The study explores a breeding selection decisions of long distance natal dispersal events in a songbird, a topic of high ecological relevance, but one which has received no serious attention due to the extreme challenges associated with measuring these dispersal events. The authors use a combination of novel approaches to tackle this question, and overall this study makes a massive contribution to the methods possible in this field, and also contribute significant understanding that stems from them testing alternative theoretical hypotheses.

Overall the manuscript reads really well, the justification is solid and the methods employed are strong. Having more site replicates would have obviously strengthened the data, but I understand the reality of such data. This does however limit the conclusions because the use of the single predictor models where other effects cannot be accounted for. I have two more 'major' comments that I will mention first, and then will detail some minor comments thereafter.

Methods: Long distance natal dispersal was defined as events classified >100km in this study, and therefore the response variables are analyzing 0s and 1s based on this threshold. Have the authors explored other threshold values, which would increase or decrease the number of long-distance dispersal individuals in the dataset under the different priors? If isotope values are limited to a 100 km resolution, it would be interesting to understand how the effects will explain this trait if the threshold were 150 km or 200 km for example. I realize the most restrictive prior might not produce enough 1s in these cases, but less restrictive priors could be run. Does this alter the results observed?

Results: I defer to other reviewers about the details of the statistical methods used, however, I do notice the 95% CIs overlap zero in almost every case. This does not provide much confidence in the impact of effects being tested, despite the estimates being consistently in the same direction for different priors. I don’t think the authors address this aspect clearly in the manuscript, and perhaps this could be done. If I’ve misunderstood this aspect, I do apologize.

Minor comments:
Line 36: “…is likely more frequent.” The qualifier seems necessary given the uncertainty in the data and the extrapolation to other songbirds. Indeed the repeated mention of ‘songbirds’ more broadly is brave, and I think the authors should justify this extension in more detail.
Line 75: change ‘than’
Lines 91-92: there is some evidence that settlement decisions could also be impacted by the quality of individuals in patches, both in terms of a reliable source of information, but also in terms of increased competition for mates or food. This could be detailed as this may provide some needed understanding to considering on density (e.g. Laland 2004 Learn .Behav; Szymkowiak et al 2016 Oikos).
Lines 143 – 147: How was R calculated with the self-recruitment? Also using the isotope data? This might be detailed in Rushing et al. 2017a, but there is need for this to be introduced in brief please.
Line 157: ‘of’… rephrase.
Line 158: what is defined as forest in this classification? Without knowing details about wood thrush requirements its difficult to evaluate if this broad measure of ‘forest’ of presumably different age and make up is relevant. Any brief information that could help the reader evaluate this would be useful.
Lines 162 – 163: missing word?
Line 174; ‘grid cell’, what was the resolution of this?
Line 178 – 179: Where does the rationale of this designation come from? How large is the north-south breeding range in this species? Again, any brief information that could inform and allow evaluation would be useful.
Line 261: a brief explanation about the inclusion of the quadratic term may help readers here.
Line 265: I’m not sure exactly how the significance test were performed, but there are always concerns about using one-tailed tests. Can the authors clearly justify the use of 1-tailed tests? Surely predictors could theoretically impact probability of long distance dispersal in both directions, and the use of 1-tailed tests are only ‘allowed’ in very clear cases.
Lines 320 – 323: I’ve never understood how the timing of this would work for first time breeders? For failed breeding adults yes, but natal dispersers in year t would be fledging and trying to survive their first months prior to migration, and only the very earliest birds would be able to prospect performance cues that could be used in year t + 1. Given the studies focus on first time breeders, I don’t think this mechanism is feasible.

Table 1: define the unit of density in the legend.

Figure 2: how do the densities expressed on the x-axis in this figure relate to the densities provided in Table 1? It seems difficult to match sites as the values are expressed differently in the figure.