Drivers of Tree Canopy Loss in a Mid-Sized Growing City: Case Study in Portland, OR (USA)

Round 1

Reviewer 1 Report

The article as a whole is interesting and relevant. But scientifically, there are questions. The research period is very short, which makes it impossible to assess long-term trends and consider the results as reliable. The composition of the factors under consideration is very diverse – economic, social, demographic, planning, etc. It is difficult to compare them, assess their significance and contribution to the processes of changing the areas of green spaces. It seems that the management of the urban area, including green spaces, their shares and distribution on the territory of the city is mainly carried out within the framework of the development of urban planning documentation. That is, it is a very subjective process. It is not entirely ethical to divide the population into whites and non-whites, rich and poor. Although, if this is accepted and allowed in the United States as part of scientific research, then the remark is removed. It is difficult to understand from the materials presented what is primary: probably, colored and not very wealthy people are forced to settle in cheaper housing and in areas where there are obviously more dense buildings and few green spaces, and not vice versa, that due to the presence of these population groups in their areas of residence there are few green spaces. For these reasons, in our opinion, the results of the work, as the authors themselves note, are contradictory and unexpected: too short a period of time is considered and very diverse factors. A small note: the words from the title of the article should not be duplicated in the "Keywords" block. However, in general, the direction of research seems promising.

Author Response

Comments from Reviewer: The article as a whole is interesting and relevant. But scientifically, there are questions. The research period is very short, which makes it impossible to assess long-term trends and consider the results as reliable. 

Author’s Response: Conducting longer-term assessments of urban tree canopy requires commensurate data that are at the same spatial resolution. The major change in available tree canopy data for the study region came about in 2012 with GeoEye and other proprietary datasets; however, they were replete with errors and did not have consistent coverage of the study area until 2013. Accordingly, when the datasets became available for the earlier period, our team started to assess these higher-resolution data. We require higher resolution data because we are aligning these data to tax lots/parcels, which are smaller in size. In addition, these datasets allow for the identification of smaller changes in the built environment that can help to explain the reasons for losses in tree canopy. For example, the relationship between planted and/or cut trees with canopy loss and between residential stability and canopy loss could be identified during this study period. In addition, recall that several major policy developments occurred in 2014, including the creation of the City’s first Urban Forestry Code, called Title 11, which aimed to regulate trees and ensure better coordination across the City. As such, this manuscript focuses more on recent changes (and explanation for those changes) to tree canopy and a policy assessment, than a longer-term description of trends. Creating a longer-term assessment is beyond the scope of this analysis.

Comments from Reviewer: 2. The composition of the factors under consideration is very diverse – economic, social, demographic, planning, etc. It is difficult to compare them, assess their significance and contribution to the processes of changing the areas of green spaces. It seems that the management of the urban area, including green spaces, their shares and distribution on the territory of the city is mainly carried out within the framework of the development of urban planning documentation. That is, it is a very subjective process. 

Author's Response: Indeed, we agree with the reviewer that decisions about urban development are subjective processes. However, the purpose of this research is to illuminate whether those subjective decisions have objective patterns that are recognizable by examining changes to the physical built environment, including the presence/absence of tree canopy. Our analysis reveals, for example, that the subject nature of these individual decisions about urban development creates a collective impact on the presence of tree canopy, which is quantifiable and attributable to specific codes and regulations. Further, if our aim is to support verdant urban green spaces, then we will need to ensure that we continuously correct the unintentional consequences of subjective decisions that, when taken together across a city, will preserve tree canopy. If the reviewer is suggesting that the explanatory factors (e.g., land use, demographics, etc.) are subjective, then we point the reviewer to other studies that have attributed some of these factors to changes in the tree canopy (see for example, Biwas et a;. 2020). We posit that examining changes to tree canopy can offer a means for understanding how the composition of factors are not accounting for changes. 

Comments from Reviewer: It is not entirely ethical to divide the population into whites and non-whites, rich and poor. However, if this is accepted and allowed in the United States as part of scientific research, then the remark is removed. It is difficult to understand from the materials presented what is primary: probably, colored and not very wealthy people are forced to settle in cheaper housing and in areas where there are obviously more dense buildings and few green spaces, and not vice versa, that due to the presence of these population groups in their areas of residence, there are few green spaces. For these reasons, in our opinion, the results of the work, as the authors themselves note, are contradictory and unexpected: too short a period of time is considered, and very diverse factors. 

Author’s Response: An extensive body of literature divides the population of the United States into Whites and non-Whites, and the present analysis is consistent with these earlier studies. In terms of the selection of variability, we selected a diverse set of variables that are intentionally not related to each other; rather, we tried to identify orthogonal variables that are known – through earlier literature  – as helping to explain the presence of canopy cover. The results of the study may not be generalizable for other cities, as the reviewer mentions, though the results could inform policymakers and researchers to broaden perspectives on canopy preservation. In terms of the length of the study, we would select a longer period, though the datasets are incommensurate, and will not yield results that are replicable or consistent.

Reviewer 2 Report

Dear Authors,

I’ve examined your manuscript with big interest. Indeed, this is the example of the very interesting and stimulating research project with clear and internationally important research questions, strong methodology, and deep interpretations. The manuscript is well-thought and well-organized. It is also written properly and illustrated appropriately. Generally, this manuscript is strong in many aspects, informative, and even provocative (in good sense) in somewhat. The authors’ story is well related to the issues of sustainability. Principally, I like this work and find it suitable to “Sustainability”. Nonetheless, I have several recommendations, which, I hope, you can find useful.

1)      Subsection 2.1: please, tell more about the geographical setting of Portland – landforms, climate, hydrology, natural vegetation, density of population, industries, urban area geometry, functional zones, etc. Do not forget to cite the basic sources of the geographical information.

2)      Figure 1: source of image?

3)      Table 1: please, either use metric units (sq. km or sq. m instead of acre) or, at least, explain what is acre. Also, source abbreviations should be explained beneath the table, and the links (citations or URLs) should be given to each of them.

4)      Subsection 3.1: what are the common trees available in this city? Can you give some photos?

5)      Figure 2: please, indicate the units for both horizontal and vertical axes on the both graphs.

6)      Figure 3 and 4 (and also the related text blocks): how reasonable is to mix metric and US units? To me, sq.m/acre is a non-sense unit.

7)      Subsection 4.6: to whom your recommendations are addressed and how they can implement the related changes? Who should pay for this? Also, move the limitations to Conclusions.

8)      Conclusions: please, move your comparisons to the previous studies to Discussion, and try to extend these comparisons via the considerations of some examples from the other countries. It would be worth to think whether your results are meaningful to the only USA.

9)      Conclusions: please, re-organize this section as follows: the numbered list of 5 main findings (2-3 from Results and 2-3 from Discussion), the limitations, and the perspectives for future research.

Author Response

Subsection 2.1: please, tell more about the geographical setting of Portland – 

Author’s Response: We added additional information about Portland, including the climate, population density, income distribution, and land use zoning. (Section 2.1 Study Area, pages 3 & 4)

Figure 1: source of image?

Author’s Response: We included the source as NAIP imagery from GIS Basemap in Figure 1. (page 4)

Table 1: please, either use metric units (sq. km or sq. m instead of acre) or, at least, explain what is acre. Also, source abbreviations should be explained beneath the table, and the links (citations or URLs) should be given to each of them.

Author’s Response: We have now explained the unit of acres(line 183 - 186), which is commonly used in the United States, and will be relevant to audiences who are attempting to apply these results to their region. We have also explained the abbreviations beneath the table. Links to the available source data are included in the citation. (Table 1, page 7 & 8)

Subsection 3.1: what are the common trees available in this city? Can you give some photos?

Author’s Response: We have not included the common names of the trees in the manuscript because the specific species of trees is not relevant to our research questions, though we can provide some information – from the City of Portland’s publicly available website – here:  (https://www.portland.gov/trees/native-and-nuisance-trees/native-tree-list), In fact, within the study area a street tree inventory indicates that many different types of trees are common, including both evergreen and deciduous trees. Based on the inventory Cherry tree(Prunus app.) was most commonly found street tree followed by Norway maple(Acer platanoides) and Red maple(Acer rubrum). 

Figure 2: please, indicate the units for both horizontal and vertical axes on the both graphs.

Author’s Response: We have included units for the canopy density, which is square meters per acre, as noted on (Figure 3, page 9) 

Figure 3 and 4 (and also the related text blocks): how reasonable is to mix metric and US units? To me, sq.m/acre is a non-sense unit.

Author’s Response: While square feet is commonly used in the United States, the unit of meters is also applied, usually in scientific articles. In fact, the unit canopy density (sq.m/acre) is explained in the method section, and offers a means for illustrating a density unit that does not have a consistent measure in urban forestry. It can be converted to canopy percentage(%) by dividing the canopy density by 400, which is the conversion (acre to sq meter) - 1025, multiplied by 100 (conversion of percentage). We provide some related information in the manuscript. Section 2.2 Data and Variables, page 5. 

Subsection 4.6: to whom your recommendations are addressed and how they can implement the related changes? Who should pay for this? Also, move the limitations to Conclusions.

Author’s Response: The intended audience for the recommendations is policy makers and urban forestesters, who are often involved in supporting preservation efforts.The entities often are involved in public administration, and they are able to access general fund dollars to ensure canopy preservation.  That said, we are interested in directing the recommendations to those generally interested in understanding the factors that contribute to losses in tree canopy, including those who serve on voluntary citizen advisory boards, arborists, and housing advocates. The Discussion section includes language that aims to provide specific locations of the city where canopy preservation policies will need to be enacted, which we expect that these audiences will take relevant action. Further, our methodology allows for other cities to pin-point the locations where losses are occurring at a disproportionately high rate, to which similar preservation policies can be applied. We have also moved the Limitations to the Conclusions. 

Conclusions: please, move your comparisons to the previous studies to Discussion, and try to extend these comparisons via the considerations of some examples from the other countries. It would be worth to think whether your results are meaningful to the only USA. + Re-organize the conclusion.

Author’s Response: We have moved several sentences from the Conclusions to the Discussion section. (page 13, line 445 - 453). Also, we have re-organized the sections accordingly. 

Reviewer 3 Report

This manuscript delves into an important topic - the importance of urban tree and tree canopy in contemporary cities. Overall, the research is well presented in a well written English.

However, I suggest some improvements.

First, I suggest including more maps on the spatial distribution of the main factors being analysed and influencing UTC losses. It would allow for a better understanding of Portland and its urban dynamics. The authors only present the maps of spatial distribution of UTC losses. Where are the areas of low income population in Portland? How is the population density distribution in the city? Etc!

Second, in lines 504-509 and 589, it is referred that “multi-family housing units and a higher floor area ratio should be encouraged”. Although, this urban solution may lead to some increase in green infrastructure and the UTC, as suggested by the authors, the solution can be easily controversial. Personally, I do not agree, as solutions of urban form should be contextualised. The recommendation of the authors implies a modernist way of building cities, which is far from being the most sustainable. This kind of recommendation should be further discussed and argued.

Third, and finally, the study has important limitations, as assumed by the authors. Conclusions are based on the “causality between identified drivers and canopy loss definitively”. I suggest there should be an effort to better support the analysis of the three issues presented: white population, residential stability, and tree data points from FOT. For example, regarding the decrease in the white population, there are certainly other studies or variables that can help understand the reasons behind the decrease.

Author Response

This manuscript delves into an important topic - the importance of urban tree and tree canopy in contemporary cities. Overall, the research is well presented in a well written English. However, I suggest some improvements.

First, I suggest including more maps on the spatial distribution of the main factors being analysed and influencing UTC losses. It would allow for a better understanding of Portland and its urban dynamics. The authors only present the maps of spatial distribution of UTC losses. Where are the areas of low income population in Portland? How is the population density distribution in the city? Etc!

Author’s Response: We have included maps of population density and income distribution (Figure 2) in Section 2.1 Study Area, page 4.

Second, in lines 504-509 and 589, it is referred that “multi-family housing units and a higher floor area ratio should be encouraged”. Although, this urban solution may lead to some increase in green infrastructure and the UTC, as suggested by the authors, the solution can be easily controversial. Personally, I do not agree, as solutions of urban form should be contextualised. The recommendation of the authors implies a modernist way of building cities, which is far from being the most sustainable. This kind of recommendation should be further discussed and argued.

Author’s Response: One recommendation is the address the relationship between horizontal and vertical density. This study is one of the first to suggest that vertical density is different than horizontal density – offering a novel approach to thinking about ensuring and preserving tree canopy amidst rapid urban growth. Our results, for example, indicate that expansion of horizontal density would accelerate canopy loss while increasing vertical density would lower the canopy loss. Densification, as such, need not be interpreted as simply more housing units (i.e., urban development), rather, the term has two distinct meanings in our results and interpretation – in this case, increasing vertical density supports modern calls for building more compact cities, which is consistent with more sustainable development(Bibri et al. 2020). Below paragraph had added in the 4.6 recommendation section, page 16 line 627 - 639.

Our study illustrates that residential developments with more multifamily housing units and a higher floor area ratio lead to less canopy loss, while developments that require more single-family lot footprint reduce canopy. For more than a decade, Portland has been known as a compact city(Dieleman and Wegener 2004; Yang 2008; Kelley et al. 2009), which is the concept of urban form that could enhance urban sustainability(Jim et al. 2018; Artmann et al. 2019; Bibri et al. 2020). The compact city concept encourages high-rise residential, mixed-land use, and efficient energy use including extensive public transportation(OECD 2012; Bibri et al. 2020). Therefore, Portland had characteristics of compact urban form with high-rise housing, and the result of the study recommends increasing multi-family housing.  However, in other places, this recommendation would not be generalizable and simply adapted as urban forestry planning should be carefully integrated into the different types of urban forms(Jim et al. 2018). At least, the result of the study could suggest that two types of density – horizontal and vertical density – could have different impacts on canopy change, so they should considered separately.

Third, and finally, the study has important limitations, as assumed by the authors. Conclusions are based on the “causality between identified drivers and canopy loss definitively”. I suggest there should be an effort to better support the analysis of the three issues presented: white population, residential stability, and tree data points from FOT. For example, regarding the decrease in the white population, there are certainly other studies or variables that can help understand the reasons behind the decrease.

Author’s Response: We have included additional information and background to support these three areas of interest. Of particular interest is our description of the changes in the White Population and residential stability in the background and Discussion sections. Our aim is not to provide an exhaustive list of reasons for these changes in demographics and land tenure, but rather to offer speculative observations in terms of statistical correlations relevant to the research questions. In addition, the FoT data were used to examine the increase in trees in the evaluation of the study City’s tree code, which aims to expand tree canopy in low income and low canopy areas. We find that the FoT data are consistent with the City’s intentions to expand tree canopy in select low income/canopy areas; however, we also observe that in those same areas, a larger loss of tree canopy is also occurring. These findings are consistent with studies noting that urban tree canopy distribution and their changes are associated with low socioeconomic status(McDonald et al. 2021; Foster et al. 2022; Koo et al. 2023)

Reviewer 4 Report

The paper is very interesting and raises important and curent issues - we need to have in cities adequate tree canopy cover and we need to consider how to monitor tree loss and where this loss is most likely to occur.
A better description of the methodology and results will broaden the possibility of comparing them in the discussion and thus improve the quality of the whole paper.
It is necessary to describe the methodology more clearly for the reader, to show the results from different sides - maybe additional data in % could be usefull? (e.g., information in the results section and then in the discussion section will make easier a comparison of the percentage of UTC loss).

Specific comments:
L 390 - 391 numbers by how much UTC is decreasing would be useful.
L 339- 401 paragraph is unclear Please supplement with numerical results that would help to understand the results
L 410 decrease - by how much? Maybe you can give data in %? Please use specific numbers more often - data in addition lower - higher etc.
L 415 reserch question  is above - here you need to answer it already
L 417 what is this goal?
L 441 - or how? Please compare to other results
L 445 lowest - extensive canopy gain - how much? Please provide numerical information
L 489 what does CBG mean?
l. 633-635 belongs in the discussion Please move it
l 641-644 this same remark
l 648-650 is an introduction section Citations in conclusions is not needed - is distracting Here should be only conclusions from the paper

Author Response

It is necessary to describe the methodology more clearly for the reader, to show the results from different sides - maybe additional data in % could be usefull? (e.g., information in the results section and then in the discussion section will make easier a comparison of the percentage of UTC loss).

Author’s Response: We have added a few more sentences about the units used in the study. For example, the unit Canopy Density (sq. m / acre) was used to normalize all the dependent and independent variables. Normalizing the variables with other units such as sq.m or sq.km made the variable difficult to compare – either too small or too large for analysis. In addition, the Canopy Density can be easily converted into Canopy Percentage (%) by diving Canopy Density with 400 - Canopy Density x (1/4047)(acre/sq.m) x 100(%). Section 2.2 Data and Variables, page 5.

Specific comments:

L 390 - 391 numbers by how much UTC is decreasing would be useful.

Author’s Response: Based on the coefficient of the regression analysis, block groups that had planted 100 more trees in an acre of land resulted in 78 sq.m/acre canopy loss in the block group; which is 0.2% canopy loss in the block group. (page12 line 421 - 422)

L 339- 401 paragraph is unclear Please supplement with numerical results that would help to understand the results

Author’s Response: We have added specific interpretations for several coefficients in the regression analysis, including population density change, building footprint change, multifamily housing unity change, and single-family housing sales. It would help to get a numeric sense of how much tree canopy would loss by the change of such variables. 3.2 Spatial Error Model Result, page 12.

L 410 decrease - by how much? Maybe you can give data in %? Please use specific numbers more often - data in addition lower - higher etc.

Author’s Response: We have included the specific numbers, which are: 6,854 sq.m, a total of less than 1% overall. Page 13 line 442. 

L 415 research question  is above - here you need to answer it already

Author’s Response: Line 415 in the paragraph is a short summary of RQ and results before getting into a detailed discussion. We removed the research question here and directly got into the discussion of the result.

L 417 what is this goal? 

Author’s Response: We’ve noted that goal of the City’s urban forestry plan is to increase tree canopy to 33% by 2035.

L 441 - or how? Please compare to other results - 

Author’s Response: As mentioned in the next paragraph, the neighborhoods with the least canopy experienced the lowest canopy gains, while those with moderate canopy levels saw extensive canopy loss. The result highlighted the need to preserve the canopy in low-income areas while planting new trees. Our findings are consistent with a few other studies nothing that the canopy change is associated with socioeconomic status indicating that canopy change is not distributed evenly across the space(Locke et al. 2017; Chuang et al. 2017; Foster, Dunham, and Bukowska 2022; Koo et al. 2023). Considering that low socioeconomic status groups tend to have low tree canopy(McDonald et al. 2021), this change would exacerbate the disproportionate canopy distribution that could amplify environmental justice concerns.

We have included additional information about these comparator studies on pages 13, lines 483 - 486

L 445 lowest - extensive canopy gain - how much? Please provide numerical information 

Author’s Response: We have noted that changes in canopy is less than 2.5%, which is calculated by converting the Canopy Density (sq.m/acre) into percentage by dividing 400 and converting an acre to sq.m. 

L 489 what does CBG mean? 

Author’s Response: We have clarified that CBG is an abbreviate of Census Block Group, which was also noted in upper section (2.2 Data and Variables).

633-635 belongs in the discussion Please move it

Author’s Response: In the conclusion, we wrote a brief summary of the result and discussion. So, some elements belonging to the Discussion are mentioned in the Conclusion, mainly to emphasize that the current study includes unexplored results on tree canopy.

l 641-644 this same remark

Author’s Response: We have moved this short paragraph to the Discussion section. 

l 648-650 is an introduction section Citations in conclusions is not needed - is distracting Here should be only conclusions from the paper

Author’s Response: Any duplicate citations and sentences have been removed, while the limitations have been moved to the Conclusion with other reviewers' recommendations.