To increase transparency, PeerJ operates a system of 'optional signed reviews and history'. This takes two forms: (1) peer reviewers are encouraged, but not required, to provide their names (if they do so, then their profile page records the articles they have reviewed), and (2) authors are given the option of reproducing their entire peer review history alongside their published article (in which case the complete peer review process is provided, including revisions, rebuttal letters and editor decision letters).
Thanks so much for your excellent rebuttal letter documenting your changes and responses to questions from reviewers or from me. I think you addressed all issues, and I think your revised mss is improved with the changes. I especially like the detailed presentation of your methods and how you presented your results. I'm pleased you chose PeerJ as an outlet for your work, and I think our ability to include your data and other supplemental information is especially valuable in a study such as yours.
I must also comment that getting such a well-written revision back makes my job as editor much easier! Thanks. -Leon
I concur with the reviewer recommendations regarding the suitability of your manuscript for publication. I think it makes a worthwhile and needed contribution to the discussion and controversy over the use of neonicotinoid insecticides. Most of my comments pertain to matters of style. PeerJ tends to be very liberal in our stylist requirements, but I strongly believe style can aid in the readability of a manuscript. I think readability is also the answer to the question you posed regarding inclusion of figures and tables directly in the manuscript versus including these as supplemental materials. My advice is to include the figures and tables in the manuscript if it helps you make a convincing argument or if they contain information you think essential to communicate to your readers. Otherwise, include information within supplemental materials. I found your division of materials between manuscript and supplement mostly appropriate, although I did have one question (regarding functional group classifications) I didn’t see answered in either place.
I have only a few “content”-based questions/observations:
First, as I mentioned, you discuss functional group choices in the M&M (lines 172-183), but this section left me with a couple questions: (1) what taxonomic level was used to make functional distinctions, family for insects and class for non-insects? (2) what were the functional choices, specifically I wanted to see a table of taxa and how you classified them, and (3) I admit this is rather anal and probably unnecessary, but given that I might disagree with some of your classifications (for instance I tend to regard Gryllidae as predators, at least in soybean based on research my advisor Larry Pedigo did years ago, and I similarly regard carabids as predators in crop fields, based on my own research experience), did you happen to do your analysis with changes in your functional designations, and if so did it substantially change your results? This last question isn’t a point for revision necessary, mostly my own curiosity.
Second, I very much like your rational (i.e., parasitoid>predator>omnivore) in testing for the influence of prey scarcity. Obviously it suffers from the potential confounding factor of these functional groups typically representing different taxa with potentially different susceptibilities, but it is, nevertheless, a valuable method in my opinion. Of course, I’d much rather see an analysis with the influence of prey density explicitly removed but, as you point out, in the absence of detailed natural enemy-prey density relationships, direct examination isn’t possible. On the other hand, I am confused about direct toxicity, and I think this is a point you must discuss. In particular, I’m not sure you actually mean direct toxicity. If the insecticide is systemically distributed in the plant, what is the mechanism of intoxication for a natural enemy? It is not direct exposure as occurs with sprayed insecticides, right? Doesn’t exposure to the toxin come from ingesting poisoned prey? In fact, I thought the argument regarding reduced impact of seed treatments generally, came from the premise that natural enemies were not exposed to the insecticide. Are neonicotinoids somehow different than other seed treatments or other systemic insecticides? So is this direct toxicity, and if so how are the natural enemies exposed? If it is indirect toxicity (through prey), then wouldn’t reduced prey abundance led to a corresponding reduction in toxicity?
Please review the “Works cited” (Please change this to “Literature Cited”). I started noting corrections but ultimately I found too many points, so I’d like to offer these three issues that need to be addressed for stylistic consistency: (1) all article titles should be listed with sentence cap format (only first word capitalized unless a proper noun), (2) all journal titles need to be consistent – either completely written out or all abbreviated (you currently have a mixture of these), and (3) scientific names in article titles must be italicized (for example, line 702 includes “Aphis glycines” rather than “Aphid glycines”) [And as a total aside, I was unfamiliar with this reference but when I saw the term “dynamic action threshold” in the title I wanted reach through time and space to slap the authors, reviewers, and editors who let that worthless term through. I thought Larry, Scott, and I put a stake through the heart of that meaningless phrase back in the ‘80s when we published our annual review article on economic injury levels and when Larry and I did our economic threshold book, but I see I was wrong. More proof that stupidity never dies and that most of my professional career was a waste of time. But I digress.]
I think you would be better served to limit italics to scientific names and either forgo emphasizing numbering and terms (“direct toxicity” and “prey scarcity”) with italics, or use bold on the terms.
In the M&M, I suggest changing your numbered sections as follows:
Use a subheading after your introductory paragraph like “Analysis Procedure”. Next number your list “1. Searching…”, “2. Defining…) etc.
Regarding your “in text” numbered lists, please use a (1), (2), (3), etc. format rather than italicized (i), (ii), etc.
What is my rationale on all of this? Well, readers are most used to numeric schemes so there is an ease and familiarity with this convention. Using a “1.” Style for your procedure headings conveys a simple message that your procedures for the meta-analysis occurred in an orderly, step-by-step fashion. Finally, choosing “(1)” over “(i)” is mostly subjective, but I find the (i) more characteristic of an outline organizational style or (God-forbid) legal documentation.
Finally, feel free to email, text, or call me (***-***, my cell and only phone), if you have any questions.
This very clear and well-reasoned manuscript reports a meta-analysis examining the effects of neonicotinoid seed treatments on predators and parasitoids, relative to pesticide-free and pyrethroid-sprayed control treatments. The methods for study selection, data extraction, and statistical analysis are clearly presented.
The research question is well defined, and it is clear that the findings fill an identified knowledge gap. The methods for study selection, data extraction, and statistical analysis are rigorous and clearly and thoroughly presented.
The methods and the data are robust. This is a thorough and clear study.
I have only minor comments:
Line 219: introduce sigma squared as the symbol for variance. I think the symbol first appears in line 345, but I don’t see where it was defined as the variance.
Line 237: there is a missing equals sign (=) in the equation for I-squared, and you need to define df as degrees of freedom (and either provide the equation for df or a reference where it can be looked up). The X in the equation, which I assume is the multiplication sign is confusing… why not simplify and just report I-squared as a proportion throughout?
Line 276: add the word collinearity here: “correlations among moderator variables (collinearity) can render…”
Lines 466-467: I would phrase a bit more neutrally: “seed-applied neonicotinoids i) reduced natural enemy abundance and ii) reduced natural enemies to about the same extent as did foliar or soil-applied pyrethroids.” The “safer” language is a bit jarring.
Line 472: Again, remain neutral in the interpretation: “Seed-applied neonicotinoids reduced the abundance of natural enemies…” rather than “negatively affected”. I realize this is a subtle distinction but do avoid anything that might seem like a value judgment here.
Figure 1 is terrific, but please explain “PRISMA” either here or in the main text and add the reference.
Figures 2 – please add the number of comparisons (n) for the subsets of studies that reported abundance of insects and non-insects.
Figure 3 – please indicate in the figure of the caption, the reduction in number of observations when the Ohnesorg et al paper is removed from the analysis.
I do not have experience w/ meta analysis, and found the methods difficult to follow because of that lack. Thus I had little expertise to judge the exp design, its validity or interpretation. (i.e. I am providing a really lame review!) Hope there is another reviewer with expertise in meta analysis.
Overall this is an interesting paper that adds to the growing evidence that NSTs have broader impacts than just on the target pest, and for longer than originally ‘sold’.
A stated goal of the paper was to test if neonicos are “more or less harmful to natural enemies than other insecticides”. A conclusion on Line 370 is that “seed applied neonicotinoids were not significantly safer for NEs than pyrethroids” and again on line 467, NSTs “were not generally safer for natural enemies than …pyrethroids” and there is ’harm’ on line 534. Does the meta analysis really measure ‘harm’, ‘safety’, toxicity, or actual exposure in a pesticide context? I would drop those words and rephrase goals and conclusions to something like - NSTs had a similar or greater impact on predator abundance than pyrethroids (ie just like the title). The words ‘safe’ or ‘safety’ are not the best choices anyway, as what is safe? In a pesticide context, we generatlly prefer words like ‘Risk’ vs safety.
I had some trouble working out how sampling time was considered in the analysis. Studies on NSTs would naturally be sampling early, and it makes sense that a prey deficit would mean a lack of NEs early too. This would be similar to a lack of NEs in a Bt crop for the targeted pest, or lack of NEs in a field of an excellent host plant resistant hybrid. The NST wears off over time, and what follows would be prey specific, correct? If a species normally comes into a system early and has one generation, but it’s wiped by the NST, then its NEs are gone too. But for something like soy aphid, the pest can build back and NE can become very abundant later. Soy aphids are also parthenogenic and overwhelm the system with abundance. In contrast, a pyrethroid spray tends to be applied later in the season, to an established NE population which likely had an impact up to that point. I simply didn’t understand how or which sampling times were used from various studies, and how these were compared to make the judgements about abundance. I think I missed something in the model that controls for this time effect.
To avoid confusion, stick to one terminology throughout instead of various terms including neonicotinoid seed treatment, neonic seed coatings (line 428), etc.
More specific comments
Line 57. Don’t’ need the words ‘of course’
Line 358 “as predicted under direct tox, the negative effect of neonic was stronger for insects than the other taxa” and in your paper summary, you claim to “Confirm lab work showing NSTs are less toxic to spiders and mites”. This seems a bit strong. This might be a dumb comment but bear with me here. Yes, insects are more susceptible to neonics in the lab, but does this data showing a stronger effect on insects really reflecting a difference in ‘direct toxicity’, or is it also reflecting a field colonization difference between the two groups of organisms? I don’t actually know this, I’m just thinking out loud. Mobile predators like ladybirds and such overwinter as adults, and move (fly) into fields early in the season. If they encounter prey scarcity early, within the 30-40 day window of neonics, they might move on or starve. Pred. mites overwinter on field edges (not sure?) & spiders as egg sacs mixed in residue or on field margins (?). Since they can’t fly, do they lag in field colonization? They would move into the field eventually, but after the NST is running out and prey might be coming back. In this data set, is spider abundance less impacted by NSTs simply because they aren’t mobile enough to get into fields in sufficient numbers early in the season to experience the prey deficit? I realize my explanation might not make any sense.
Line 317: I had assumed the soil applied pyrethroid was Capture LFR, surprised to read it was granular pyrethroid.
Line 385 spelling of neonic
LIne 547-8 You say its ‘Worth noting that pyrethroids have lower acute toxicity than neonics.’ As written, this sentence seems to be adding on to the previous sentence, giving yet another advantage of using a foliar pyrethroid instead of a neonics. By lower acute tox, did you mean they are less toxic or that they have a LOWER LD50, which actually = MORE toxic. The latter, not the former, is true. The oral LD50 of formulated Warrior II w/ Zeon Tech (300 mg/kg -ish) is much lower than IMI or cruiser (closer to 5000 mg/kg). 4th gen pyrethroids not only have high toxicity, they also tend to cause skin rashes and other handler-related impacts.
Tables/Figs: Instead of the terms ‘insect’ and ‘non-insect’ (which could mean anything like birds, amphibs, etc), suggest the terms ‘Other arthropods’ or ‘non-insect arthropods’ for clarification?
Table 1 Table heading was cut off on my version after the word arachnid
Table 2: Suggest spelling out the insecticide names (the other lines are spelled out)
• Don’t know the specific citation requirements for this journal, but may need to replace long journal names w/ the approved acronyms
• Some sci names were italicized but others not
• Line 677, misspelling of ‘commission’
Legend to Figure 1. Please define the acronym PRISMA.
The meta-analyses were conducted very carefully; the authors should be commended for doing such a thoughtful and rigorous analysis.
Figure 3. It might be interesting to address the possibility that foliar sprays versus applications of granular materials to the soil might have different effects on natural enemies (or, at least, particular groups of natural enemies).
I agree with the authors that it is tricky to interpret the suppression of natural enemy populations following neonicotinoid seed treatments. If the suppression of natural enemies is due to prey (or host) scarcity, then this may not really be a problem for biological pest control. If, however, the seed treatments cause direct toxicity to the natural enemies, then we could have a real problem. The authors note that direct toxicity and prey scarcity could both be contributing. The discussion of predator-prey ratios is a useful step, but there just aren’t many studies the provide data to evaluate predator-prey ratios. This is, I think, a key interpretational challenge, but I think the authors do a good job of discussing this in a forthright manner with the readers.
Lines 101-102. Under the prey scarcity hypothesis, it is still possible that arachnid populations might be somewhat less strongly suppressed by seed treatments, if arachnids can survive or persist during periods of prey scarcity better than insect predators can. Some spiders can survive surprisingly long periods of prey absence or scarcity, even if their growth rates are reduced; this may be especially true for web-building spiders. Predatory insects can do this too, but I think less effectively. I’m not sure where phytoseiids fit in here (perhaps more similar to predatory insects in their need for prey, although many phytoseiids are omnivorous).
The magnitude of the mean effect size reported here (d = -0.3, or a 16% drop in natural enemy abundance) is quite small. This may explain why previous studies arrived at mixed results. If some of this reduction is due to prey scarcity, then it might not be unreasonable to suggest that seed treatments are unlikely to seriously disrupt biological pest control. However, the authors do point out from their earlier work that, at least in one case, this effect on natural enemy suppression was sufficient to disrupt biocontrol and result in yield loss. It may be that early-season effects are important, even if they are diluted by analyzing season-long mean effects (as they do here). The authors might address, earlier in the paper, why they chose to address season-long average effects, when seed treatments are most likely to exert their effects early.
All text and materials provided via this peer-review history page are made available under a Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.