Annotator causing squished plots, am I missing something?

[simbru]

I'm finding that sometimes if I have the Annotator set with loc='outside', it will for a lack of a better word "squish" the y axis. I've had similar experiences with horizontally oriented boxplots on the x axis too. I am wondering if there is something that I'm obviously doing wrong?

I cant share the data as its proprietary but I'm sure I could create a minimally reproducible example with demo datasets if its necessary, just let me know. Was hoping perhaps there is something obviously wrong in the code logic below.

Normal (no annotations):

Normal (with annotations): Squsihed (with annotations):

# data load and prep above, looping through parts of dataset
...
from statannotations.Annotator import Annotator

fig, axes = plt.subplots(1, 4, figsize=(CONTENT_WIDTH_IN*.5, 3 * CM_TO_IN),
                        sharey=True)

# Get colors from custom_palettefor each channel
custom_colors = plotting.custom_palette

# Define statistical comparisons (all pairwise)
stat_pairs = [("off", "on"), ("off", "opp"), ("on", "opp")]

for i, color_channel in enumerate(hue_order):
    color_data = data[data['color_display'] == color_channel]
    channel_color = custom_colors[i]  # Use the specific color for this channel

    sns.boxplot(data=color_data, x='category', y=value_col,
               order=order, color=channel_color, ax=axes[i])

    sns.stripplot(data=color_data, x='category', y=value_col,
                  color='k', ax=axes[i], jitter=True, dodge=True,
                  alpha=.2, marker='.')

    # Add statistical annotations
    annotator = Annotator(axes[i], stat_pairs, data=color_data,
                        x='category', y=value_col, order=order)
    annotator.configure(test='Mann-Whitney', text_format='star', loc='outside',
                    hide_non_significant=True)
    annotator.apply_and_annotate()

    axes[i].set_xlabel('')
    if i == 0:
        axes[i].set_ylabel(f'{metric_name}')
    else:
        axes[i].set_ylabel('')

    # Capitalize x-axis labels (off -> Off, on -> On, opp -> Opp)
    labels = [label.get_text().capitalize() for label in axes[i].get_xticklabels()]
    axes[i].set_xticklabels(labels)
sns.despine()
plt.tight_layout()