource]=target", data=data)
+        assert response.status_code == status.HTTP_201_CREATED
+        foreign_key_source = ForeignKeySource.objects.first()
+        assert foreign_key_source.name == "Test"
+        assert response.json() == {
+            "data": {
+                "id": str(foreign_key_source.pk),
+                "type": "ForeignKeySource",
+                "relationships": {
+                    "target": {
+                        "data": {
+                            "id": str(foreign_key_target.pk),
+                            "type": "ForeignKeyTarget",
+                        }
+                    }
+                },
+            }
+        }
+
 
 class TestReadonlyModelViewSet:
     @pytest.mark.parametrize(