**3. Results**

The 2018 beekeeping season was generally a weak active season for Romania in terms of honey production, (via oilseed rape and sunflower), this situation being registered in the monitored locations by means of the electronic hives. The weak beekeeping season had a negative impact on honey sample collection. The weight gain represented the first indicator of honeybee colony activity and, mainly, of the existence of a nectar flow from targeted fields that is necessary for honey sample collections. This evaluation offered preliminary information about the probability of collecting samples from the targeted fields, but this information needed to be correlated and confirmed by melissopalinological studies on collected samples. The weight gain in the monitored hives is shown in the following images (Figures 4 and 5):

**Figure 4.** The weight gain (kg) monitored by experimental electronic hives in the rape honey flow in two locations (Albota and Tandarei) from 20 April to 5 May 2018.

**Figure 5.** The weight gain (kg) monitored by experimental electronic hives in the sunflower honey flow in four locations from 29 June to 14 July 2018.

As can be noticed from the gain weight registered by the electronic hive, the general activity of honeybee colonies and the development of the colonies in honey flow conditions were relatively low, with the causes being not well understood.

Concerning the results on honeybees, it is important to mention that in general, in the Fundulea, Tandarei and Baneasa locations, a weak activity was noticed. Additionally noticed were signs of depopulations and/or honeybees in front of the hives with specific symptoms of acute toxicity (walking on the ground, paralysis, and dying) during the oilseed rape honey flow period. The results showed that imidacloprid was present in a concentration of 0.1 ng/bee, as well as under the limit of quantification (LOQ) in these locations.

The presented protocol that was used to collect and prepare honey and pollen samples for neonicotinoid residues analyses, included an important step: melissopalinological analyses for the identification the mono-floral honeys (Table 1) and mono-floral pollens selection. Following these preliminary analyses, all rape honey samples were framed in the standards of mono-floral honeys, and the results showed that these samples were set according to the standard internal laboratory references for rape honey (minimum 40% rape pollen grains). However, in the case of sunflower honeys, only one of the collected samples was framed in the internal standards of typical sunflower honey.

These results indicate that the analyzed honey samples in the case of multi-floral honeys (e.g., sunflower honey) could not correctly reflect the residues of neonicotinoids found in the envisaged flora, and this led to their sub-evaluation.

The results on the neonicotinoid residues analyzed on different matrices and laboratories are highlighted in Table 2.

The raw data show that 48% of the total samples (*n* = 50) sent to the two laboratories contained one or more, detected or quantified, neonicotinoid residues, with the quantifiable residues being found in 38% of samples.

It can be emphasized that, in the case of the EU reference laboratory (ANSES), 43.3% of the total analyzed samples (*n* = 30) contained registered, quantifiable residues of one or more neonicotinoids, and 33.3% contained registered, detectable amounts of one or more residues of neonicotinoids. In the QSI laboratory, 30% of the total analyzed samples (*n* = 20) contained registered, quantifiable residues.

One very important aspect needs to be mentioned regarding the analytical references of the two laboratories (Table 3) for the residue quantification limit (LOQ), which could influence the results and their interpratation; for example, in the honey analyses, the different residue LOQs (ANSES = 1.0–4.0 ng/g, QSI = 1.0–5.0 ng/g) need to be analyzed on different active substances between the two laboratories, while for the pollen analyses, the LOQs (ANSES = 0.5–1 ng/g, QSI = 10.0 ng/g) are more favorable in ANSES laboratory.

Taking into account the LOQ of the residue analysis in the involved laboratories, the following neonicotinoids were found in quantifiable levels: imidacloprid in honeybee samples, acetamiprid and thiacloprid in honey samples, and acetamiprid, imidacloprid and thiacloprid in pollen samples.

The minimum–maximum levels of the quantified neonicotinoids were:


Regarding the three neonicotinoids (imidacloprid, clothianidin and thiamethoxam) that are banned at European level but are the subject of derogation in Romania, the following percentages of residues were found on different matrices and crops in the two laboratories:


Out of all analyzed active substances, clothianidin was not found in any sample.

Concerning the monitored crops, the percentages of samples with one or more quantifiable neonicotinoid residues in different matrices are presented as it follows:


With only one exception found in oilseed rape pollen (795.1 ng/g thiacloprid), all residues were under the maximum residue limit (MRL) established for human consumption.

Looking to the percentages of samples with one or more neonicotinoid residues over the detection limit in different matrices and crops, the situation is presented as follows:


One can notice the presence of neonicotinoids in detectable amounts in three of the four sunflower honey samples analyzed in the ANSES laboratory, while in sunflower pollen from the same location (Albota, Tandarei, and Secuieni), neonicotinoids were not detected, even in the pollen samples that were mono-floral selected. An inverse situation was noticed in one oilseed rape sample analyzed in the same laboratory (ANSES), where neonicotinoid residues were not identified in honey (R-M-F-1) from the same location (Fundulea), but they were very well identified in a pollen sample (R-P-F-1). This result shows the importance of analyzing residues in both matrices of honey and pollen collected from the same hive.

Another aspect of residue analysis is related to the presence in detectable and/or quantifiable amounts of di fferent active substances of neonicotinoids in the same sample that could conduct the so-called "cocktail" e ffect. In this study, this was the case of oilseed rape honey (50%) and pollen samples (100%), as well as in corn samples (33.3%), though only in the samples analyzed in the ANSES laboratory.

Another important aspect that is worth mentioning is that in every sample preparation, the corn and sunflower pollens pellets were separated from the same multi-floral sample with di fferent levels of mixtures with other pollens (not analyzed). Out of the obtained results, one can highlight that the neonicotinoid residues found in corn pollens did not influence the sunflower pollen residues because all corn pollens registered di fferent levels of residues and sunflower pollen was free of residue from an analytical point of view. This fact indicates us that neonicotinoids from a pollen pellet do not contaminate other pollen pellets.

Taking into account the eight locations where the samples were collected from, the results showed that every location had at least one sample with detected or quantified neonicotinoids in honey or pollen, so the distribution of neonicotinoids in the environment was relatively large in the areas with intensive agriculture.
