diff --git a/.github/workflows/test.yml b/.github/workflows/test.yml
index fb6df95..ae4b021 100644
--- a/.github/workflows/test.yml
+++ b/.github/workflows/test.yml
@@ -41,7 +41,22 @@ jobs:
         if: matrix.python-version == '3.15'
         run: sudo apt-get update && sudo apt-get install -y libglpk-dev swig
       - name: Install
-        run: pip install -e ".[dev,evotune]"
+        run: pip install -e ".[dev]"
 
       - name: Run pytest
         run: pytest -q
+
+  import-without-extras:
+    name: import without the optional extras
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v5
+      - uses: actions/setup-python@v6
+        with:
+          python-version: "3.12"
+          cache: pip
+          cache-dependency-path: pyproject.toml
+      - name: Install without extras
+        run: pip install -e .
+      - name: Import geckopy
+        run: python -W "error:Could not import matplotlib:UserWarning" -c "import geckopy"
diff --git a/CHANGELOG.md b/CHANGELOG.md
index b9dc4f6..617eaf4 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -2,6 +2,7 @@
 
 ## Unreleased
 
+* `cma` is a required dependency of geckopy and the extra `evotune` is removed. `import geckopy` imports `cma`.
 * `enzyme_usage` reports `measured_protein`, `demanded_protein` and `demand_ratio`, the usage of the enzymes with a measured concentration divided by their measured concentrations, and the same restricted to enzymes with a usage above 1e-7 mg/gDCW (`measured_protein_active`, `demand_ratio_active`). `report_enzyme_usage` carries `demand_ratio` and `demand_ratio_active`. GECKO's `getEnzymeUsage` and `reportEnzymeUsage` give the same values.
 * geckopy is licensed under the GNU General Public License v3 (GPL-3.0), as RAVEN and raven-toolbox. `LICENSE`, the `license` field and the classifier of `pyproject.toml`, and the README state it. Releases up to 4.0.0b2 were published under the MIT licence.
 * `load_prot_data` takes `zeros_as_measured`: by default an abundance of 0 is read as not measured (NaN) before any filter, also with `filter_data=False`, so the mean of a protein with a zero replicate leaves the zeros out; with `True` a 0 counts as a measured replicate, is kept in the mean and is exempt from `min_val` and `cut_lowest`, as GECKO's `loadProtData` option `zerosAsMeasured`.
diff --git a/README.md b/README.md
index ac07062..7f05af0 100644
--- a/README.md
+++ b/README.md
@@ -28,7 +28,7 @@ Both geckopy and raven-toolbox are pre-releases, so pip needs `--pre`:
 pip install --pre geckopy
 ```
 
-Extras: `geckopy[tutorial]` (matplotlib), `geckopy[evotune]` (cma, for kcat tuning) and `geckopy[dev]` (pytest and ruff). Python 3.11 or newer.
+Extras: `geckopy[tutorial]` (matplotlib), `geckopy[dev]` (pytest and ruff). Python 3.11 or newer.
 
 ## Quick start
 
diff --git a/docs/evotune_kcat_tuning.md b/docs/evotune_kcat_tuning.md
index e03fedc..0292ef8 100644
--- a/docs/evotune_kcat_tuning.md
+++ b/docs/evotune_kcat_tuning.md
@@ -41,8 +41,6 @@ least-trusted sources, has.
 
 - A working ecModel and `ModelAdapter` project (an existing
   `model_adapter.toml` and `models/*.yml`).
-- `pip install geckopy[evotune]` -- the CMA-ES search
-  (`cma`) is an optional dependency, not part of the base install.
 - Experimental data: at least one of a set of measured exchange fluxes
   per condition, or a set of measured maximum growth rates per carbon
   source. Both together works and is preferred -- flux data pins the
diff --git a/pyproject.toml b/pyproject.toml
index e31452d..96453fc 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -35,6 +35,7 @@ classifiers = [
     "Topic :: Scientific/Engineering :: Bio-Informatics",
 ]
 dependencies = [
+    "cma>=3.3",
     # >=0.31.1: earlier cobra versions carry a Reaction.__copy__/__deepcopy__
     # override that recurses infinitely (RecursionError) on Python 3.14 for
     # any model with a reaction<->metabolite reference cycle -- i.e. any real
@@ -68,9 +69,6 @@ dev = [
 tutorial = [
     "matplotlib>=3.5",
 ]
-evotune = [
-    "cma>=3.3",
-]
 
 [project.urls]
 Homepage = "https://github.com/CompMetEng/geckopy"
diff --git a/src/geckopy/kcat_tuning/__init__.py b/src/geckopy/kcat_tuning/__init__.py
index 59eac2e..71e9c33 100644
--- a/src/geckopy/kcat_tuning/__init__.py
+++ b/src/geckopy/kcat_tuning/__init__.py
@@ -14,29 +14,7 @@
 from .sigma_fitter import SigmaFitterResult, fit_sigma, sigma_fitter
 from .truncate_values import truncate_values
 
-try:
-    from . import evotune
-except ImportError as exc:  # pragma: no cover - exercised
-    # only when the optional `cma` dependency is absent.
-    # Bound to a module-level name: Python clears an `except ... as`
-    # target when the block exits, so the class below cannot close over
-    # `exc` itself.
-    _evotune_import_error = exc
-
-    class _MissingEvotune:
-        """Stand-in for the `evotune` subpackage when its
-        `cma` dependency isn't installed. Raises only on first attribute
-        access, not on import, so plain `geckopy.kcat_tuning`
-        stays usable without the optional extra."""
-
-        def __getattr__(self, name: str):
-            raise ImportError(
-                "Evotune: CMA-ES kcat tuning requires the optional "
-                "'evotune' extra: pip install "
-                "geckopy[evotune]"
-            ) from _evotune_import_error
-
-    evotune = _MissingEvotune()  # type: ignore[assignment]
+from . import evotune
 
 __all__ = [
     "Correction",
diff --git a/src/geckopy/kcat_tuning/evotune/__init__.py b/src/geckopy/kcat_tuning/evotune/__init__.py
index 7243d87..abb3672 100644
--- a/src/geckopy/kcat_tuning/evotune/__init__.py
+++ b/src/geckopy/kcat_tuning/evotune/__init__.py
@@ -4,13 +4,6 @@
 :func:`~.tuning.screen_kcat_leverage`, then
 :func:`~.tuning.select_tunable_mask`, then
 :func:`~.tuning.cmaes_kcat_tuning`.
-
-Modules in this package that need ``cma`` import it unconditionally --
-it's a hard requirement here, gated one level up instead (see
-``kcat_tuning/__init__.py``) so importing plain
-``geckopy.kcat_tuning`` never requires the optional
-``evotune`` extra. Install it with
-``pip install geckopy[evotune]``.
 """
 from .data import EvotuneData, load_evotune_data
 from .distance import (
diff --git a/src/geckopy/kcat_tuning/evotune/tuning.py b/src/geckopy/kcat_tuning/evotune/tuning.py
index 9585e68..3d68b09 100644
--- a/src/geckopy/kcat_tuning/evotune/tuning.py
+++ b/src/geckopy/kcat_tuning/evotune/tuning.py
@@ -57,12 +57,12 @@
 from __future__ import annotations
 
 import logging
+import warnings
 from contextlib import nullcontext
 from dataclasses import dataclass, field
 from itertools import combinations
 from typing import TYPE_CHECKING, Callable, Optional, Sequence
 
-import cma
 import cobra
 import numpy as np
 import pandas as pd
@@ -77,6 +77,11 @@
 from .simulate import simulate_evotune_dataset
 from .tying import isozyme_tie_map
 
+with warnings.catch_warnings():
+    # cma warns when matplotlib is absent; the plotting functions of cma are not used here.
+    warnings.filterwarnings("ignore", message="Could not import matplotlib", category=UserWarning)
+    import cma
+
 if TYPE_CHECKING:
     from ...adapter import ModelAdapter
     from ...adapter.params import EvotuneParams
diff --git a/tutorials/full_ecModel/protocol.py b/tutorials/full_ecModel/protocol.py
index f673bc0..6ee2645 100644
--- a/tutorials/full_ecModel/protocol.py
+++ b/tutorials/full_ecModel/protocol.py
@@ -270,8 +270,7 @@
 # %% [markdown]
 # Besides step-wise sensitivity tuning, kcats can be fitted to
 # experimental growth rates and exchange fluxes with CMA-ES
-# (`geckopy.kcat_tuning.evotune`, installed with the `evotune` extra;
-# see docs/evotune_kcat_tuning.md):
+# (`geckopy.kcat_tuning.evotune`; see docs/evotune_kcat_tuning.md):
 #
 # ```python
 # from geckopy.kcat_tuning.evotune import (