Pest Rodents’ Responses to Rice Farming in Northern Peninsular Malaysia

Round 1

Reviewer 1 Report

This manuscript is meant to aid crop production and rodent pest management through the use of models with a limited number of variables. However, we have no idea of the range of these values across the study area and their relation to rodent numbers. There is a total lack of data on rodent food habits and the type and amount of damage occurring to crops. Also, the units are not given in tables: are rodent "means" are the number captured per trap? And are bund height, width, and distances to bushes, water, and residential areas in meters, and again, what is the range in those values and how are distributed across the study area. How were rodent traps arranged (grids, traps lines; and what was the spacing between traps? In line 114, it should say "live traps" if that is what was used. Also, tables are not numbered in the order of their reference in the text. For example, table 2 isn't mentioned in the text until long after tables 3 and 4 are mentioned in the text. In table 3, what is "Example morphology" referring to? In line 156, it refers to 9 variables in table 2, but there are only 6 items listed in the table and these are mainly the rodent species list. In line 168, add words to read: ..."were used to complete"...  In Figure 2, it is clear that bandicoots are, by far, the major rodent species in the study area, even though the text often refers to damage by multiple species. Also, the title of figure 2 should give the months for the dry and wet seasons as is done in the title of figure 3. Ir seems that tables 5-7, should say the models are determining variables influence to relative rodent abundance? The titles of table 5 and 6 should read: ..and highest adjusted R2."  Line 232: it seems that figures 2 and 3 suggest that overall, the numbers of rodents is similar in both seasons, and perhaps a little higher in the wet season due to bandicoot numbers. Line 235, use "were" not wear. Lines 236-237: it's common to have multiple rodent species on most areas around the world...and, again, your study area is very dominated by bandicoots, with much lower numbers of the other species. Line 248-249, again, it's a shame that rodent damage wasn't assessed in this study. Lines 274-276, figure 2 doesn't seem to support this statement about greater rodent abundance in the dry season. Line 291, use "sufficient", not suffice? Lines 292-299, it would have been useful to show rodent numbers in relation to distances to residential  areas, just as it should show rodent numbers relative to heights and widths of bunds, i.e., not just model result values. I agree with your statement in lines 310-313, but unfortunately, your study doesn't contribute much to this needed info. Line 320, eliminate "Jacob, 2008"  Line 333, should say "burrows built under permanent..." Line 243, Use "eco-friendly" rather than green? Line 354, change to read: "...for rodents to nest." Line 365: use "asset" instead of assist? Lines 377-379, don't use capitol letters for common names of species. Lines 398-403, this statement is very true, but I feel that your manuscript doesn't contribute much to this need. I also note 91 citations is a lot for a scientific manuscript; although it would be fine for a review article. Some sentences cite 4-6 references for support of the statement which seems rather excessive. 

Author Response

Point 1: This manuscript is meant to aid crop production and rodent pest management through the use of models with a limited number of variables. However, we have no idea of the range of these values across the study area and their relation to rodent numbers. There is a total lack of data on rodent food habits and the type and amount of damage occurring to crops. Also, the units are not given in tables: are rodent "means" are the number captured per trap? And are bund height, width, and distances to bushes, water, and residential areas in meters, and again, what is the range in those values and how are distributed across the study area.

Response 1: Due to lack of resources and time, rodent food habits, type and amount damage was not counted as variables in GLMMS. “There is a total lack of data on rodent food habits and the type and amount of damage occurring to crops” has been inserted to limitation section. The unit for the parameters has been updated and the range has been inserted in table 3.

Point 2: How were rodent traps arranged (grids, traps lines; and what was the spacing between traps?

Response 2: Rodent traps was arranged randomly in line with min distance 1m from each other.

Point 3: In line 114, it should say "live traps" if that is what was used.

Response 3: live traps was inserted in line 125

Point 4: Also, tables are not numbered in the order of their reference in the text. For example, table 2 isn't mentioned in the text until long after tables 3 and 4 are mentioned in the text. In table 3, what is "Example morphology" referring to?

Response 4: Tables has been corrected and numbered in order. Example morphology is the graphical explanation explaining the field condition comparing situation throughout rice growth cycle. Since it caused confusion and taking space, the row has been removed.

Point 4: In line 156, it refers to 9 variables in table 2, but there are only 6 items listed in the table and these are mainly the rodent species list.

Response 4: Table 3 was only mentioned significant variables that accounted parsimoniously using GLMMs. More information on remaining additional variables was inserted in table 3.

Point 5: In line 168, add words to read: ..."were used to complete"... 

Response 5: Line 168 has been edited to “GenStat 12th Edition (VSN International, Hemel Hempstead, UK) were used to complete our analysis”

Point 6: In Figure 2, it is clear that bandicoots are, by far, the major rodent species in the study area, even though the text often refers to damage by multiple species.

Response 6: Bandicoot rats are omnivore and was majorly captured when used protein-based baits. We assume the damage was done by multiple species as the others were also identified pests in agrosystem.

Point 7: Also, the title of figure 2 should give the months for the dry and wet seasons as is done in the title of figure 3.

Response 7:  Month for wet and dry season for figure 2 has been inserted.

“Figure 2. Rodent species composition throughout dry (May-Sept) and wet rice planting season (Oct-Feb) in Jitra, Kedah. Protein based was used in trapping”

Point 8: It seems that tables 5-7, should say the models are determining variables influence to relative rodent abundance? The titles of table 5 and 6 should read:  ...and highest adjusted R2." 

Response 8: Table titles has been updated as follows.

Table 5. Best subsets from candidate models. The most parsimonious model (labeled with* and bolded) has six explanatory stand-level variables with a lowest Mallow Cp and highest adjusted R² influence to relative rodent abundance.

Table 6. Best subsets from candidate models using dredging method. The most parsimonious model (labeled with * and bolded) has two landscape-level explanatory variables with a lowest Mallow Cp and highest adjusted R2 influence to relative rodent abundance.

Table 7. Slopes and back transformed means for explanatory stand- and landscape-level variables determining variables influence to relative rodent abundance.

Point 9: Line 232: it seems that figures 2 and 3 suggest that overall, the numbers of rodents are similar in both seasons, and perhaps a little higher in the wet season due to bandicoot numbers.

Response 9: Figure 2 and 3 has been replaced with a better graphical aid on rodent abundance.

Point 10: Line 235, use "were" not wear.

Response 10:  The word has been replaced. Sentence 235 à 253

Point 11: Lines 236-237: it's common to have multiple rodent species on most areas around the world...and, again, your study area is very dominated by bandicoots, with much lower numbers of the other species.

Response 11:  The sentence (253-255) has been edited into “The main finding to emerge from the analysis is that multiple rodent species were trapped from the Jitra rice fields, although it seems to be dominated by the greater bandicoot rats, with much lower numbers of the other species.”

Point 12: Line 248-249, again, it's a shame that rodent damage wasn't assessed in this study.

Response 12: This study was conducted without grant support. With substantial financial and human resources, more detailed future study can improve this modelling.

Point 13: Lines 274-276, figure 2 doesn't seem to support this statement about greater rodent abundance in the dry season.

Response 13: Figure 2 has been replaced with monthly rodent abundance over time which gives a much better support for the statement.

Point 14: Line 291, use "sufficient", not suffice?

Response 14: Suffice has been changed into sufficient. (Line 291à310)

Point 15: Lines 292-299, it would have been useful to show rodent numbers in relation to distances to residential areas, just as it should show rodent numbers relative to heights and widths of bunds, i.e., not just model result values. I agree with your statement in lines 310-313, but unfortunately, your study doesn't contribute much to this needed info.

Response 15: Please provide your response for Point 2. (in red)

Point 16: Line 320, eliminate "Jacob, 2008" 

Response 16: Jacob, 2008 removed. (Line 320à339).

Point17: Line 333, should say "burrows built under permanent..."

Response 17: “burrows built under permanent..." has been inserted. (Line 333-> 352).

Point 18: Line 243, Use "eco-friendly" rather than green?

Response 18: green has been changed to eco-friendly. (Line 243-> 362).

Point 19: Line 354, change to read: "...for rodents to nest."

Response 19: "...for rodents to nest." Has been changed. Line 354 à373)

Point 20: Line 365: use "asset" instead of assist?

Response 20: Assist has been changed to asset (Line 365 à 384)

Point 21: Lines 377-379, don't use capital letters for common names of species.

Response 21: capital letters has been removed for common names of species (line 397-400).

Point 22: Lines 398-403, this statement is very true, but I feel that your manuscript doesn't contribute much to this need.

Response 22: Sentence has been removed.

Point 23: I also note 91 citations is a lot for a scientific manuscript; although it would be fine for a review article. Some sentences cite 4-6 references for support of the statement which seems rather excessive.

Response 23: sentences with 4-6 references for support of the statement has been edited. Max 4 references

Author Response File: Author Response.docx

Reviewer 2 Report

Re: Pest rodent’s responses to rice farming in Northern Peninsular Malaysia.

This is a wonderful research article; the authors have significantly contributed to the understanding of changes in rodents distribution in terms of space and time in Malaysia. It is also impressive that the authors have made comparisons with other Asian countries than other distant geographical locations. The authors have carefully designed the study and logically presented their findings.

However, there are a few major issues that could potentially flaw the study. The change is bait was clearly explained in the method section but it wasn’t listed at a potential limitation of the study. The authors should have found a way of making their baits consistent because the changes in rodent abundances could significantly be affected by the change in bait especially for the different species of rodents. Although the number and size of traps were clearly stated, the authors did not sufficiently explain the nearest distances between traps and why they chose that threshold. In addition, it was not clear in more traps were placed close to bunds  or in interior parts of the rice farm, what if most traps were kept close to bunds and more rats were trapped there? Is such was the case the model itself and the general conclusions may not be relied upon, thus I suggest that if the authors should add it to the limitations of the study if they had not clearly considered that during the project design but if they did, it should be clearly stated in the methods section. If accurate spacing of the traps was considered during the project design, the authors would have been able to estimate density which would have been a better predictor how rodents associate with different habitat characteristics. Results from the model table (Table 5) generally showed a weak R² value, which questions the validity of models tested and the best model chosen. The best model may have been over-parametrised. I suggest, adding interactive terms and use backward elimination process to determine the best model. The results of the best model needs to be clearly stated in a table so that the readers can clearly see which variables where more important in the study as this will clarify the points in the discussion section.

Other minor comments are:

Title: no need to change if authors don’t want to but I suggest the title should be changed as “Abundance of rodent pests in rice farms of Northern Malaysia”.

The ‘pests rodents’ through out the manuscript should be changed to ‘rodent pests’ because that’s the correct usage in English.

Lines 45-58: While the authors have clearly stated that rodent pests significantly affect rice production, no numbers where provided (e.g. up to 10% of rice loss…). Your audience may could simply conclude that this is an assumption since an idea of their effect is unknown. The rodent species known to exert the greatest impact should also be noted, and compared with the results at the discussion section.

Table 1: Looks messy, probably because of how it was saved as a PDF, the authors need to make tidy the table to make rge

Figure 1: it would be nice if there was an inset to show trapping grid, but not necessary.

Table 2: needs a clearer heading. Does the table represent number of traps rodents per night or per trap? If yes, then that should be included in the table. Need to also add location, both sites or just a single site. “relative abundance” needs to be taken off that table and can be included in the table title. Table 2, also seems like a table that should be included in the result section.

Table 3: Are those numbers in the table number of days? It should be clearly stated in the table title or column title.

Table 4: In the first column, the unit of measurement for each variable needs to be stated e.g. ‘Bund height (cm)’.

Lines 152-160: It would be nice if you could state whether your response variable was tested to be normally distributed through QQplots, shapiro-wilks or other tests.

Lines 162-169: It would be nice to include AIC too to see how it compares. The fact that Mallows’ Cp has been used a multitude of times doesn’t mean that it would be the best for your data set.

Figures 3 and 4: I suggest you take out the grid lines inside the plot and also those values close to the error bars, it makes the plot look messy.

Lines 234-249: From your result it appears the greater bandicot was the most abundant species, thus you need to say a bit whether it was also the dominant species in other studies in rice farms in Asia and whether there’s any evidence that shows that it significantly feeds on rice because not all correlation is causation. Then you can connect the dominant rodent species in other studies with their effects on rice.

The discussion section was unnecessarily long, the authors spent more time discussing other articles than discussing the major findings of their work and how it compares to other articles. The major findings should be the topic sentence of each paragraph in the discussion section.

Author Response

Point 1: The change is bait was clearly explained in the method section but it wasn’t listed at a potential limitation of the study. The authors should have found a way of making their baits consistent because the changes in rodent abundances could significantly be affected by the change in bait, especially for the different species of rodents.

Response 1: inconsistent bait choice has been mentioned in the limitation

Point 2: Although the number and size of traps were clearly stated, the authors did not sufficiently explain the nearest distances between traps and why they chose that threshold. In addition, it was not clear in more traps were placed close to bunds or in interior parts of the rice farm, what if most traps were kept close to bunds and more rats were trapped there? It is such the case the model itself and the general conclusions may not be relied upon, thus I suggest that if the authors should add it to the limitations of the study if they had not clearly considered that during the project design but if they did, it should be clearly stated in the methods section. If accurate spacing of the traps was considered during the project design, the authors would have been able to estimate density which would have been a better predictor of how rodents associate with different habitat characteristics.

Response 2: Further details on the trapping method have been mentioned under subchapter 2.2-rodent sampling. The following sentences have been added: The traps were installed alongside bunds or between the intersection of the bunds in the rice field with a minimum 5-meter distance from each other. 10 traps were installed over 3 locations in Ketol Village and 3 locations in Raja village. These were distributed so that each trapping location contained traps both with and without burrow entrances, and fairly distributed for an even spread.

Point 3: Results from the model table (Table 5) generally showed a weak R² value, which questions the validity of the models tested and the best model is chosen. The best model may have been over-parametrized. I suggest, adding interactive terms and using a backward elimination process to determine the best model. The results of the best model need to be clearly stated in a table so that the readers can clearly see which variables were more important in the study as this will clarify the points in the discussion section. 

Response 3: we have done the right thing, and reported the adjusted R², not the normal R² in order to deal with the overparameterizations. R-square is used to complement the model selection but it is not the main criterion for selecting the best model. The best model is only through Mallow cp with the lowest value. We disagree to use interactive terms because it will make the results more difficult to interpret in the end. We also disagree to backward elimination because it's a traditional approach but we used information-theoretic approach to pick the best model among other candidate models.

Point 4: Title: no need to change if the authors don’t want to but I suggest the title should be changed to “Abundance of rodent pests in rice farms of Northern Malaysia”.

Response 4: We disagree with the reviewer’s suggestion as the current title highlights the nature of the research with respect to habitat requirements of rodent pests in the rice field.

Point 5: The ‘pests’ rodents’ throughout the manuscript should be changed to ‘rodent pests’ because that’s the correct usage in English.

Response 5: ‘pest rodents’ has been changed to ‘rodent pests’ throughout the manuscript.

Point 6: Lines 45-58: While the authors have clearly stated that rodent pests significantly affect rice production, no numbers were provided (e.g. up to 10% of rice loss…). Your audience may could simply conclude that this is an assumption since an idea of their effect is unknown.

Response 6: Due to resource and time limitations, the model was unable to compute the damage % in the rice field during the study. I would encourage future studies to include damage assessment as a parameter for this modeling.

Point 7: The rodent species known to exert the greatest impact should also be noted, and compared with the results in the discussion section.

Response 7: Please provide your response for Point 2. (in red)

Point 8: Table 1: Looks messy, probably because of how it was saved as a PDF, the authors need to make tidy the table to make rge

Response 8: Table 1 has been tidied up.

Point 9: Figure 1: it would be nice if there was an inset to show the trapping grid, but not necessary.

Response 9: further details on trapping distances has been inserted in subchapter 2.2 rodent sampling.

Point 10: Table 2: needs a clearer heading. Does the table represent a number of trap rodents per night or per trap? If yes, then that should be included in the table. Need to also add location, both sites, or just a single site. “Relative abundance” needs to be taken off that table and can be included in the table title. Table 2, also seems like a table that should be included in the result section.

Response 10: Table 2 explains the rodent variable input to the GLMMs model. It has been changed to table 4 and placed under the statistical analysis subchapter.

Point 11: Table 3: Are those numbers in the table number of days? It should be clearly stated in the table title or column title.

Response 11: Unit ‘Days’ has been inserted in the 1^st column.

Point 12: Table 4: In the first column, the unit of measurement for each variable needs to be stated e.g. ‘Bund height (cm)’.

Response 12: Unit measurement has been inserted.

Point 13: Lines 152-160: It would be nice if you could state whether your response variable was tested to be normally distributed through QQplots, shapiro-wilks or other tests.

Response 13: GLMMs deal with non-normally distributed data and hence, there is no need to check the normality of the response variable because count data is always not normal and we have to use Poisson distribution in the modeling.

Point 14: Lines 162-169: It would be nice to include AIC too to see how it compares. The fact that Mallows’ Cp has been used a multitude of times doesn’t mean that it would be the best for your data set.

Response 14: Model selection based on AIC and Mallow cp would give similar results. We think using mallow cp to select the best model is already sufficient and reliable as much as AIC

Point 15: Figures 3 and 4: I suggest you take out the grid lines inside the plot and also those values close to the error bars, it makes the plot look messy.

Response 15: the grid lines inside the plot has been removed for both figures 3 and 4. The values close to the error bars has been adjusted and those which was too close has been also removed.

Point 16: Lines 234-249: From your result it appears the greater bandicoot was the most abundant species; thus, you need to say a bit whether it was also the dominant species in other studies in rice farms in Asia and whether there’s any evidence that shows that it significantly feeds on rice because not all correlation is causation. Then you can connect the dominant rodent species in other studies with their effects on rice.

Response 16: This flaw has been inserted in the limitation sub-topic. It is unclear whether the bandicoot rats are overtaking the rice field rat as the dominant species since protein-based bait in this study favors the bandicoot rat species. For a better species composition study, we would suggest using bait that effectively favors all species or using supporting observation using track plates or camera traps.

Point 17: The discussion section was unnecessarily long; the authors spent more time discussing other articles than discussing the major findings of their work and how it compares to other articles. The major findings should be the topic sentence of each paragraph in the discussion section

Response 17: The discussion section has been edited. Major findings have been inserted as topic sentences in each paragraph.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

It appears that this manuscript has been adequately revised for publication. 

Author Response

It appears that this manuscript has been adequately revised for publication. 

Response:  Thank you for your time and effort in evaluating this paper. we really appreciated it.