Effects of Wildfire and the Presence of the Invasive Paulownia tomentosa on the Regeneration of Native Tree Species in North-Central Appalachia
Reviewer 1 Report
Review of “Effects of wildfire and the presence of the invasive Paulownia tomentosa on the regeneration of native tree species in north central Appalachia”
In this study the authors analyse a rather old dataset regarding the effects of an invasive species and fire on the regeneration density and initial growth of native species. The dataset is interesting, although I have some concerns about the measurements of height in only three individuals per plot and the counting of branches below 137cm as different individuals. The statistical analysis applied needs to be better described and probably changed to a glm with poisson error for the regeneration density and to a mixed effect models using all height measurements. For the above reasons I believe that this manuscript needs to be substantially revised, before we can appropriately evaluate the findings and discussion
L36”biogeochemical processe” please be more specific for example productivity? Nutrient recycling?
L 45” botanical and silvical characteristics commonly associated” … such as?
L47 “in areas”
L51 “Paulownia was found to invade two different areas that experienced fire, and 51
the native plant communities changed as a result .” Please explain a bit more.
L102 “Paulownia stems that split below 137 cm height were counted as multiple stems.” Is this a valid assumption? Usually these are considered as a single individual. Please explain?
L105. The native species are five.
L110 Just three height seems to be rather low
L113 -L115: Please be more specific on the use of ANOVA. Which were the response variables and which were the factors. When comparing count data (regeneration) maybe it is better to use GLM with a poisson error distribution, as probably the assumptions of the simple ANOVA do not hold.
Also I think the proper way to do the height analysis is to use a mixed effect model.
See more details below
L118-L120. Move to description of stats in Methods?
L123-L126. Better fits to Discission
Section 3.2. Are you using a two way ANOVA here? If not I think it would for be more informative. The two factors would be Pauwlonia presence/absence and fire presence absence. You should also check for interactions.
Again, regeneration density should be assessed with a glm with a poisson error distribution
L211 please correct the name
I think it would be better to use all data (not just the mean height per plot) and run a mixed effect model with fixed effects the fire (presence/absence), the Pauwlonia (presence/absence) and use plot as a random effect.
I am not sure about the utility of results summary. You could probably fuse these findings in the previous results sections
L324: “seedling responses cannot be attributed to fire alone but rather must be analyzed together”. This is why I suggest using a two way anoa with interaction, id you have not done so
L328-342. Considering short (<137) stems as different individuals might have an impact on your findings. I would rerun the analysis with single individulas to be sure. Also use the appropriate statistical model.
L343: “Whether the number of paulownia stems will continue to increase to significantly more numbers can only be known by revisiting the study area some years later.” I totally agree. To be honest as this is a rather old dataset I expected that the authors would have visited the areas and provided these data.
The responses to your comments are contained in the uploaded file.
Author Response File: Author Response.pdf
Reviewer 2 Report
This study evaluated post-fire regeneration of an invasive tree species, paulownia, and assessed the degree to which paulownia presence was associated with the regeneration densities of three native hardwood species. The authors present evidence that paulownia presence was associated with lower numbers of regenerating native species in both burned and unburned areas. The study has an appealing, straightforward design, and for the most part, the material is presented clearly. My two main concerns are that the authors appear to attribute casualty to the presence of paulownia (more on this below), and that the stems ha-1 do not appear to meet the basic assumptions for ANOVA. Based on Figure 2, the count data modeled in the study are not normally distributed and the variance between groups is not homogeneous. In general, ANOVA are not appropriate/optimal for count data. I recommend the authors provide an explanation for why ANOVA is an appropriate approach in this case, or better yet, consider an alternative approach (e.g., negative binomial GLM). Stroup 2014 may be a good resource if the authors are willing to consider an approach more suited to zero-bound, non-normally distributed count data. The authors report approximately 15 p-values and my understanding is that many more statistical tests were performed but not reported. The authors may consider a Bonferroni correction to reduce the risk of making type 1 errors.
Stroup, W. W. (2015). Rethinking the analysis of non-normal data in plant and soil science. Agronomy Journal, 107(2), 811–827. https://doi.org/10.2134/agronj2013.0342
Many readers will not be familiar with paulownia biology or fire regimes in the author’s study area. I recommended that the authors include a bit more detail about paulownia biology in the introduction (there is some great stuff in the discussion that could be adapted to flesh out the intro). It would also be useful for the authors to generally characterize the fire regime in their study region, especially given the target journal (fire). Does fire frequently interact with paulownia populations in NC Appalachia? Is 1215 ha “large” fire by regional standards? Are fires human or lightning caused? What is the seasonality (this is touched on in the discussion) of fire? Was fire historically more/less common?
The discussion feels a little “light” to me – I’d suggest that the paper could be more impactful if the authors more deeply explore the implications of their work rather than re-report results, as the authors have done in several places in the discussion and almost all of the conclusion. The lowest hanging fruit here in my mind would be to describe the management implications – especially since the study occurred in a publicly managed forest. Are there implications here for logging? Prescribed fire? Fire prevention? Invasive control? Post-fire replanting? The fate of native hardwoods?
ADDITIONAL MINOR COMMENTS
Line 31 – I agree that invasive plants in some systems are the most numerous and pose the greatest concern, but I tend to think of the native species most harmed by those invasives as the most “important.”
Lines 37-40 – From my understanding of the (a) taxonomy of non-native plants (see below), invasive species by definition have a negative impact on their introduced environments. The authors certainly don’t need to adopt the taxonomy below, but the sentence would benefit for more clarity/precision.
From Richardson et al. 2000; USDA Natural Resource Conservation Service
Non-native, exotic, alien plants: Plants whose presence is due to intentional or accidental introduction as a result of human activity
Casual plants: Non-native plants that may flourish and reproduce occasionally in an area, but do not form self-replacing populations, and rely on repeated introductions for their persistence.
Naturalized plants: Non-native plants that reproduce consistently and sustain populations over time without direct intervention by humans
Invasive Plant: A non-native plant that grows quickly, and spreads to the point of disrupting plant communities or ecosystems likely to cause economic or environmental harm or harm to human health
Line 166: “The presence of paulownia in the burn areas (F-P) further reduced this number significantly” suggests that the number of regenerating white oak stems was significantly lower than the F-NP scenario. Although the last clause of this sentence is accurate (e.g., “when compared to the control”), this sentence is somewhat confusing/misleading. Please clarify that there was no difference in regenerating white oak stems between the F-NP and F-P plots.
Lines 301-320 – The summary figure (Figure 6) is very useful. However, this “results summary” appears to simply repeat many (all?) of the same results reported earlier in the results section. I would recommend consolidating the results, so each result is only reported once. Otherwise, the reader will be confused as to whether or not they are reading new results, a subset of important results, or simply a rehash of previously reported results. Key results can be revisited in the discussion.
Lines 322-327 – This strikes me as a strange way to start a discussion. I’m also unclear by what you mean by “seedling responses cannot be attributed to fire alone but must be analyzed together.” My understanding was that you were intending to attribute seedling responses to not just fire, but also the presence of paulownia? In what way were seedling responses analyzed together? Do you mean that you didn’t categorize burned plots as logged/not logged, or along a gradient of fire severity? Perhaps this would be more appropriate in the methods section? Or later in the discussion, in a paragraph describing study limitations? I find that the first paragraph of the discussion is most impactful when it cements the reader’s understanding of most important results.
Lines 118-126 and Lines 329-330 – The finding that paulownia trees, but not red maple and oaks, were taller in unburned areas than in burned areas did not strike me as a finding worth reporting until I got to lines 337-341, where I see that juveniles are capable of very rapid growth rates in favorable conditions. I’d recommend providing the reader with some of this great biological context (lines 337-353) in the introduction.
The study consistently uses language like “reduced” (e.g., abstract, line 179, line 387, line 384) “impacts” (line 382) to describe the association between paulownia presence and native tree seedling densities. Do the authors feel that this language is appropriate given that this was an observational study without pre-fire data in which treatments were not randomly assigned? Can the differences between treatments be definitely ascribed to the presence or absence of paulownia? In these types of studies, I generally see such relationships described as negative or positive “associations” or “relationships.” In my view, lines 424-428 provide a more tempered interpretation of the results, and I would recommend a similar framing throughout. If the authors insist on ascribing causality to paulownia presence or absence, at the bare minimum they should provide a compelling biological explanation for why the presence of paulownia (as few as a single individual) would result in significantly reduced native regeneration. Lines 388-404 very generally describe how invasives can out compete natives, but this text does not provide any information about a specific mechanism by which paulownia presence would reduce native regeneration.
Figure 2: It is not clear to me what the letters next to the boxplots mean, could you clarify? I see that “means with the same letters are not significantly different,” but I’m still confused. Same question for Figure 4.
Figure 6: This composite figure is impactful. I generally like to use colors in figures to differentiate between categories. The authors may prefer to optimize their figure appearance for greyscale, but I personally think these figures would be more easily intelligible if they were color-coded.
The response to your comments are contained in the uploaded file.
Author Response File: Author Response.pdf
Reviewer 1 Report
The authors have adequately addressed my comments, and I appreciate their effort for trying a slightly different statistical approach. The paper would be a good contribution to the field. Congrats!
Reviewer 2 Report
The authors have thoughtfully addressed all my comments. In particular, I appreciate the additional silvics background/interpretation, as well as the authors' willingness to consider an alternate statistical approach for the tree count data. I believe this paper is a good contribution in its present form. Well done!