General comments:

This is a very nice paper that provides an elegant solution to habitat mapping needs for species of conservation concern. The manuscript is extremely well written and the approach is timely, though the results are grim. I think this manuscript will be well received by readers and managers. The authors provide a clear blueprint for automated habitat mapping building off past work, and the utility for other systems I think is clear. Excellent work. I have a handful of methodological questions below. My lone major criticism surrounds the definition of “fisher potential reproductive habitat” which is used throughout the manuscript but only somewhat defined for the first time on line 447. It remains unclear to me what is actually being mapped here and how the documented changes will actually affect fishers. It is revealed late in the paper that this analysis was also done with 2500 den site locations (7x more than the presented detections), but the results are said to be comparable and therefore habitat maps here may represent reproductive habitat. The authors acknowledge this limitation in a closing paragraph, but this strikes me as insufficient. I have detailed my concerns below, and I think the term and interpretation of trends need further explanation and interpretation.

Specific comments:

L20 – no and, just “highlighting”

L21-26 – With multiple “eg’s” in one sentence this gets hard to read and digest. Suggest breaking into 2 sentences.

L39 – “in” local ecosystems

L116 – Presumably non-detections could have also been used (e.g., Davis et al. 2007) and would have improved your model. Why use pseudoabsences when considerable data on non-detections exist?

L116-120 – Why the focus on “reproductive” habitat, and does focusing on detections of female fishers achieve this goal? I think this is interesting since you have the genetic data to support sex ID, but if this is to be replicated by other monitoring programs it’s unlikely that they will have the same level of detail. Given that male and female fisher habitat selection is largely equivalent (Zielinski et al. 2004, Sauder and Rachlow 2015, Kordosky et al. 2021), would you expect to see (or did you see) a different pattern if using all detections vs a single sex?

L119- Probably need to define or cite approach for sex ID since that is critical to the data being used.

L136 – What is the ultimate mapping resolution? 30m pixels? Also, is data extracted only from the pixel where fishers are detected, or buffered to a broader scale representing 3rd or 4th order habitat selection? Habitat and climate data I presume matches the timing/year of detection? Think these details need clarifying.

L166 – Don’t see it in this section, but did submodels from each region only use samples from that region? If so, how did those 330 samples break down across areas and could this have contributed to the relative lack of transferability between regions. Fig 1 looks like Kern is pretty underrepresented relative to the other 2 regions.

L193 – I think “fisher potential reproductive habitat” needs to be formally defined somewhere as it is routinely used. The authors are careful to define this as “potential” habitat given the presence-only modeling framework (which should also be made clearer), but “reproductive habitat” seems far too ambiguous. Is any habitat where female fishers are detected “potential reproductive habitat” or is it just habitat?

L231-245 – This is pretty hard to follow, particularly the first half of the paragraph.

L302-312 – Is the placement of monitoring stations in FIA sites problematic in that it biases detections to forested habitats? Is fisher monitoring maintained post-fire? If all detections are in forests, then of course high intensity stand replacing fire will lead to extreme reductions in that habitat, but presumably fishers also benefit from fire to some degree through the creation of early seral characteristics that favor small mammal and insect prey (Zielinski et al. 1999, Smith et al. 2022, Kuntze et al. 2023, Pilgrim et al. 2023). They also use more open and early seral habitats, admittedly to a much smaller degree. I expected to see some discussion of this later on but did not.

L344 – How do 153 environmental covariates specifically model fisher potential reproductive habitat as opposed to mature forest vegetation? What makes this uniquely different, other than focusing on females alone (who may not be using said locations for reproduction either)? It’s still unclear to me what reproductive habitat even is (denning, resting, kit rearing?) much less how this is distinctive from other habitat types, but the authors clearly see a difference so that should be explained.

L447 – Here it is! Needs to go much, much earlier in this manuscript.

L454 – Why not present the data in its entirety? Think there has to be more justification than this, as it’s effectively asking readers to take your word for it. A quantitative comparison is justified given the emphasis on reproductive habitat throughout. It seems that 2500 known denning locations, even if spatially biased, is a much better data source to test changes in reproductive habitat than 300 female detections. Not sure I understand the logic. Also think this gets back to the mechanism of habitat loss – is the issue loss of denning trees, prey, all of the above? I understand it’s not being tested directly, but a more detailed interpretation of potential drivers seems prudent to inform local management. Some discussion of what these changes in habitat actually represent is needed.

Fig 4 – A and B definitions needed in fig caption

References
Davis, F., C. Seo, and W. Zielinski. 2007. Regional variation in home-range-scale habitat models for fisher (Martes pennanti) in California. Ecological Applications.

Kordosky, J. R., E. M. Gese, C. M. Thompson, P. A. Terletzky, K. L. Purcell, and J. D. Schneiderman. 2021. Landscape use by fishers (Pekania pennanti): Core areas differ in habitat than the entire home range. Canadian Journal of Zoology 99:289–297.

Kuntze, C. C., J. N. Pauli, C. J. Zulla, J. J. Keane, K. N. Roberts, B. P. Dotters, S. C. Sawyer, and M. Z. Peery. 2023. Landscape heterogeneity provides co-benefits to predator and prey. Ecological Applications 33:1–18.

Pilgrim, K. L., R. E. Green, K. L. Purcell, T. M. Wilcox, E. L. McGregor, L. E. Gleason, S. K. Wasser, and M. K. Schwartz. 2023. Shifts in fisher (Pekania pennanti) diet in response to climate-induced tree mortality in California assessed with DNA metabarcoding. Journal for Nature Conservation 73:126408.

Sauder, J. D., and J. L. Rachlow. 2015. Forest heterogeneity influences habitat selection by fishers (Pekania pennanti) within home ranges. Forest Ecology and Management 347:49–56.

Smith, G. B., J. M. Tucker, and J. N. Pauli. 2022. Habitat and drought influence the diet of an unexpected mycophagist: fishers in the Sierra Nevada, California. Journal of Mammalogy 103:328–338.

Zielinski, W. J., N. P. Duncan, E. C. Farmer, R. L. Truex, A. P. Clevenger, and R. H. Barrett. 1999. Diet of fishers (Martes pennanti) at the southernmost extent of their range. Journal of Mammalogy 80:961–971.

Zielinski, W. J., R. L. Truex, G. A. Schmidt, F. V. Schlexer, K. N. Schmidt, and R. H. Barrett. 2004. Home range characteristics of fishers in California. Journal of Mammalogy 85:649–657.

I found this to be a pretty hostile response to what was overall a very friendly review. As a result, I did not find the responses to the comments raised by either reviewers or the AE to be compelling, and it appears the authors made no real attempt to justify their position or analyses related to a number of valid questions. Did the authors make any real changes, or do they simply think they are beyond reproach here? Moreover, it is difficult to assess what if any changes were made because they did not document line numbers or quoting altered text in their response. I still believe this to be a valuable contribution, but many of the original criticisms remain insufficiently addressed.

L116, Author Response - “Many, many non-presence locations are needed to produce a reliable model, and no dataset exists that is sufficiently large and spatially balanced to meet this need, particularly for our effort of modeling potential reproductive habitat. Using pseudo-absences was the only realistic approach for our purposes.”

This may be the case, but the authors make absolutely no attempt here to justify their decision and are expecting everyone to trust them based on past papers. I don’t think this is a valid answer, and I would have expected to see some form of data or analysis justifying the decision.

L116-120, Author Response – “We have added further justification in the text to clarify this point…. We believe that is the case because including males included multiple habitat types that fishers may select for that are unrelated to reproduction, and so the model had difficulty identifying what was “habitat” and what was not. For these reasons, using female locations was the best way to achieve our goal of modeling potential reproductive habitat.”

Please point to these additions by citing line numbers or quoting altered text. Second, I still think the authors are assuming that any female detection is representative of “reproductive habitat” and conflating probability of use with the factors actually dictating reproduction. This response does nothing to assuage those concerns.

L166, Author Response – “Yes, submodels were limited to each region, this was stated in the original text: “We developed three species distribution models within the GEE environment, trained on data from each subregion.” We have added in the revised text the number of locations per region. The uneven sampling intensity in each subregion certainly could have contributed. Regardless, even with small samples in some regions (Kern), model performance metrics were superior across all regions. Moreover, we have reason to believe, based on previous studies and research in this large bioregion, that fisher habitat selection might vary spatially, and that such habitat-environment relationships may be particularly unique in the Kern Plateau.”

When a reviewer asks a question, the appropriate response is to acknowledge that there may be some confusion and that you, the authors, have the responsibility to clarify this so that your study can be as clear and transparent as possible. Once again, the authors have used their response to make the bare minimum adjustments (not cited with line numbers, again) and have effectively told us to trust them because they know better than anyone else.

L123 – “we can assume that female fishers do not stray far from suitable denning and reproductive habitat” and Author Response “We have added further clarification in the text, and see the above comment that explains reproductive habitat and why we used female locations”.

Can you assume this? Females disperse up to 18km (Matthews et al. 2013) and have home ranges that vary from 5 to 50 km2 (Zielinski et al. 2004, Sauder and Rachlow 2015, Kordosky et al. 2021). Sure, core home ranges are smaller and probably representative of “reproductive habitat” as defined, but what evidence is there that the non-invasive detections from baited cameras or hair snares are at all representative of core home ranges? Given the longevity of this work, I suspect there is data that could be used to test the questions raised in the initial review, but once again there is no supporting data presented here to justify the positions being taken by the authors.

L302 Author response – “Certainly fishers use non-forested habitat. But they would not use these types of habitats to support reproductive behavior, which is what we are modeling...”

Why not? The authors claim in their response that “reproductive habitat includes not only den sites but also the surrounding habitat, such as travel corridors and foraging habitat.” If the argument was that its only den sites then I get why you would ignore more open habitats, but they specifically say it is also travel and foraging and numerous papers show that fishers use early seral habitat for foraging. So which is it, can’t have it both ways.

L454, Author Response – “However, modeling habitat that is suitable for den sites will only capture that—denning—and will not capture the full spectrum of forest structural features and environmental conditions that can support successful dens (such as foraging habitat) which is what is of primary concern for managers.”

How does modeling habitat based on non-invasive detections address foraging habitat at all? You have shown no correlation between the two at all. This argument has now been used multiple times and it remains unconvincing. See comments above. I don’t think the question about what is actually being modeled was ever answered, and although “reproductive habitat” is now defined, after reading these responses it apparently captures denning, resting, foraging, and movement. So all habitat for female fishers? To me this term means effectively nothing other than predicted habitat for females (which does not mean it is important at all for reproduction) and I certainly don’t believe the approach captures any reasonable definition of foraging habitat based off 300 hair snare detections.

References

Kordosky, J. R., E. M. Gese, C. M. Thompson, P. A. Terletzky, K. L. Purcell, and J. D. Schneiderman. 2021. Landscape use by fishers (Pekania pennanti): Core areas differ in habitat than the entire home range. Canadian Journal of Zoology 99:289–297.

Matthews, S. M., J. M. Higley, K. M. Rennie, R. E. Green, C. a. Goddard, G. M. Wengert, M. W. Gabriel, and T. K. Fuller. 2013. Reproduction, recruitment, and dispersal of fishers ( Martes pennanti ) in a managed Douglas-fir forest in California. Journal of Mammalogy 94:100–108.

Sauder, J. D., and J. L. Rachlow. 2015. Forest heterogeneity influences habitat selection by fishers (Pekania pennanti) within home ranges. Forest Ecology and Management 347:49–56.

Zielinski, W. J., R. L. Truex, G. A. Schmidt, F. V. Schlexer, K. N. Schmidt, and R. H. Barrett. 2004. Home range characteristics of fishers in California. Journal of Mammalogy 85:649–657.

The authors have addressed my primary concern that “potential reproductive habitat” was not clearly defined nor clearly tied to the modeling process. The authors do a much better job of articulating what is being modeled and what it represents ecologically in L121 onwards – I think this is much clearer and well-reasoned than previous iterations, and it now provides a more logical and ecologically informed justification for focusing on female fisher habitat writ large.