While you should be highly commended for all the captive and field effort and hard work trying to learn the most out of these release interventions for such a critical bear species, I think that so many technical issues impede to your manuscript to reach the standards of a high quality, academic wildlife journal. I fully realize that most of these issues do not really depend on you (e.g., high failure rates of GPS collars, high inaccessibility of your study area), but I am afraid this realization alone is not sufficient for your manuscript to gain publication status. There are a few general comments that highlight the major weaknesses of the manuscript:
- to assess the outcome of a release program at the individual level, 8-12 weeks of tracking (at the most) should be considered insufficient. It is not by accident that you originally planned to have those GPS collars to survive for 60-90 weeks.
- the home range concept should be applied only to individuals displaying residency and its estimate of size gains significance if supported by additional ecological information (e.g., composition of the home range n terns of habitat types). For released individuals that are not residential yet, the area they use (what you define home range) may not be the most practical nor informative indicator of space-use patterns. Also, the distance travelled between successive GPS locations may not be the most informative measure of space use following release, as this does not account for the total distance travelled from the release site, possibly discriminating between an exploratory/dispersal phase and a successive residential status.
- The analyses (i.e., choice of the variables, their treatment and statistical inference) should be guided by research or working hypotheses, in your specific case linked to the assessment of the outcome of the release. In other words, in order to assess the outcome (success, partial success, or failure) of this conservation intervention, you should formulate benchmarks of conditions or reference values upon which to compare your results. How (and why) would you judge the outcome of a release based on home range size (or distance traveled) alone?
- Measures of feeding and foraging ecology, nutritional conditions, or landscape used (with particular reference to anthropogenic conditions) by release bears should be an integral component of post-release monitoring, not to mention causes of death. If these data are impossible to get using GPS devices alone, other complementary means could be adopted (e.g., camera trapping).

In addition to the above, the following detailed comments are meant to highlight aspects that could be further addressed should the authors be willing to work on an improved version of their manuscript (the numbers refer to the corresponding line of the text).

Abstract
7: ‘rescued’ would be preferable over ‘wild caught’.
10: ‘behaviour’ here would require some further specification as to what type of behaviour (e.g., post-release movement’ or ‘post-release spatial patterns’).
14: the home range concept should be limited to the area that is effectively used routinely by an animal during its normal daily activities; that is, it should exclude areas visited only once or only temporally, as it may happen with released individuals that may undergo an exploration period prior to settlement. In addition, quantification of home rage size per se does not necessarily hold any relevant information if not related to other ecological parameters that affect the animal’s fitness. You may have (released) bears that use smaller or larger home ranges without gaining any knowledge of their fitness and survival probabilities.
15-16: apart from the small sample size (both in bears and tracking period), gender is just one of many factors that may affect home range size. What about age, ecological conditions within the home range, interaction with other species (including humans), etc.?
16: how did you measure (and define) home range stabilization? This is not clear from your manuscript.
17: what is a ‘significant amount’? What is the underlying research hypothesis behind this statement? Why would you expect these bears to spend time together?
19: how would you define the ‘ultimate success’?
20: not clear why the inaccessibility of the terrain would preclude assessing the ‘ultimate success’ of the release, please clarify.
20-21: not clear on what results of your study this conclusion is based. You just maintained that the ultimate success of these releases could not be assessed.
22-23: not clear from your manuscript which are the findings that provide support for this closing statement.

Introduction
49-57: assessing sun bear release interventions does not necessarily equate with assessing their spatio-temporal patterns after release. Being surely part of the picture, this information conveys an insufficient amount of data to assess the success of such interventions. What about their survival or cause of death? Their post-release reproductive success or other measures of fitness?

84-87: this is fundamental information for this study, yet is it highly unspecific. ‘Behaviour’ in this context is too vague and little informative. How did you define the ‘success of these reintroductions’? What are the specific insights you gained from this study and what the connection with the results you presented?

88: please note that the study site has not been defined (why you deem relevant that bears move within and outside the study site?).

89: what is the hypothesis underlying the second objective (i.e., ii)? Why do you believe that monitoring these interactions is interesting or useful?

90: what do you exactly mean by ‘outcome’? If you could be more specific that would be beneficial for the kind of insights you may gain from these release interventions. In addition, collaring success is relevant but not only for release interventions. Collar failure on bears is a daunting and unfortunately already known issue (and perhaps not something to be analyzed but rather reported?)

Methods
156: understanding the rationale behind the choice of the release areas would be beneficial. What were the environmental and anthropogenic factors which made these areas ideal release sites?

164: same date and simultaneously on the same site? That would be important to know, as well as the rationale behind releasing bears together (what expectations did you have in doing so)?

174: To issues here: (a) the home range concept has a relevant temporal dimension, markedly affecting its size among other things (you cannot compare a 1-week to a 3-month home range, and a 3-month home range is not representative of an annual home range); (b) the home range as a concept applies to areas usually and routinely used by a resident animal and does not apply equally well to migrating or dispersing individuals or during exploratory travels. I’m wondering how you considered and accounted for these important aspects in choosing the home range as an indicator of post-release behaviour.

179-180: I believe that 1-week home range estimates are too unstable and sample size-dependent to be of any relevant meaning. In addition, you should motivate the choice of the parameters you are estimating as a function of what you are trying to assess. This is not a descriptive study but, based on the stated aims in the introduction, you are trying to assess the outcome of the release intervention.

183-184: I’m not sure at all that inter-fix distance is an ideal indicator of post-release spatial behaviour. But again, what are the specific aims of your analysis and what the properties of an ideal space-use indicator? What are you expecting these indicators (inter-fix distance and summation across weeks) would reflect that is useful or informative to assess the outcome of the releases?

186: why you believe that temporal patterns are important to assess the outcome of the releases? What hypotheses you have in mind reflecting the connection between temporal budgets of released bears and the outcome of the release interventions?

206-208 (and Tables S1- S2): not clear in Table S1 if bear ID (see Table S2) were included as predictors as well in the three tested models. Please specify.

216: this is unexpected as in the Methods section you said that collars were programmed to acquire one location every 4 hours. Please clarify.

218-223: not clear what is the underlying hypothesis here. In addition, it is not clear how inter-bear distances are analyzed to make inferences on expected behaviour (e.g., any benchmark? Any test?).

Results:
232: but you deployed 12 collars, right? The remaining six collars then were not configured to include a mortality signal or what? Please clarify.

232-242: this is critical information to understand the performance of the Vectronic collars and the fate of the bears. However, the text reads somewhat confusing and is not immediate to understand which bear/collar is which. I strongly suggest moving this information into a Table with individual bears/collars in rows and their state/fate in columns.

236-237: this is an example that providing information regarding the habitat composition of individuals’ home ranges, rather than their simple size, is informative of the behaviour of the released individuals.

238: reasons why? I guess that if you adopted mortality signal you were interested in understanding causes of mortality. A 2-week delay in field surveying the mortality site greatly reduces the chance to detect mortality causes.

245-249: this is not truly relevant for the scope of this study. The bears’ immediate reactions upon release provide no indication whatsoever about their future fitness in the wild. Why do you think this information is relevant to be include in te Results? Please advance any relevant hypotheses in the Introduction.

250: this is hard to conclude, as most of your bears have been tracked for less than two months!

252-253: distance travelled each week cannot be, per se, indication of stabilization. How did you assess stabilization (no hints in the Methods section)? In addition, please note that the home range concept does not apply prior to spatial stabilization, and other measures should be adopted (see comment above).

Figure 2: In addition to the relevant individual variation, the lack of reference to the distance travelled from the release site makes this graph of very little informative value concerning the spatial behaviour of the released bears.

264-267: up to 13 weeks of tracking is highly insufficient (i.e., inadequate) to estimate the home range of individual bears! These estimates do not even cover a seasonal home range, excluding the fact that the released bears perhaps spent some days after their release to move away or explore unknown terrains (i.e., not part of their home range).

268: how large is the reserve compared to an average annual home range?

270: how large is the reserve compared to an average annual home range?

Figure 3: weekly and ‘total’ home ranges are very poor indicators of spatial patterns of post-released bears (see comments above). Most importantly, home range size per se does not convey information about where the home range is located (i.e., stabilization) and its configuration.

313-315: the step interval (i.e., the time between consecutive GPS locations) should be constant across time and bears to be comparable. In addition, are you sure this measure provides useful information on spatial patterns by released bears? You should at least explain why and how.

316-317: there is an error here as it is not possible to have an average value which is smaller that its minimum.

343-378. it is difficult to fully interpret these results in the absence of any conceptual model that would associate the higher of lower extent of interaction to any (biological, social, ecological) feature of the released bears. What kind of interaction would you have expected and why? And how this information is relevant to assess the outcome of the release intervention?

Discussion:
387-392: this is a good reason why to exclude these periods off the home range estimate (or to choose a different indication of spatial behaviour).

391: you did not provide evidence of ‘stabilization’ nor you described methods to measure it.

396: individual past experiences (e.g., length of the rehabilitation time) could be accounted for in your GLMs exploring home range effects.

403-410: I’m not sure about actual accessibility in your study area, but gaining information about the bears’ feeding behaviour could have been acquired through field surveying clusters of GPS locations (possibly using a more frequent acquisition rate), providing very important information about the ecology and behaviour of these released bears.

411: should be ‘phenological’

477: how did you measure or assess ‘success’?

517-518: given the high variability of your results, the many uncovered factors affecting behaviour (and survival) of released bears, I cannot see how your findings provide ‘key insights’. More than your quantified results, I guess it is your own personal and pluriannual experience (which is indeed of high value) in recovering sun bears that provides ‘key insights’. Of course, by no means personal experience is of lower quality and importance conservation-wise, but the fact is that this paper simply does not provide sufficient hard data or evidence to meet that specific scope.

519: I see no reference to this in your results.

520: what do you exactly mean by ‘understanding the in situ wild bear population’? Would this be this feasible and under what circumstances?

Can you suggest ways to enhance our knowledge on mortality and/or reproductive success of released bears? Those comments could be of extreme value for those conservationists working through releasing rehabilitated individuals.