Effect of Anthropogenic Disturbance on Floristic Homogenization in the Floodplain Landscape: Insights from the Taxonomic and Functional Perspectives
Round 1
Reviewer 1 Report
In my opinion, this work is very thorough and well conceived. While many papers have examined urbanization effects on homogenization, the focus on floodplains, the location (Japan), and the meticulous methodology (sampling protocols, data collection and analysis) make this a significant contribution to the literature on urban impacts. The authors have a solid understanding of the literature and put their results into an appropriate context. I can find no significant problems with the paper except to suggest some heavy editing for quite a few places where the English has some awkward grammatical constructs. Other than that, I congratulate you on an excellent paper.
Author Response
Please see the attachment.
Author Response File: 
Author Response.docx
Reviewer 2 Report
Summary
This interesting study examines patterns of taxonomic and functional trait diversity of herbaceous plants and shrubs in floodplain forests of a watershed with varying levels of urbanization. The techniques used to explore these patterns are interesting, and the results intriguing.
However, greater clarity is needed in the methods section to be able to determine whether the methods used to select sites, sample vegetation, and classify urbanization are appropriate to the questions asked and conclusions drawn. It appears that a single remotely-sensed measure of vegetation was used as a proxy for urbanization, which would greatly limit the strength of conclusions about effects of urbanization that can be drawn from the comparison.
While the writing is generally clear, the manuscript would benefit from additional English-language review to avoid ambiguous word choices and resolve some repeating grammatical errors.
Introduction
The literature cited in this section represents current thinking in functional and taxonomic diversity. This area of ecological thought is presented in detail, as appropriate to the topic of the study.
Statements regarding urban environments (e.g. lines 45, 49-50) could be improved by consideration of studies in urban landscape ecology that have shown considerable fine-scale patch heterogeneity. Homogenization should also be considered in relationship to patch heterogeneity. In urban environments, this may depend on type of vegetation sampled, due to the degree of difference in biotic and abiotic conditions between patches. Conclusions based upon findings of entire urban floras (such as those of McKinney that are heavily relied upon in the introduction and discussion) differ from those that can be drawn from a single vegetation type such as lawns or, as in this case, floodplains.
In some instances, statements are not accompanied by support from the literature where such support would both lend weight to the argument being made and provide the reader with links for further exploration (e.g. lines 65, 79-80).
The introductory presentation of concepts surrounding beta diversity and scale would benefit from reorganization into one concise section. Similarly, the terms “environmental factors” and “urban matrix” should be clearly defined in the introduction because they are important to the rest of the study.
Objectives and Hypotheses
- Urbanization “high intensity” (107) should be defined more clearly.
- Wording here, “roles of non-native species”, implies that multiple species will be examined separately.
- “Environmental matrices” and “urbanization matrix” need to be explicitly defined.
Methods
Study Area
The maps and description are very helpful to a reader unfamiliar with the area. Please clarify “offers a continuous forested strip associated with gravel floodplain or cultivated areas” (138).
Field Sampling
The method for site selection is not described beyond that it should “fully cover” the basin area of the studied river. Details should be included, as this may influence outcomes. Related questions include: How do disturbance regimes and width of the channel and floodplain vary along the length of the river, and how was this accounted for in sampling? If the sampling was done for other purposes and this analysis is added because of the availability of functional and taxonomic data, the original study should be referenced.
Methods for sampling vegetation are not supported by citations to precedent studies. Description of these methods are imprecise (e.g. “one or two” transects/plots, lines 156-157) and need clarification, including 1) how the number of transects and plots relates to floodplain width and “accessibility”; 2) what methods were use to “measure” visually (this might be better termed estimation); and 3) how individuals were determined for clonal and multi-stemmed species in order to count the number of individuals in each 1m2 subplot. While the importance of the ground layer in forested systems is provided, no rationale for excluding trees is provided. This key element of competitive dynamics and local-scale biotic and abiotic conditions of forests is missing.
Urban, Suburban and Rural Classifications
Methods for classification of sites based on degree of urbanization are based on only one measure: NDVI. This is neither a widely used measure of urbanization intensity (on its own), nor is a persuasive argument made that this measure alone is appropriate to this task. Other landscape features that may indicate degree of urbanization are included as predictive variables in the analysis, but are not included in the classification and do not appear to have been used for grouping sites for analysis. The definition of urbanization intensity is a subject of much discussion as pertains to studies in urban ecology that compare along this gradient (see McDonnell, M.J.; Hahs, A.K. in Landscape Ecol 2008, 23, 1143–1155, doi:10.1007/s10980-008-9253-4, and responses to this paper).
Functional Traits
The rationale for inclusion of each of the functional traits should be provided, including the direct relevance of each one to the study objectives and hypotheses.
LULC
Rationale for differing buffer widths should be provided; here 200 m is used for LULC, while NDVI was performed at 500 m. Methods for mapping the three land classes used are not given; it is unclear whether this was done by hand or using an existing classification (or some other method). Methods for soil texture are not described, and rationale for inclusion is not provided.
Distance to stream is not necessarily a good measure of flood disturbance in urban environments. As mentioned in the Discussion, many urban streams have hardened banks, and are often incised due to high volume and high velocity flows from impermeable surfaces, preventing flood waters from leaving the channel except under extreme flows. (These flows are also not mentioned in relationship to use of impermeable surface as a predictor.) See work on “urban stream syndrome” addressing these hydrologic effects.
Statistical Analyses
Precedent works using similar analyses for similar purposes and/or documentation showing appropriateness of methods for the data are provided in most instances but not all and should be provided throughout. Note that Bray-Curtis is a measure of dissimilarity rather than distance.
Values of predictor variables introduced in section 2.6.3 should be shown in a table. Rationale for selecting these variables should be included.
Results
The presentation of results is generally clear and the diagrams appropriate. A few items could use more information or clarification.
Urbanization levels presented in Table 4 (324) do not appear to address all three classes.
It is not clear how the homogenization effect (331) is quantified and compared; is this a comparison among sites within an urbanization class? If so, where are the data presented?
Not all traits listed in the methods appear to be reported on in the results, except as part of the functional diversity analyses. Were there differences among trait states between sites with differing levels of urbanization?
It should be made clear what comprises the “total flora” (339) – is this the complete species list of all sampled plots, or does it include additional species in the regional flora from other sources?
The way that results related to impervious surface and other variables (344) are presented suggest direct effects on taxonomic beta diversity, while many of these effects would be indirect. Data for all predictor variables (e.g. range, mean, median, S.D.) should be presented in a table for the reader to interpret.
The graphs included in Figure 6 should be interpreted in the text.
Discussion
It is not possible to evaluate the strength of statements related to urbanization intensity based on the use of a single measure of vegetation as proxy for “urban”. Going along with the use of NDVI alone for this purpose, there remain some questions pertaining to the interpretation of these results.
The relationship between non-native species and urban conditions is not clearly driven by the species as stated (391). Urbanization is a complex process that unfolds over time, with both changes to the conditions for plant life and the introduction and extirpation of species. Urbanization is treated as monolithic in this section (389-402) and its effects are simplified to “disturbances” (398). The contradictory findings here may be an artifact of classification, or (more interestingly perhaps) could be an indicator or extinction debt in more recently transformed ecosystems described in lines 412-414.
The definition of non-native plants in lines 408-409 is not comprehensive (it excludes invasive woody plants, for example); the circular logic here supports assertions by defining the problem in terms of the conclusion drawn.
Lines 416-417 should be revised for clarity.
Predictor variable results are discussed, but as they are not presented quantitatively, the reader has no basis on which to evaluate.
Statements related to connectivity should be very clear regarding which species might benefit and how.
Thinking about “disturbance” would benefit from being drawn together into one section and what is meant by this clearly defined. Stream systems are subject to frequent ecological disturbance, and relationships between riparian disturbance regimes and urban disturbance regimes would enrich the interpretation of results.
To state that this study can draw conclusions about the effects of flooding as a homogenizing factor (446) using only the proxy variable of distance to stream is a stretch, especially considering flood control and bank hardening that is common in urban environments.
The discussion of the influence of forest within 200 m is unclear. Forest is equated with heterogeneity (including grasslands), though landscape heterogeneity is not treated elsewhere. Heterogeneity is also discussed as a positive in relation to implications for conservation. Engaging with the literature on urban landscape heterogeneity and considering its implications in this study would also enrich the interpretation.
Grammatical Errors
Number agreement, particularly lack of “s” where nouns should be plural
Overuse of “the” when referring to general concepts
Use of “besides” when “in addition” or similar term would be appropriate; “besides” has a meaning that implies contradiction and in these instances, none is implied.
Examples of instances where word choices could be examined for clarity (not exhaustive):
|
Line |
Word/phrase |
|
36 |
in an ecological process |
|
36 |
coexists with |
|
38 |
in ecology and evolution processes |
|
42 |
effective |
|
44 |
Use of “mentioned” throughout should be reviewed; “previously mentioned” may be more appropriate. |
|
61 |
refer to |
|
96 |
sustainable mechanisms |
|
120-123 |
Remove “in” to improve clarity |
|
278 |
Investigated |
|
278 |
investigated |
|
293 |
The riparian plant species, such as |
|
333 |
facilitated |
|
335 |
facet |
|
367 |
Unlike the consistent results were obtained |
|
444 |
Coarse textured floodplain soil always linked to frequent flush flooding events. |
|
484 |
refers to |
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Round 2
Reviewer 2 Report
This revision has resulted in greater specificity regarding methods, improvement to clarity of the writing, and greater detail in the results, all of which improve the reader’s ability to assess the work and the meaning of its outcomes. However, some important shortcomings remain.
The most important of these is that the sampling design does not explicitly include variables related to urbanization in selection of sites, though these are a central focus of the analysis. It is still unclear what the basis for selection of individual field sampling sites was. A list of remotely sensed criteria are now provided and a much clearer description of methods used is now provided, which is helpful, but the total number of sites that would meet these criteria is not provided and how the specific sites were selected from among all potential sites is not described. Variables related to urbanization appear to be applied only post-hoc. If this study utilizes sites from another work by these authors and the rationale for site selection for that study applies, this does not make the findings invalid, but it should be clearly stated, the other work cited, and conclusions adjusted accordingly.
A related problem is that only the more urbanized portion of the watershed of the studied river catchment is sampled. The map showing location of sampling sites appears to have no sampling points beyond where major roads leave the river floodplain. Even the most rural category for sampling has at least 5% impervious surface within 500 m2 of the sampling point. This could explain why some trends in “rural” conditions are more similar to “suburban” patterns seen in some other studies. A direct comparison using the continuous variables related to urbanization intensity might give a clearer picture than these categories. The definition of what is “urban” or “rural” certainly varies from place to place. A quick look at aerial images of the study area shows nearly continuous use of the floodplain along the entire channel in the sampled zone, which appears to become more agricultural in nature with distance from the largest city. Agricultural effects on plant composition are not addressed.
The decision not to include trees in the vegetation analysis due to their “artificial character” needs more support. In human-ecological systems, the action of people upon the landscape is part of the system. Excluding a key type of plants (e.g. trees in a forest) due to potential human influence on composition needs a clear justification. Management by people is included as a variable (though criteria for inclusion are not enumerated) and could be a predictor for differences in this forest stratum.
Additional variables still need explanation. First, soil texture does not include a key biologically active texture class typically found in floodplains: silt. Second, the relationship between sampled plots and the regional floodplain flora beta diversity is not clear.
Organization of the manuscript could also still be improved. Explanations for the inclusion of specific functional traits have now been added, which adds to clarity. However, these are hypotheses and should be stated as such, as well as backed with citations to relevant literature and included in the introduction. In some cases, results are reported in the Methods section.
Methods for vegetation sampling are now much clearer, though the method for density using number of individuals for shrubs is subjective – the unit being “obvious clump” per square meter. While this might be obvious for small plants, this would yield little meaningful information for large stands of tall shrubs. Since this is not included in the analysis, it should be removed.
The Results are now clearer with additional explanations and addition of a table with predictor values. This table could be improved by grouping the predictors by type, explaining units and methods for each in brief in a caption, and removing “Natural riverbank”, which is the inverse of the previous variable, “Artificial management”.
Concepts related to disturbance and invasion have been integrated, though they need to cite the literature on which statements are based.
English language usage is clearer though some word choices are still ambiguous.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
