After the publication of the article, a number of concerns were raised about several of the methodological aspects of the study and about the conclusions drawn. In the light of the concerns raised, the PLOS ONE editors solicited further advice about this study from a member of the editorial board, independent from the Academic Editor who handled the submission. This editorial board member acknowledged that the study design has limitations and that these could have been more clearly outlined, however considered that the article’s contributions stand. PLOS ONE encourages post-publication evaluation of research and in line with this goal we are sharing the concerns raised and the evaluation received from our editorial board member.

The concerns raised to the attention of the editorial office were as follows:

-The field sizes for the field trials are too small (2ha) for 6 colonies to feed upon, raising the possibility that bees fed elsewhere.
-Lack of information about the surrounding area - the sites were not near arable crops raising the possibility of locations in flower-rich areas so that the bees did not fed on the treated crops.
- Time of exposure was short (a few days) compared to real situations (crops usually flower for one month).
- The treated and non-treated crops in the Multiple Exposure Study were separated by only 2 km. The foraging range of bees is up to 10 km, and thus bees located in the control field may have been foraging on the treated crop, and vice versa, confounding the results.
- The article does not report statistical analysis. Figure 6 appears to show large treatment effects in 2006 and 2008, and Figure 11 in 2007 and 2008, however no error bars or associated statistical analyses are included.
-The article does not adequately acknowledge the limitations of the study.

The evaluation by our editorial board member is outlined below:

This is a rather comprehensive study assessing the residue levels of thiametoxan and its primary metabolite (i.e., clothianidin) in pollen and nectar collected by honeybees in maize and oilseed rape grown from treated seeds determining subsequent exposure and residue persistence brought back to the colony. The contribution also aimed at investigating the potential detrimental effect of insecticide residues in pollen and nectar on the honeybee colonies. The concerns expressed basically refer to the second portion of the manuscript targeting the field experiment (i.e., field size with too many colonies, low exposure time and small between-field distances), lack of statistical analysis, and strong conclusions raised by the authors.

Small field size (2 ha): typical farming size varies greatly with region; I see no need of plots reflecting farm size since what is required on experimentation is decent sample sizes. Two hectares seem fine to me.
Too many colonies per field (6 colonies/2 ha): indeed it seems a relatively high colony density. As the different treatment plots are relatively isolated (woodland site with extensive agriculture), I imagine that the study plots were overexploited by the bees, which may have explored other areas to compensate for lack of enough resources potentially minimizing insecticide exposure.

Surrounding area: the critics are assuming that the surrounding areas are flower-rich, but the authors suggest otherwise (Multiple Exposure Study section).

Length of exposure: reported as 5 days according to the critics. This seems to be an oversimplification, although a mention of 5 days inspection is indeed made, at the exposure sites the exposure was between 19 and 23 days for adult insects. As the assessments were for adult insects of compounds with a rather strong acute effect on such stage, 5-day exposure would allow the detection of some responses. Nineteen to 23 days exposure is a very good and long exposure time.

Plot distance: the authors used a plot distance of 2 km and bees exhibit a foraging range of up to 10 km, which is accurate. However, the bulk of foraging takes place a much smaller distance than 10 km and as the plots are isolated among themselves by allegedly flower-poor woodplants, I believe the range of distance used (i.e., 2 km) is reasonable.

Lack of statistical analysis: the problem is not really the lack of statistical analysis, the limitation is the lack of descriptive statistics. The use of a measure of data dispersion would have been important and very welcome it is a fault of the manuscript. The overall trend exhibited is of a similar fluctuation between treated and untreated plots, but the concerns regarding the specific years of 2006 and part of 2008 from Fig. 6, and a few months of 2007 and 2008 from Fig. 11 are justified. However, the general trend suggests small differences between treatments.

Strong conclusions and limited data: the authors’ conclusions are consistent with their work. Too strong or too light is a question of perception and individual personality and style. However, as any scientific contribution, this particular one does have limitations and some are indicated above. The authors could have delimited the key concerns, but this does not invalidate or compromise the potential value of the contribution. The effort was comprehensive and seems honestly described, the relative importance of the findings is a matter of debate.