Changing bird communities of an agricultural landscape: declines in arboreal foragers, increases in large species

Birds are declining in agricultural landscapes around the world. The causes of these declines can be better understood by analysing change in groups of species that share life-history traits. We investigated how land-use change has affected birds of the Tasmanian Midlands, one of Australia's oldest agricultural landscapes and a focus of habitat restoration. We surveyed birds at 72 sites, some of which were previously surveyed in 1996–1998, and tested relationships of current patterns of abundance and community composition to landscape and patch-level environmental characteristics. Fourth-corner modelling showed strong negative responses of aerial foragers and exotics to increasing woodland cover; arboreal foragers were positively associated with projective foliage cover; and small-bodied species were reduced by the presence of a hyperaggressive species of native honeyeater, the noisy miner (Manorina melanocephala). Analysis of change suggests increases in large-bodied granivorous or carnivorous birds and declines in some arboreal foragers and nectarivores. Changes in species richness were best explained by changes in noisy miner abundance and levels of surrounding woodland cover. We encourage restoration practitioners to trial novel planting configurations that may confer resistance to invasion by noisy miners, and a continued long-term monitoring effort to reveal the effects of future land-use change on Tasmanian birds.

1. Study design. Using effectively two data points (occurrence patterns of birds in time period 1 and 2, twenty years apart) makes it very hard to identify trends, especially for an assemblage of animals characterized for their high vagility and often broad habitat tolerances. Rather than using this temporal dimension to frame their analyses, I would suggest running the same sets of analyses for the two datasets and looking instead at how determinants of occurrence / richness have changed, and exploring how those differences might relate to the dramatic land use changes that have occurred in the intervening period. 2. While I appreciate the method used to estimate bird occurrence was effectively decided by MacDonald and Kirkpatrick for the 1996/97 survey period, many of the concerns I have with the data relate to how it has been handled, rather than initially collected. Variable effort was used which is not necessarily problematic. Rather, by adjusting their data based on proportion of the site surveyed, a systematic error forcing a specie-area relationship might be the real problem. This can be avoided by using estimated richness rather than observed richness, following the methods of Colwell in his freely available EstimateS software. They can then look at how much the predicted richness diverges from observed richness and, if there are consistent effects of sampling effort, they can be quantified and more precisely dealt with. As for density, I have even more concern with the response variable used, and would recommend the authors follow the work of Mac Nally and others and uses incidence or reporting rate instead. That is, the proportion of surveys conducted in that site in which the species was detected as present. It is coarse, but far less prone to the many confounding effects of variable detectability, (between species, sites and seasons). 3. The analyses used were complex and, although well explained, the authors side-stepped what I took away as one of their main findings-the avifauna changed very little between sampling periods, and most of the factors measured had little explanatory power, either at the individual species scale or for variously-defined groups. There was also a very large number of potential relationships to explore--figure 4 alone is a distillation of approximately 600 univariate relationships. The determinants of species richness very murky, with more than half the variation unexplained. Yes, this is a negative result, but if that result remains using predicted richness where comparability across sites is assured, then the authors should stand by it, and discuss this lack of signal more explicitly. The fact that these site-based factors had so little explanatory power suggests that landscape scale factors might be more important, that interspecific differences and the masking effects of simultaneous positive and negative relationships might conceal any assemblage-wide trajectories. 4. Related to point 3, I would suggest the authors be a little more circumspect about their inferences. In figure 5, temporal changes for 33 species are depicted of which more than half have confidence intervals that overlap with zero. To me, this suggests that either most species aren't changing or the approach used to monitor them across this period is unable to discern any change. Likewise, in figure 6: just four of the 21 trait-defined groups exhibited significant changes in density: waterbirds, raptors, large birds and plant eaters. Rather than any trait-based influence, these four groups likely reflect the clear increases noted in swamp harriers and shelducks.
I would recommend the authors consider subdividing this large manuscript into three more narrowly-defined contributions. The noisy miner story might well be worth carving off and exploring on its own. [Oh, and please don't use the term "reverse keystone". Keystones are by definition rare in their communities, in terms of numbers of individuals / proportion of biomass--the term "despotic" is far more appropriate]. The determinants of diversity in two different time periods would also stand alone, leaving perhaps a third manuscript considering what you've found overall, what appears to be important and not important in structuring these bird assemblages in a highly modified system, how you might expect birds to respond to planned restoration, and some guiding predictions to revisit in another twenty years. This is all up to you but by lumping all of this into one paper, not only does it become unwieldy, but you simply don't have the space to consider your findings. There's plenty of interesting comments made in passing about particular sites, land-use practices, positive and negative responses to shifting land use. Reworking the paper into more clearly defined narratives will give you the structure and space to explore these in more depth, and make better use of all the hard work that went into compiling all of this information.

Review form: Reviewer 2
Is the manuscript scientifically sound in its present form? Yes

Are the interpretations and conclusions justified by the results? Yes
Is the language acceptable? Yes

Do you have any ethical concerns with this paper? No
Have you any concerns about statistical analyses in this paper? No

Recommendation?
Accept with minor revision (please list in comments)

Comments to the Author(s)
Review: Long -Long-term change in bird communities of an agricultural landscape: declines in arboreal foragers, increases in large species. Bain et al use bird surveys (past and present) to compare species and their traits with habitat characteristics in the Tasmanian midlands. Over all I think the paper is well written, the methods appropriate and well executed and the results interesting and well interpreted. I particularly like the use of the fourth corner approach (and the resulting figures) to addressing the question of how to link species occurrence data with complex habitat variation. My comments on are relatively minor, but I would suggest that the authors look out for over-long sentences in their revision, and attempt to clarify some parts of the methods as suggested below. The manuscript could be trimmed in some places to save words. Methods -generally fine, but more detail is needed about the method used in the original surveys -given the time that has passed since the first surveys, it is important to understand the possible factors that could confound the results between the past and now. E.g. was observer error quantified and if not, then this assumption should be spelled out more (the first mention I see is at L428 which is a bit late). I see GB did all the field work in the contemporary surveys-who did them in the past? Is it possible to quantify detection error between the surveyors? L157: presumably the orientation and location of transects was different due to change in vegetation cover (or not?) between survey periods? If yes, say so. L158: if I understand correctly, small patches received 1 * 100m survey transect and large ones received 4 * 200m transects back-to-back over 800m, plus an addition 20min/2ha? was the orientation random in large woodlands? Why you used 2 survey approaches (transects and then 2ha searches) -did you pool the data for analysis or were they used for different parts of the analysis? It's not clear to me how these different methods are used in analysis L198: "as a proxy for land use intensity." L204: specify here where you sourced miner density estimates -is this derived from your own models? L265: I understand this limitation of mvabund but think it would be helpful for readers if you provided some explanation of what this might mean for interpreting your potential results, especially since the large forest patches had quadruple the effort at transects twice as long as those in small patches L270: Clarify species richness -is that just the number of species per survey? Or number per site over the survey periods? Fig 4. It would help to have the environmental variable names on both the top and bottom for readability, but not critical L413-17: long and slightly clunky sentence. Reword. L440 and L671: remove 'productive' -plantations are referred to in other ways too elsewhere in the paper -be consistent L465: "Granivores and raptors appeared…" L468: re 'common in urban environments' either needs a citation or delete (relevance to the section seems tangential anyway) L611: flock sizes are results

6
We'd like to give you the choice of either revising the current manuscript to address as far as possible the concerns raised during this round of review and so get the paper across the line largely in its current format (we'll ask the existing reviewers to comment on the changes made).
Or you may take the opportunity to divide your work as suggested by the reviewer and use the resubmission as the basis for one of the two papers, and then preparing a second manuscript for submission and consideration alongside the resubmission (again, we'll ask the current reviewers for their thoughts).
Whichever approach you take, we'd be grateful if you could contact the editorial office to let them know, so we can take the appropriate steps post-revision.
Good luck! Reviewers' Comments to Author: This manuscript, by Bain and collaborators, summarises a retrospective series of comparisons of bird occurrence in the midlands of Tasmania, as part of ongoing restoration efforts aimed to improve wildlife habitats in the region. The basis of the work is two sets of bird surveys collected at thirty three sites in 1996/1997 and 2015/2017, with another 39 sites added for the recent round of surveys. This manuscript summarized a tremendous amount of work, both in the field conducting bird surveys and vegetation data, compiling site attributes in a GIS, and analysing determinants of bird occurrence.
I have four sets of substantive comments, relating to overall study design, bird surveying methods, analysis and inference as well as various minor quibbles.
1. Study design. Using effectively two data points (occurrence patterns of birds in time period 1 and 2, twenty years apart) makes it very hard to identify trends, especially for an assemblage of animals characterized for their high vagility and often broad habitat tolerances. Rather than using this temporal dimension to frame their analyses, I would suggest running the same sets of analyses for the two datasets and looking instead at how determinants of occurrence / richness have changed, and exploring how those differences might relate to the dramatic land use changes that have occurred in the intervening period. 2. While I appreciate the method used to estimate bird occurrence was effectively decided by MacDonald and Kirkpatrick for the 1996/97 survey period, many of the concerns I have with the data relate to how it has been handled, rather than initially collected. Variable effort was used which is not necessarily problematic. Rather, by adjusting their data based on proportion of the site surveyed, a systematic error forcing a specie-area relationship might be the real problem. This can be avoided by using estimated richness rather than observed richness, following the methods of Colwell in his freely available EstimateS software. They can then look at how much the predicted richness diverges from observed richness and, if there are consistent effects of sampling effort, they can be quantified and more precisely dealt with. As for density, I have even more concern with the response variable used, and would recommend the authors follow the work of Mac Nally and others and uses incidence or reporting rate instead. That is, the proportion of surveys conducted in that site in which the species was detected as present. It is coarse, but far less prone to the many confounding effects of variable detectability, (between species, sites and seasons). 3. The analyses used were complex and, although well explained, the authors side-stepped what I took away as one of their main findings-the avifauna changed very little between sampling periods, and most of the factors measured had little explanatory power, either at the individual species scale or for variously-defined groups. There was also a very large number of potential relationships to explore--figure 4 alone is a distillation of approximately 600 univariate relationships. The determinants of species richness very murky, with more than half the variation unexplained. Yes, this is a negative result, but if that result remains using predicted richness where comparability across sites is assured, then the authors should stand by it, and discuss this lack of signal more explicitly. The fact that these site-based factors had so little explanatory power suggests that landscape scale factors might be more important, that interspecific differences and the masking effects of simultaneous positive and negative relationships might conceal any assemblage-wide trajectories. 4. Related to point 3, I would suggest the authors be a little more circumspect about their inferences. In figure 5, temporal changes for 33 species are depicted of which more than half have confidence intervals that overlap with zero. To me, this suggests that either most species aren't changing or the approach used to monitor them across this period is unable to discern any change. Likewise, in figure 6: just four of the 21 trait-defined groups exhibited significant changes in density: waterbirds, raptors, large birds and plant eaters. Rather than any trait-based influence, these four groups likely reflect the clear increases noted in swamp harriers and shelducks.
I would recommend the authors consider subdividing this large manuscript into three more narrowly-defined contributions. The noisy miner story might well be worth carving off and exploring on its own. [Oh, and please don't use the term "reverse keystone". Keystones are by definition rare in their communities, in terms of numbers of individuals / proportion of biomass--the term "despotic" is far more appropriate]. The determinants of diversity in two different time periods would also stand alone, leaving perhaps a third manuscript considering what you've found overall, what appears to be important and not important in structuring these bird assemblages in a highly modified system, how you might expect birds to respond to planned restoration, and some guiding predictions to revisit in another twenty years. This is all up to you but by lumping all of this into one paper, not only does it become unwieldy, but you simply don't have the space to consider your findings. There's plenty of interesting comments made in passing about particular sites, land-use practices, positive and negative responses to shifting land use. Reworking the paper into more clearly defined narratives will give you the structure and space to explore these in more depth, and make better use of all the hard work that went into compiling all of this information.

Reviewer: 2
Review: Long -Long-term change in bird communities of an agricultural landscape: declines in arboreal foragers, increases in large species.
Bain et al use bird surveys (past and present) to compare species and their traits with habitat characteristics in the Tasmanian midlands. Over all I think the paper is well written, the methods appropriate and well executed and the results interesting and well interpreted. I particularly like the use of the fourth corner approach (and the resulting figures) to addressing the question of how to link species occurrence data with complex habitat variation. My comments on are relatively minor, but I would suggest that the authors look out for over-long sentences in their revision, and attempt to clarify some parts of the methods as suggested below. The manuscript could be trimmed in some places to save words. Methods -generally fine, but more detail is needed about the method used in the original surveys -given the time that has passed since the first surveys, it is important to understand the possible factors that could confound the results between the past and now. E.g. was observer error quantified and if not, then this assumption should be spelled out more (the first mention I see is at L428 which is a bit late). I see GB did all the field work in the contemporary surveys-who did them in the past? Is it possible to quantify detection error between the surveyors? L157: presumably the orientation and location of transects was different due to change in vegetation cover (or not?) between survey periods? If yes, say so. L158: if I understand correctly, small patches received 1 * 100m survey transect and large ones received 4 * 200m transects back-to-back over 800m, plus an addition 20min/2ha? was the 8 orientation random in large woodlands? Why you used 2 survey approaches (transects and then 2ha searches) -did you pool the data for analysis or were they used for different parts of the analysis? It's not clear to me how these different methods are used in analysis L198: "as a proxy for land use intensity." L204: specify here where you sourced miner density estimates -is this derived from your own models? L265: I understand this limitation of mvabund but think it would be helpful for readers if you provided some explanation of what this might mean for interpreting your potential results, especially since the large forest patches had quadruple the effort at transects twice as long as those in small patches L270: Clarify species richness -is that just the number of species per survey? Or number per site over the survey periods? Fig 4. It would help to have the environmental variable names on both the top and bottom for readability, but not critical L413-17: long and slightly clunky sentence. Reword. L440 and L671: remove 'productive' -plantations are referred to in other ways too elsewhere in the paper -be consistent L465: "Granivores and raptors appeared…" L468: re 'common in urban environments' either needs a citation or delete (relevance to the section seems tangential anyway) L611: flock sizes are results Reviewer: 3 A very good paper. See attached file for some comments. On behalf of the Editor, I am pleased to inform you that your Manuscript RSOS-200076 entitled "Changing bird communities of an agricultural landscape: declines in arboreal foragers, increases in large species." has been accepted for publication in Royal Society Open Science subject to minor revision in accordance with the referee suggestions. Please find the referees' comments at the end of this email.
The reviewers and Subject Editor have recommended publication, but also suggest some minor revisions to your manuscript. Therefore, I invite you to respond to the comments and revise your manuscript.
• Ethics statement If your study uses humans or animals please include details of the ethical approval received, including the name of the committee that granted approval. For human studies please also detail whether informed consent was obtained. For field studies on animals please include details of all permissions, licences and/or approvals granted to carry out the fieldwork.
• Data accessibility It is a condition of publication that all supporting data are made available either as supplementary information or preferably in a suitable permanent repository. The data accessibility section should state where the article's supporting data can be accessed. This section should also include details, where possible of where to access other relevant research materials such as statistical tools, protocols, software etc can be accessed. If the data has been deposited in an external repository this section should list the database, accession number and link to the DOI for all data from the article that has been made publicly available. Data sets that have been deposited in an external repository and have a DOI should also be appropriately cited in the manuscript and included in the reference list.
If you wish to submit your supporting data or code to Dryad (http://datadryad.org/), or modify your current submission to dryad, please use the following link: http://datadryad.org/submit?journalID=RSOS&manu=RSOS-200076 • Competing interests Please declare any financial or non-financial competing interests, or state that you have no competing interests.
• Authors' contributions All submissions, other than those with a single author, must include an Authors' Contributions section which individually lists the specific contribution of each author. The list of Authors should meet all of the following criteria; 1) substantial contributions to conception and design, or acquisition of data, or analysis and interpretation of data; 2) drafting the article or revising it critically for important intellectual content; and 3) final approval of the version to be published.
All contributors who do not meet all of these criteria should be included in the acknowledgements.
We suggest the following format: AB carried out the molecular lab work, participated in data analysis, carried out sequence alignments, participated in the design of the study and drafted the manuscript; CD carried out the statistical analyses; EF collected field data; GH conceived of the study, designed the study, coordinated the study and helped draft the manuscript. All authors gave final approval for publication.
• Acknowledgements Please acknowledge anyone who contributed to the study but did not meet the authorship criteria.
• Funding statement Please list the source of funding for each author.
Please note that we cannot publish your manuscript without these end statements included. We have included a screenshot example of the end statements for reference. If you feel that a given heading is not relevant to your paper, please nevertheless include the heading and explicitly state that it is not relevant to your work.
Because the schedule for publication is very tight, it is a condition of publication that you submit the revised version of your manuscript before 12-Feb-2020. Please note that the revision deadline will expire at 00.00am on this date. If you do not think you will be able to meet this date please let me know immediately.
To revise your manuscript, log into https://mc.manuscriptcentral.com/rsos and enter your Author Centre, where you will find your manuscript title listed under "Manuscripts with Decisions". Under "Actions," click on "Create a Revision." You will be unable to make your revisions on the originally submitted version of the manuscript. Instead, revise your manuscript and upload a new version through your Author Centre.
When submitting your revised manuscript, you will be able to respond to the comments made by the referees and upload a file "Response to Referees" in "Section 6 -File Upload". You can use this to document any changes you make to the original manuscript. In order to expedite the processing of the revised manuscript, please be as specific as possible in your response to the referees.
When uploading your revised files please make sure that you have: 1) A text file of the manuscript (tex, txt, rtf, docx or doc), references, tables (including captions) and figure captions. Do not upload a PDF as your "Main Document". 2) A separate electronic file of each figure (EPS or print-quality PDF preferred (either format should be produced directly from original creation package), or original software format) 3) Included a 100 word media summary of your paper when requested at submission. Please ensure you have entered correct contact details (email, institution and telephone) in your user account 4) Included the raw data to support the claims made in your paper. You can either include your data as electronic supplementary material or upload to a repository and include the relevant doi within your manuscript 5) All supplementary materials accompanying an accepted article will be treated as in their final form. Note that the Royal Society will neither edit nor typeset supplementary material and it will be hosted as provided. Please ensure that the supplementary material includes the paper details where possible (authors, article title, journal name).
Supplementary files will be published alongside the paper on the journal website and posted on the online figshare repository (https://figshare.com). The heading and legend provided for each supplementary file during the submission process will be used to create the figshare page, so please ensure these are accurate and informative so that your files can be found in searches. Files on figshare will be made available approximately one week before the accompanying article so that the supplementary material can be attributed a unique DOI.
Once again, thank you for submitting your manuscript to Royal Society Open Science and I look forward to receiving your revision. If you have any questions at all, please do not hesitate to get in touch. It is a pleasure to accept your manuscript entitled "Changing bird communities of an agricultural landscape: declines in arboreal foragers, increases in large species." in its current form for publication in Royal Society Open Science. The comments of the reviewer(s) who reviewed your manuscript are included at the foot of this letter.
Please ensure that you send to the editorial office an editable version of your accepted manuscript, and individual files for each figure and table included in your manuscript. You can send these in a zip folder if more convenient. Failure to provide these files may delay the processing of your proof. You may disregard this request if you have already provided these files to the editorial office.
You can expect to receive a proof of your article in the near future. Please contact the editorial office (openscience_proofs@royalsociety.org) and the production office (openscience@royalsociety.org) to let us know if you are likely to be away from e-mail contact --if you are going to be away, please nominate a co-author (if available) to manage the proofing process, and ensure they are copied into your email to the journal. Due to rapid publication and an extremely tight schedule, if comments are not received, your paper may experience a delay in publication.
Please see the Royal Society Publishing guidance on how you may share your accepted author manuscript at https://royalsociety.org/journals/ethics-policies/media-embargo/.

By Glen Bain et al for Roy Soc Open Science
This is an excellent paper on a subject of global interest. It is well written, and the literature view is very thorough for recent studies. The paper can be published with very little change. However, the following points could add some value: 1. The term "response ratio" could imply a response to an identified process such as agricultural intensification, whereas in fact it just refers to a change over time between the two survey periods (with agricultural intensification as one of several possible causal factors). Would "percent change" be a less misleading term? 2. The trait-based approach is clearly useful, and quite similar to the guild approach used in various similar studies. The latter approach would enable conclusions to be made about changes in the abundance of whole guilds, e.g. nectarivores, canopy-foraging insectivores or small birds and large birds, rather than just saying that several species with those traits increased or decreased. It seems that calculations were made along these lines ( Fig. 6) and perhaps more could be made of them in the text? I found only one reference to Fig. 6, and that just related to large herbivores such as ducks (line 465). 3. The basic results (as shown in Figure 2) are similar to those from a study of fragmented forests in south-eastern Victoria (Loyn 1985(Loyn , 1987, which showed the role of noisy miners, and gave estimates of likely declines in numbers of species over time. The study was repeated 22 years later, and the further declines were found to be less than predicted at a set of sites where eucalypt plantations had been established nearby (MacHunter et al 2006). It may be worth comparing these results.

Some specific comments (re line numbers):
Title. Not sure that 20 years is long-term. Perhaps "Changes over 20 years…."?
27 & 28. "have" not "has" (subject is plural, "levels"), one of just two typos noticed. 207. Could also refer to the early work on this subject: Dow (1977) showed that Noisy Miners exclude other birds, and Loyn (1985) showed the importance of this process in fragmented forests in rural landscapes (in Victoria).
298. Interesting that Striated Fieldwrens were recorded at all (albeit offsite). In Victoria I have also found them in rural environments (in weedy young pine plantations and woody weeds in farmland) but quite rarely (and not for many years): they remain common in or near saltmarsh and sparsely treed heath, but otherwise rare in farmland away from the coast. [This is just a comment for 512. Similar effects were also found by Carla Catterall in peri-urban and rural environments in southeast Queensland (Catterall et al. 1997(Catterall et al. , 1998

Manuscript ID: RSOS-191405
Long-term change in bird communities of an agricultural landscape: declines in arboreal foragers, increases in large species.
We thank the three reviewers of this manuscript for their helpful comments. Please see our detailed responses to their suggestions below.
Please note that while revising this manuscript, the paper was also reviewed by an additional two reviewers as part of the first authors (GCB) PhD thesis submission. The thesis examiners were also experts on land-use change effects on birds. In addition to addressing the comments from reviewers for Open Science, we have also made some changes at the suggestion of thesis examiners. These are listed at the foot of our response.

Associate Editor Comments to Author:
The reviewers have provided the Editors -and hopefully yourselves -with some food for thought. Each reviewer sees value in your work, and each recommend modifications of varying degrees to improve the current paper, the most critical of the reviewers recommends a more radical step: splitting the manuscript into two parts to tell the two distinct but related stories they have identified in your submission. It is on this basis that a reject/resubmit decision has been issued (not a reflection of the quality of the work per se).
We'd like to give you the choice of either revising the current manuscript to address as far as possible the concerns raised during this round of review and so get the paper across the line largely in its current format (we'll ask the existing reviewers to comment on the changes made).
Or you may take the opportunity to divide your work as suggested by the reviewer and use the resubmission as the basis for one of the two papers, and then preparing a second manuscript for submission and consideration alongside the resubmission (again, we'll ask the current reviewers for their thoughts).
Whichever approach you take, we'd be grateful if you could contact the editorial office to let them know, so we can take the appropriate steps post-revision.

Good luck!
Thank you for giving us the option to resubmit the manuscript as two separate works. After considerable thought, though, we have decided that the paper is stronger as one. Please see comment 5 below for our reasoning.

Reviewer: 1
This manuscript, by Bain and collaborators, summarises a retrospective series of comparisons of bird occurrence in the midlands of Tasmania, as part of ongoing restoration efforts aimed to improve wildlife habitats in the region. The basis of the work is two sets of bird surveys collected at thirty three sites in 1996/1997 and 2015/2017, with another 39 sites added for the recent round of Appendix B surveys. This manuscript summarized a tremendous amount of work, both in the field conducting bird surveys and vegetation data, compiling site attributes in a GIS, and analysing determinants of bird occurrence.
I have four sets of substantive comments, relating to overall study design, bird surveying methods, analysis and inference as well as various minor quibbles.
1) Study design. Using effectively two data points (occurrence patterns of birds in time period 1 and 2, twenty years apart) makes it very hard to identify trends, especially for an assemblage of animals characterized for their high vagility and often broad habitat tolerances. Rather than using this temporal dimension to frame their analyses, I would suggest running the same sets of analyses for the two datasets and looking instead at how determinants of occurrence / richness have changed, and exploring how those differences might relate to the dramatic land use changes that have occurred in the intervening period.
We agree that trends in abundance are difficult to identify with data collected from only two survey periods and acknowledge this in the manuscript (e.g. lines 600-605). It is our hope that this study provides the impetus for a continued long-term monitoring effort.
We would not necessarily expect any difference in those factors underlying patch occupancy and abundance of birds in Tasmania between survey periods, perhaps only their relative strength. We had, nonetheless, considered repeating our multivariate statistical analyses on the original survey data as suggested by the reviewer. Unfortunately, the historical data was not recorded in such a way that the mvabund analysis could be used; bird data was recorded as densities and the raw abundance data was not available. We could have run the analysis for the two data sets using only presence/absence information but feel that this would not make use of the most interesting source of variation in the data.
2) While I appreciate the method used to estimate bird occurrence was effectively decided by MacDonald and Kirkpatrick for the 1996/97 survey period, many of the concerns I have with the data relate to how it has been handled, rather than initially collected. Variable effort was used which is not necessarily problematic. Rather, by adjusting their data based on proportion of the site surveyed, a systematic error forcing a species-area relationship might be the real problem. This can be avoided by using estimated richness rather than observed richness, following the methods of Colwell in his freely available EstimateS software. They can then look at how much the predicted richness diverges from observed richness and, if there are consistent effects of sampling effort, they can be quantified and more precisely dealt with. As for density, I have even more concern with the response variable used, and would recommend the authors follow the work of Mac Nally and others and uses incidence or reporting rate instead. That is, the proportion of surveys conducted in that site in which the species was detected as present. It is coarse, but far less prone to the many confounding effects of variable detectability, (between species, sites and seasons).
Transect surveys were proportional in their length to the size of the woodland surveyed and so reflect a proportional survey effort. If we were to standardise the length of transects across woodland remnants of different size, we would fail to account for the greater diversity of habitat types, and therefore birds, in larger patches of woodland. Our measure of species richness was equal to the accumulated total number of species from our seasonal surveys (now clarified on lines 280-281).
We only used species richness in one analysis. This was concerned with detecting changes in richness within woodlands and between survey periods and then identifying potential factors that may explain these changes (e.g. changes in woody veg cover & noisy miner density). Because we were only interested in change in species richness and we used the same bird survey techniques as in the earlier survey period, we do not think it necessary to calculate predicted richness in this instance. We are thankful to the reviewer, however, for introducing us to EstimateS; the software looks very useful! Reporting rates are most appropriate for large data sets with more than two data points (i.e. not ours, as indicated by the reviewer in their first comment). Reporting rates have, for example, proven very useful for analysing data collected through citizen science projects but this metric requires all count data to be simplified to presence/absence. It is in our opinion that this would result in a loss of the most interesting variation in our data set.
Densities for some species did not change between the two survey periods (20 years apart), providing some indication that our comparisons are robust to differences in detectability between survey periods. After consulting with an ecological statistician, we decided that occupancy modelling, which accounts for differences in species detectability, was unnecessary. Our study sites were largely open habitat types, surveyed repeatedly and there are very few cryptic bird species in the region.
3) The analyses used were complex and, although well explained, the authors side-stepped what I took away as one of their main findings-the avifauna changed very little between sampling periods, and most of the factors measured had little explanatory power, either at the individual species scale or for variously-defined groups. There was also a very large number of potential relationships to explore--figure 4 alone is a distillation of approximately 600 univariate relationships. The determinants of species richness very murky, with more than half the variation unexplained. Yes, this is a negative result, but if that result remains using predicted richness where comparability across sites is assured, then the authors should stand by it, and discuss this lack of signal more explicitly. The fact that these site-based factors had so little explanatory power suggests that landscape scale factors might be more important, that interspecific differences and the masking effects of simultaneous positive and negative relationships might conceal any assemblage-wide trajectories.
Changes in avifauna were very much dependent on the field site. Some woodlands experienced a dramatic change in species richness, but yes, for many sites there was little to no difference in this measure ( Figure S2). We disagree, however, with the wider premise that there was little change in avifauna between survey periods. Some species such as cockatoos and ravens have significantly increased in number (reflected in Figures 5 & 6) while others appear to have declined (e.g. pardalotes, Fig. 5). Similarly, we do not feel that the determinants of change in species richness were murky, constituting a negative result. We modelled change in richness using just three predictor variables and achieved an adjusted R-squared value of 0.45. This is a good fit when considering such complex ecological interactions (see Møller & Jennions 2002).
We understand that by grouping species into assemblages based on diet, body size etc. this masked simultaneous positive and negative relationships and increased the associated confidence intervals in our analyses of change in bird density. This was, however, a key aim of the study -to test for common trends among groups of birds that share life history traits.
4) Related to point 3, I would suggest the authors be a little more circumspect about their inferences. In figure 5, temporal changes for 33 species are depicted of which more than half have confidence intervals that overlap with zero. To me, this suggests that either most species aren't changing or the approach used to monitor them across this period is unable to discern any change. Likewise, in figure 6: just four of the 21 trait-defined groups exhibited significant changes in density: waterbirds, raptors, large birds and plant eaters. Rather than any trait-based influence, these four groups likely reflect the clear increases noted in swamp harriers and shelducks.
5) I would recommend the authors consider subdividing this large manuscript into three more narrowly-defined contributions. The noisy miner story might well be worth carving off and exploring on its own. [Oh, and please don't use the term "reverse keystone". Keystones are by definition rare in their communities, in terms of numbers of individuals / proportion of biomass-the term "despotic" is far more appropriate]. The determinants of diversity in two different time periods would also stand alone, leaving perhaps a third manuscript considering what you've found overall, what appears to be important and not important in structuring these bird assemblages in a highly modified system, how you might expect birds to respond to planned restoration, and some guiding predictions to revisit in another twenty years. This is all up to you but by lumping all of this into one paper, not only does it become unwieldy, but you simply don't have the space to consider your findings. There's plenty of interesting comments made in passing about particular sites, land-use practices, positive and negative responses to shifting land use. Reworking the paper into more clearly defined narratives will give you the structure and space to explore these in more depth, and make better use of all the hard work that went into compiling all of this information.
We have given this suggestion some serious thought and acknowledge that the manuscript is quite lengthy. Nonetheless, we have decided to leave the paper as one for the following reasons.
First, relationships between noisy miners and Australian bird communities have been extremely well-studied at the habitat patch and landscape levels (see Maron  Second, our data set is not strong enough to justify a stand-alone study of temporal comparisons in bird species presence or abundance, with only 33 sites where surveys were repeated 20 years apart and no intervening records.
Lastly, we believe the advantage of the manuscript as it currently stands is that our analysis of those environmental factors correlated with patterns of bird species presence and abundance is complemented by findings from the temporal comparison in bird data. The following comments were received from two reviewers of the PhD thesis that this manuscript draws from: "What I liked most about it was the multi-pronged approach used to explore the different mechanisms behind patterns of bird occurrence in the Tasmanian Midlands, going well beyond a focus on describing landscape and habitat correlates. This is done too, though, and yields more insight than usual as it combines an analysis of contemporary patterns with a temporal comparison, and together these approaches paint a much richer picture of the conservation ecology of this bird assemblage." "The [manuscript] is very comprehensive and having analysed very similar data to this in the past, I commend the student for doing an excellent job on data which can be quite challenging to analyse well." We have further reduced the length of the manuscript as far as possible.

Reviewer: 2
Bain et al use bird surveys (past and present) to compare species and their traits with habitat characteristics in the Tasmanian midlands. Over all I think the paper is well written, the methods appropriate and well executed and the results interesting and well interpreted. I particularly like the use of the fourth corner approach (and the resulting figures) to addressing the question of how to link species occurrence data with complex habitat variation. My comments on are relatively minor, but I would suggest that the authors look out for over-long sentences in their revision, and attempt to clarify some parts of the methods as suggested below. The manuscript could be trimmed in some places to save words.
6) Methods -generally fine, but more detail is needed about the method used in the original surveys -given the time that has passed since the first surveys, it is important to understand the possible factors that could confound the results between the past and now. E.g. was observer error quantified and if not, then this assumption should be spelled out more (the first mention I see is at L428 which is a bit late). I see GB did all the field work in the contemporary surveyswho did them in the past? Is it possible to quantify detection error between the surveyors?
Added: Historical surveys were also completed by a single observer (MM).
Michael Macdonald (co-author) completed all the original surveys. We were not able to quantify observer error. The strip-transect survey method used in both survey periods is described on lines 160-165. 7) L157: presumably the orientation and location of transects was different due to change in vegetation cover (or not?) between survey periods? If yes, say so.
Transects location and orientation was the same between survey periods, even at sites where clearing had occurred. Cleared areas never overlapped with the location of the original transects. The only site where transects did differ in their location (Woodstock Lagoon) was excluded from all analyses of change in bird communities/species richness. This is stated on line 169.

8) L158
: if I understand correctly, small patches received 1 * 100m survey transect and large ones received 4 * 200m transects back-to-back over 800m, plus an addition 20min/2ha? was the orientation random in large woodlands? Why you used 2 survey approaches (transects and then 2ha searches) -did you pool the data for analysis or were they used for different parts of the analysis? It's not clear to me how these different methods are used in analysis Correct, these are good points. The orientation of transects in large woodlands either replicated that of the earlier survey or was randomly located but always began at the edge and was directed towards the woodland interior. We have added the following on line 162 to clarify this point: These [transects] were placed randomly if their location had not already been established in the historical survey period.
We have deleted the reference in our methods to 2ha surveys in woodlands because this was also a source of confusion for other reviewers and, ultimately, data collected in these concurrent surveys was not analysed in this manuscript. Mention of 2ha surveys is now restricted to a single sentence in the Results: Seven additional species were detected and identified from calls offsite, eight more were observed during concurrent 2-hectare/20-minute bird surveys in woodlands (not analysed in this study, Table S2) and four were found only in either pastures or grasslands (Table S3).
Additional 2ha surveys were conducted for two purposes. First, to provide data that was useful for Birdlife Australia (2ha-20mins is the standard survey method for their data base). Second, these were conducted with the hope of comparing detection probabilities of birds using different survey methods. That is, comparing species lists derived from 2ha searches in the middle of woodland sites to transect and acoustic surveys. We did not have time to analyse data from these surveys and as such have deleted the relevant lines from the manuscript. 9) L198: "as a proxy for land use intensity." Change accepted.
10) L204: specify here where you sourced miner density estimates -is this derived from your own models?
Miner densities were derived from the bird surveys conducted in the 2017 survey period. We have added the following: Density of miners was determined from bird surveys as described above (see 2.3).

11) L265
: I understand this limitation of mvabund but think it would be helpful for readers if you provided some explanation of what this might mean for interpreting your potential results, especially since the large forest patches had quadruple the effort at transects twice as long as those in small patches We contacted an ecological stats consultant regarding this matter to see if there was any workaround. It was explained to us that offsets (to account for differences in the area surveyed) are not able to be used only when using LASSO for model fitting and selection but that this method was still the most suitable for our data. We have added this detail to line 272. In addition, we now include the following… We agree but understand that the figure would no longer meet the formatting requirements of the journal (too large for the page).
Changed to: The number of pivot irrigation systems increased dramatically between 1997 and 2017, from three irrigators within 1 km of two survey sites, to 33 irrigators at 12 survey sites. Yet, there was no relationship between the number of nearby irrigation systems and changes in species richness.
Hopefully this is a little better.
15) L440 and L671: remove 'productive' -plantations are referred to in other ways too elsewhere in the paper -be consistent We have removed "productive" from lines 447 & 680. We have left the term in the first instance (line 199) to help establish that commercial eucalypt plantations are separate from the small eucalypt plantings established for biodiversity or shelter purposes on farms.
17) L468: re 'common in urban environments' either needs a citation or delete (relevance to the section seems tangential anyway) The comment is relevant because it identifies the bird as a species associated with land use change, in this case urbanisation. This is a personal observation.

18) L611: flock sizes are results
These flock sizes were not recorded during our formal surveys, however, and so we did not want to conflate these with our more substantive survey data.

Reviewer: 3
A very good paper. See attached file for some comments.
I have copied the comments from the pdf and address them here.
This is an excellent paper on a subject of global interest. It is well written, and the literature view is very thorough for recent studies. The paper can be published with very little change. However, the following points could add some value: Figure captions: The term "response ratio" could imply a response to an identified process such as agricultural intensification, whereas in fact it just refers to a change over time between the two survey periods (with agricultural intensification as one of several possible causal factors). Would "percent change" be a less misleading term?
We understand the problem with this terminology but the log-response ratio well as in meta-analyses. For this reason, we feel that it would cause more confusion if we were to change the term in our paper. "Response ratio" only describes the statistical metric used and is not meant to imply the cause of change. Percent change is a different measure of effect magnitude.
20) The trait-based approach is clearly useful, and quite similar to the guild approach used in various similar studies. The latter approach would enable conclusions to be made about changes in the abundance of whole guilds, e.g. nectarivores, canopy-foraging insectivores or small birds and large birds, rather than just saying that several species with those traits increased or decreased. It seems that calculations were made along these lines (Fig. 6) and perhaps more could be made of them in the text? I found only one reference to Fig. 6, and that just related to large herbivores such as ducks (line 465).
We tend to agree. Specifically, we have realised that the Abstract had failed to mention the results of our trait analysis altogether. We have rewritten the abstract to include some details of the trait analysis.
Fourth-corner modelling showed strong negative responses of aerial foragers and exotics to increasing woodland cover; arboreal foragers were positively associated with projective foliage cover; and small-bodied species were reduced by the presence of a hyperaggressive species of native honeyeater, the noisy miner (Manorina melanocephala).
Having said this, the large confidence intervals (noted by Reviewer 1) for most trait groups in Fig. 6 overlap zero. Therefore, we did not feel it appropriate to comment too much on those insignificant results.
21) The basic results (as shown in Figure 2) are similar to those from a study of fragmented forests in south-eastern Victoria (Loyn 1985(Loyn , 1987, which showed the role of noisy miners, and gave estimates of likely declines in numbers of species over time. The study was repeated 22 years later, and the further declines were found to be less than predicted at a set of sites where eucalypt plantations had been established nearby (MacHunter et al 2006). It may be worth comparing these results.
We were not aware of these studies. They are very interesting but for the purposes of keeping the manuscript as short as We have changed the title such that it no longer includes "long-term".
Thanks, change made. We have added the appropriate years.
26) Could also refer to the early work on this subject: Dow (1977) showed that Noisy Miners exclude other birds, and Loyn (1985) showed the importance of this process in fragmented forests in rural landscapes (in Victoria).
We have now included Dow (1977) among our references.
27) Interesting that Striated Fieldwrens were recorded at all (albeit offsite). In Victoria I have also found them in rural environments (in weedy young pine plantations and woody weeds in farmland) but quite rarely (and not for many years): they remain common in or near saltmarsh and sparsely treed heath, but otherwise rare in farmland away from the coast. [This is just a comment for interest: no changes needed unless you wish to say something about the habitat where they were found.] Yes, we believe them to be quite rare on farmland in our study region. For your interest, we have observed striated fieldwren at two locations: near a very isolated single shrub in the centre of a bare paddock; and in another paddock where tall grasses and gorse shrubs were present. We have rewritten this sentence to avoid misinterpretation. 31) L614: Did you class corellas as introduced species? I've tended to think of them as natural recent colonists (as they are in southern Victoria), and this is supported by the first record of Little Corella being in 1983 (a major drought year in SE Australia). However, I notice they are listed as introduced to Tasmania in the State of Tasmanian Birds reports: could be worth checking.
The Tasmanian government list corellas as an introduced species, it was for this reason that we also classed them as introduced. This is noted in Table 2, paragraph 5 of the Results and in the Discussion. We cannot definitively say whether corellas were self-introduced to Tasmania or not (much like debate over other parrots such as galahs in Launceston).
32) L635-665: A study in Victoria found that Noisy Miners rarely use young commercial or amenity eucalypt plantations ), though parallel work found that they were often present in mature linear planted shelterbelts. Thanks for the references, they will be useful in future. We now cite the Loyn (2007)   Appendix -Responses to additional comments made by thesis examiners since original manuscript submission. I have removed any reference to "plots" when describing the collection of bird survey data. This term is now reserved for the first paragraph of "Environmental Data", in which I explain methods of collecting vegetation data. The term "transects" is only used in reference to bird surveys. I have removed "sampling points" from the caption of  Table 3.3. The new analysis now identifies Landcover diversity and Leaf litter cover as having significant effects on community composition of birds at survey sites.

References
The order of significance of other environmental variables is also different but the overall interpretation of my data has not changed dramatically.

12) [Thesis Rev 1] L 1651 -density is per unit area -indicate this (is it per ha?)
Correct. I have added the appropriate units (hectare -1 ).

13) [Thesis Rev 1] L 1772 -what do you mean 'should be avoided' -can you explain how?
Was: Thus, such dramatic changes in community composition are consistent with the impact threshold proposed by Thomson et al. (2015) and should be avoided in restored habitat. Now: Thus, such dramatic changes in community composition are consistent with the impact threshold proposed by Thomson et al. (2015); this represents an effective target for the maximum density of noisy miners that could be tolerated by other birds in restored habitats.
14) [Thesis Rev 1] L 1872 -these were both short-term studies -not long-term Correct. I have changed "long-term" to "short-term".  30 Changes in species richness were best explained by changes in noisy miner 31 abundance and levels of surrounding woodland cover. We make encourage 32 restoration practitioners to trial novel planting configurations that may be 33 resilientconfer resistance to invasion by noisy miners, and practical 34 recommendations to restore habitat for terrestrial birds of Tasmania and encourage 35 a continued long-term monitoring effort such that theto reveal effects of future 36 land-use change on Tasmanian birds can be examined.

Introduction
37 Agricultural intensification is a major cause of global biodiversity loss [1].    225 Table 1. Summary of landcover types within 1 km of sampling pointsthe centre of 226 bird survey transects at woodland (n = 52) and planting sites (n = 5).

262
Candidate environmental variables were reduced to a final set of 10 (Table 3) by 263 generating a correlation matrix (Spearman's rank) and eliminating those variables 264 that demonstrated high levels of collinearity (r s > 0.6). One exception was that we 265 included both patch shape and woody vegetation cover (r s = 0.78) in our models. We 266 did this for three reasons: bird species respond differently to each of these habitat 267 characteristics; both were considered to be biologically informative; and removing 268 either variable would lead to biased estimates of importance for other predictors.

269
Patch shape was also strongly associated with patch size (r s = 0.94), as was the cover

295
We also used traitglm to model bird count data without specifying a trait matrix.

296
This method effectively fits a multivariate species distribution model and assumes a 297 different environmental response for each species. Unfortunately, mMvabund does 298 not yet support the use of offsets when modelling fourth-corner problems and using 299 LASSO in the model fitting process. While the traitTherefore, while the results of this 300 analysis does not, therefore, account for differences in the area of each transect 301 surveyed, the results remain informative. remain informativeObserved traits were 302 unrelated to woodland patch size and so this limitation is expected to have minimal 303 influence., they do not account for differences in the area of each transect surveyed.

304
In our fourth analysis, we used GLMs to assess which elements of landscape change

316
We used an information theoretic approach to assess model performance and 317 ranked models by Akaike's information criterion corrected for a small sample size

319
Finally, we used the log-response ratio (lnRR) to assess changes in species densities

328
A total of 91 species was recorded across all our bird surveys (Tables S1-S3) miners and a greater range in patch size, woodlands classed as medium size varied 347 greatly in bird community composition (Fig. 2). Some medium-sized remnants 348 supported species that were typical of large intact woodlands, while others were 349 most similar in species composition to small and degraded remnants. Thus, when 350 species were pooled across sites, medium remnants were the most species-rich (58 351 species, n = 25) followed by large (52, n = 12) and small remnants (48, n = 15), native 352 grasslands (31), plantings (25), gorse (23) and pastures (20).

404
Life-history traits of birds explained their response to environmental variables. The

405
negative influence of noisy miners on small birds was once again highlighted by our 406 trait analysis (Fig. 3). Large birds showed no response to miner density but were 407 more common at lower elevations and at sites with a greater diversity of 408 surrounding landcover types. The positive association between miners and birds 409 introduced to Tasmania reflects a higher abundance of little and long-billed corellas 410 (Cacatua sanguinea, Cacatua tenuirostris) and laughing kookaburras (Dacelo 411 novaeguineae) in miner-dominated woodlands (Fig. 3). These three species have 412 been introduced to Tasmania from elsewhere in Australia. However, species exotic 413 to Australia were negatively associated with miners. Exotic species also showed a 414 strong negative response to increasing cover of woody vegetation and were more 415 common in woodlands with a greater perimeter-area ratio (i.e. shape) and where 416 the basal area of tree stands was high.

417
The diet and foraging habits of birds also moderated their response to 418 environmental variables (Fig. 3). Arboreal foragers were more abundant at sites with 419 greater FPC, more leaf litter, fewer miners and at higher elevations. Aerial foragers 420 were negatively associated with woody vegetation cover and FPC but preferred sites 421 with a simplified groundcover and with a more diverse composition of surrounding 422 landcover. Aerial foragers were also associated with woodlands that had a simplified 423 groundcover and a higher perimeter-area ratio. Birds with a mainly granivorous diet

432
The diet and foraging habits of birds also moderated their response to 433 environmental variables (Fig. 3). Arboreal foragers were more abundant at sites with 434 greater FPC, more leaf litter and at higher elevations. Aerial foragers were negatively associated with woody vegetation cover and preferred sites with a more diverse 436 composition of surrounding landcover. Aerial foragers were also associated with 437 woodlands that had a simplified groundcover and a higher perimeter-area ratio.

438
Birds with a mainly granivorous diet were more common in woodlands at lower 439 elevations and where FPC and woody vegetation cover were low. In contrast,

440
generalists and nectarivores were more abundant with increasing cover of woody 441 vegetation.

442
Our species distribution model indicates more complex relationships between 443 environmental variables and individual species (Fig. 4). For example, sulphur-crested 444 cockatoos (Cacatua galerita) were more likely to be found in areas of high woodland 445 cover but also preferred higher landcover diversity, higher groundcover diversity and recorded than previously (Fig. S2). The best model included just these three 472 predictors and explained 45% (Adj. R 2 ) of the variation in the net change in species 473 richness. The top 3 models, however, were all within two Δ AICc units and carried 474 38%, 23% and 16% of the weight respectively (Table 4).

475
Only three survey sites experienced a reduction in patch size from clearing. Thus, our   (Fig. 5). This was also reflected in the lnRR 530 of birds with a herbivorous diet and very large body size, which were mostly species 531 of ducks (Fig. 6). Granivoresous and raptorsial birds also appeared to be more

539
Of the 10 species with the lowest response ratios (i.e. less abundant in the recent 540 survey), two are endemic to Tasmania (yellow-throated honeyeater Lichenostomus 541 flavicollis and yellow wattlebird Anthochaera paradoxa) and 8 are arboreal foragers.

542
The exceptions were eastern rosellas and the yellow-rumped thornbill (Acanthiza    that were negatively associated with leaf litter cover tended to be more common in 632 degraded woodlands where pasture grasses were the dominant groundcover (e.g.

634
Elevation had a strong influence on community composition but only one species, 635 the grey fantail, was found to contribute significantly to this effect. The apparent 636 negative relationship between grey fantails and elevation could be the result of 637 seasonal migratory behaviour in this species, which is poorly understood in Tasmania

638
[73]. Some birds, such as the crescent honeyeater, are known to migrate altitudinally 639 in response to changes in food availability but individual trends for these elevational 640 migrants could have been masked in our analysis because we combined survey data 641 from different seasons. Given that our study sites were mainly restricted to the rift 642 valley of the Midlands, our surveys were not designed to reveal the full effects of 643 elevation on Tasmanian bird assemblages. Nonetheless, our models showed that 644 endemics like the black currawong, black-headed honeyeater and yellow-throated 645 honeyeater were more common at higher altitudes.

647
The bird community of Tasmania is distinct from those of mainland Australia but is 648 also relatively depauperate [25]. This could result from a dispersal filter to the island 649 or Tasmania's climatic suitability. Of the terrestrial bird species that settled in 650 Tasmania, and when compared to those of mainland Australia, few have been 651 recognised as of 'conservation concern', 'decliners' or as sensitive to the area of

656
-species of conservation concern are yet to be identified -or might otherwise 657 indicate that local species are more resilient to the impacts of land-use change and 658 are not at overall risk.

659
Of the 51 species that we examined, seven showed significantly lower current 660 densities than in the 1997 survey period, but without data collected from the 661 intervening years, we cannot be certain if these differences truly reflect population Populations of raptors and other large carnivorous species like the forest raven may 687 have also increased. This could be because of a greater availability of animal value of restoration efforts, however, competition between birds and the specific 783 habitat preferences of individual species must also be addressed.