Overall, this is a solid and well-structured manuscript with strong methodological grounding. The integration of dynamic occupancy modelling with practical monitoring outputs is timely and relevant, particularly in the context of large carnivore range expansion. However, I have identified some points you might consider addressing or clarifying before approval.

Main concerns

Lines 152-154: Given the nature of the monitoring program—particularly its targeted focus on areas with known or suspected wolf presence—it seems likely that the dataset is affected by presence-biased sampling. This would result in an underrepresentation of truly unoccupied sites, especially in regions with low or no survey effort. Combined with the potential circularity in using PPAs for threshold calibration (see also the next comment to lines 212-213), this may lead to an overestimation of model performance, particularly for predicting persistence. May this imbalance also help explain why the model appears less reliable in predicting absences or local extinctions? I suggest clarifying how sampling bias may influence the threshold calibration.

Lines 212–213: The calibration of the occupancy threshold using PPAs appears to rely on the same dataset used to fit the occupancy model. This raises concerns about circularity or overfitting, since both the modelled probabilities and the PPA-based metrics are derived from the same detection data. I suggest clarifying the extent of overlap and, if possible, recommending a data partitioning strategy to ensure a less biased threshold calibration.

Lines 227–231: The use of a custom composite metric (C) to calibrate binary occupancy maps provides a flexible way to balance conservation goals, but it lacks the interpretability and comparability of standard classification metrics. Given the constraints of the dataset, the approach is understandable, but I recommend either justifying the choice more explicitly or considering additional, complementary performance metrics such as Kappa, TSS... Numerous metrics exist to quantify classification accuracy, with some emphasising true positives or true negatives, and others designed to balance both aspects.

Lines 372–375: This passage reflects the well-documented ecological principle that occupancy and abundance are nonlinearly and scale-dependently related. The observed divergence, where abundance increases faster than occupied area after 2015, suggests that occupancy becomes less predictive of abundance as population density increases. It would be helpful to emphasize whether this divergence indicates a saturation point and whether this limits the future utility of occupancy modelling as a proxy for abundance under continued population growth. This could be especially relevant for the application of the approach in other geographical areas.

Lines 549–554: While the discussion of a dual-frame approach is interesting, this idea is not entirely new. I suggest citing relevant prior work, such as Gervasi et al. (2025) and the sampling design therein.
Gervasi, V., Aragno, P., Salvatori, V., Caniglia, R., Angelis, D. D., Fabbri, E., La Morgia, V., Marucco, F., Velli, E., & Genovesi, P. (2024). Estimating distribution and abundance of wide-ranging species with integrated spatial models: Opportunities revealed by the first wolf assessment in south-central Italy. Ecology and Evolution, 14(5), e11285. https://doi.org/10.1002/ece3.11285

Minor Comments

• Lines 136–137: The term _monitoring-focus scenarios_ is not clearly defined at this point. It would help to include concrete examples (e.g., prioritising predictions of both wolf presence and absence equally vs. mainly predicting wolf presence) to clarify the meaning and practical relevance of these scenarios.
• Line 181: The phrase _“more criteria”_ is vague. Please consider specifying.
• Line 189: The phrase _“the calibrated occupancy model”_ might be misleading, as the calibration in this study seems to refer primarily to the transformation of continuous occupancy probabilities into binary maps, rather than to the model fitting itself. Clarifying this distinction would help avoid confusion.
• Line 199: Please specify the resolution of the reference grid used.
• Line 205: It would be worth noting that the reliability of presence signs is assessed according to established criteria for large carnivores (e.g., C1, C2 classifications) - just a brief mention of this would reinforce the credibility of the dataset.
• Line 278: Minor typo — add a closing parenthesis after _“model”_.
• Line 397: Typo — _“change”_ should be _“changed.”_

Valentina La Morgia

The authors have made several useful revisions, and the manuscript is generally improved. E.g., the occupancy-abundance divergence is now discussed in more detail. However, a few of my main concerns remain only partially addressed:

1. The authors emphasise that their model is based on presence-only data, which indeed avoids some of the pitfalls of class imbalance in a presence–absence framework, and effort is also taken into account. However, low-density/low-effort regions may remain underrepresented. I still wonder if this effort bias may inflate apparent model performance when predictions are calibrated against PPAs (which are themselves presence-rich products) and may help explain the model’s limited ability to predict absences or local extinctions. While the discussion on rapid recolonisation in dense areas (lines 548–561) helps contextualise why missed local extinctions may have limited ecological impact, it does not fully address my concern. I would be grateful if the authors could explicitly address this and clarify the distinction: the ecological buffering effect of recolonisation is important, but it does not remove the methodological risk of inflating apparent performance.
2. The authors clarify that PPAs are naïve products while occupancy maps account for imperfect detection, arguing they are “not redundant.” While this is a useful clarification, it does not directly address the risk of circularity, given that both outputs ultimately derive from the same dataset.

In summary, while the manuscript is stronger after revision, I believe the above points still require clearer acknowledgement or discussion.