Foraging in maize field areas: A risky business?

Science of the Total Environment 601–602 (2017) 1522–1532

Contents lists available at ScienceDirect

Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv

Foraging in maize field areas: A risky business? Monique Boily ⁎, Philippe Aras, Catherine Jumarie Centre de Recherche en toxicologie de l'environnement (TOXEN), Département des sciences biologiques, Université du Québec à Montréal (UQAM), C.P. 8888, Succursale Centre-Ville, Montréal, QC H3C 3P8, Canada

H I G H L I G H T S

G R A P H I C A L

A B S T R A C T

• Maize pollen contains several metals that are potentially toxic to bees. • Consumption of maize pollen may contribute to higher metal contamination in bee. • Conventionally grown maize could be a challenge for the bee antioxidant system. • Retinoid acid metabolites could be used to monitor the oxidative damages in bee. • Discriminant analysis was relevant to associate bees to their maize field exposure.

a r t i c l e

i n f o

Article history: Received 2 April 2017 Received in revised form 2 June 2017 Accepted 2 June 2017 Available online xxxx Editor: Jay Gan Keywords: Apis mellifera Metals Carotenoids Retinoids Oxidative damages

a b s t r a c t In Quebec, Canada, the cultivation of maize dominates the agricultural territory. This crop requires a sustained supply of fertilizers from different sources: chemical, natural or from residual materials (sludge). These amendments contain metallic trace elements, which may lead to metal-contaminated maize pollen, a possible source of prooxidants for the foraging bees. Our objective was to determine whether maize fields environment influences the oxidation processes and the accumulation of metals in bees. A few days prior to pollen shedding, beehives were installed in maize fields: one organically grown (site A) and three conventionally grown (sites B, C and D). Soil, maize pollen and bees were analyzed for metal content. Every 15 days, bees were collected and analyzed for peroxidation of lipids, metallothionein-like proteins (MTLPs), proteins, retinoids and lipophilic antioxidants (carotenoids and α-tocopherol). The compound β-carotene was the most abundant in bees from all sites, followed by α-carotene, β-cryptoxanthin, α-cryptoxanthin, zeaxanthin and lutein. Retinaldehyde and retinol varied according to times and sites without demonstrating clear trends. However, significant differences between sites were noted in 13-cis-retinoic acid and two retinoic acid metabolites measured in bees, suggesting alteration in the reduction-oxidation processes. In line with these results, the level of lipid peroxidation was globally higher in sites B, C and D compared with the organic site. Higher concentrations of metals were observed in soil and pollen from the field A, but bees metal contents were equal or less than those measured in bees from other sites. Higher bee MTLP levels were measured in sites B, C and D. For most sampling times, the discriminant analysis revealed that the conditions were distinguished by the oxidation processes in bees. Our data suggest that bees foraging in conventionally grown maize fields are at risk of increased oxidative damages which can alter the fine regulation of retinoids. © 2017 Elsevier B.V. All rights reserved.

⁎ Corresponding author at: Département des sciences biologiques, C.P. 8888, Succ. Centre-Ville, Montréal, Québec H3C 3P8, Canada. E-mail address: [email protected] (M. Boily).

http://dx.doi.org/10.1016/j.scitotenv.2017.06.014 0048-9697/© 2017 Elsevier B.V. All rights reserved.

M. Boily et al. / Science of the Total Environment 601–602 (2017) 1522–1532

1. Introduction

2. Materials and methods

In North America, the health of honey bees is a constant concern. In Quebec, as in Canada, maize is a dominant crop. In addition to consuming large areas of agricultural land, maize crops are largely associated with seed coating and repeated applications of herbicides. In a previous study, we exposed honey bees to atrazine, metolachlor and glyphosate (Hedrei Helmer et al., 2014). Bees exposed to atrazine and glyphosate had lower concentrations of β-carotene and retinol (vitamin A derived from β-carotene). In contrast, higher levels of retinol were observed in bees exposed to metolachlor (Hedrei Helmer et al., 2014). At present, we do not know the exact mechanisms that trigger unbalanced retinoid in the bee, but we believe that oxidative damage induced by herbicides could interfere with the oxidation reactions known to maintain the balance between different forms of retinoids. Retinoids are compounds whose molecular structure is similar to retinol (vitamin A). Essential for cell differentiation, vision, reproduction, growth and development, retinoids are also known for their antioxidant properties. What we know about retinoids essentially comes from studies on mammals (Alvarez et al., 2006; Blomhoff and Blomhoff, 2006). In insects, the retinoid system has mainly been studied in Drosophila, and although obvious physiological characteristics distinguish vertebrates from invertebrates, the system appears relatively well preserved between the two phyla. For example, the enzymes that metabolize β-carotene to retinoids (β-carotene-15-15′ monooxygenase1, BCMO1 and β-carotene-9′-10′-dioxygenase 2, BCDO2) have their equivalent in insects, the transmembrane protein NinaB (Neither inactivation nor afterpotential B) (Oberhauser et al., 2008). In Drosophila, mutations in the gene encoding for NinaB lead to the inhibition of this enzyme and ultimately to blindness (von Lintig, 2012). In addition to herbicides, maize cultivation requires regular inputs of fertilizers in the form of manure, slurry or fertilizing residual materials (FRM). In Quebec, applications of FRM represent a significant source of metals, up to 400 mg/kg Cu or 700 ppm Zn (Hébert, 2015). Without necessarily reaching lethal concentrations for the bee, metals can be toxic by competing with essential elements such as calcium involved in olfaction and memory processes in bees (Perisse et al., 2009). According to Valko et al. (2005), several metals such as Fe, Cu, Cr and Co have the ability to produce free radicals, which may alter DNA, cause lipid peroxidation and decrease protein content. In experiments conducted in our laboratory, caged bees were exposed to increasing concentrations of Cd, Pb, Al and Fe. Our data revealed high levels of Cd accumulation in bees and a net increase in metallothionein-like proteins (MTLPs) (Gauthier et al., 2016). We also observed significant lipid peroxidation in bees exposed to Al, Fe, atrazine and glyphosate with or without 0.03 ppm Cd (Gauthier et al., 2016; Jumarie et al., 2017). In recent years, we have developed or adapted several methods to detect biochemical changes in bees exposed to herbicides or metals. The results, however interesting, come from laboratory experiments using sugar solutions spiked with one contaminant at a time or a mixture of two metals (Fe\\Cd) or three contaminants (atrazine-glyphosate-Cd). In the pursuit of our long-term goal of developing biomarkers for bees in agricultural environments, we conducted field studies with beehives exposed in four sites, one organically grown and three conventionally grown maize fields. We assumed that the biochemical parameters measured in the bees would reflect the prooxidant potential between the two conditions. Bees were captured every 15 days and analyzed for carotenoid, retinoid, αtocopherol, lipid peroxidation, protein and MTLP contents. Metals were measured in soil samples, in maize pollen and in bees. Multivariate statistical models were used to identify variables that could better discriminate the maize field conditions and predict bees' belonging to these groups.

2.1. Chemicals

1523

Standards for carotenoids (lutein, zeaxanthin, α-cryptoxanthin, βcryptoxanthin, α-carotene and β-carotene) were purchased from DHI Labs (Hørsholm, Denmark). Standards for retinoids (all-trans-retinol, 9-cis-retinoic acid, 13-cis-retinoic acid, all-trans-retinoic acid and retinaldehyde) and vitamin E (α-tocopherol) as well as antipain dihydrochloride, pepstatin A, 2 thiobarbituric acid (TBA), Tris-(2Carboxyethyl) phosphine hydrochloride (TCEP), 7-fluorobenzo-2-oxa1,3-diazole-4-sulfonic acid ammonium (SBD\\F) were from SigmaAldrich Ltd. (Oakville, ON, Canada). Butylated hydroxytoluene (BHT), Triton X-100 (t-octylphenoxy-polyethoxy-ethanol), malondialdehyde (MDA), ascorbic acid, sodium deoxycholate, trichloroacetic acid, sodium dodecylsulfate and bovine albumin serum (BSA) were from Fisher Scientific (Montreal, QC, Canada) and the rabbit metallothionein-1 was from Enzo Life Science (Brockville, ON, Canada). HPLC-grade solvents were used.

2.2. Field experimental design Field observations took place in the southern area of the Laurentides region (province of Quebec, Canada), where agricultural activities are dominated by bovine and milk productions (MAPAQ, 2014). In 2009, the bee loss reported for this region was in the order of 26% (Bilan de la mortalité hivernale 2008–2009 au sein des colonies d'abeilles du Québec, https://www.agrireseau.net/apiculture/documents/Enqu%C3%AAte_ mortalit%C3%A92009_Bilan.pdf, accessed 15/05/17). Twenty beehives were purchased from a certified organic beekeeper (Api Culture Hautes-Laurentides Inc., Ferme-Neuve, Quebec, Canada), which relies on ISO/IEC (International Organization for Standardization/International Electrotechnical Commission) Guide 65: 1996. The colonies were installed at four maize producers designed A, B, C and D. Composted manures, uncoated seeds and non-synthetic products were practices associated to the organic site (A), while others sites used coated seeds, herbicides and mineral fertilizers (B and D) or fertilizing residual materials from municipal wastewater treatment plants (C). The distance between the sites is reported in the Table 1. In July of 2013, bee colonies (5 hives per site × 4 sites) were installed in the four experimental sites, a few days prior to tassel emergence. One hive per site was equipped with a pollen trap to monitor inputs of maize pollen to the hives. To keep track of possible diseases or high mortality due to pesticide poisoning, a tray (61 cm × 45.7 cm) made of Coroplast® was placed in front of each hive.

2.3. Collection of soil, pollen and honey bees For each site, 10 soil sub-samples were collected randomly in the maize field. The sub-sample was obtained by cleaning a rectangular surface and by excavating a V-shaped area to a depth of 25 cm. The sub-samples (~ 1 kg) were then combined, mixed and cleaned of larger debris (pebbles, roots, etc.) and transferred into a plastic bag and transported to the laboratory. The samples were tested for

Table 1 The distance (km) between the sites selected for this study. Sites

A

B

C

D

A B C D

– 1.04 2.78 3.84

– 2.16 2.97

– 4.10

–

1524

M. Boily et al. / Science of the Total Environment 601–602 (2017) 1522–1532

pH (pH/conductivity meter AP85, Accumet Engineering Corporation, Hudson, MA, USA) and stored in the refrigerator until metal analysis. Pollen was collected 9 times between July 29 and August 29. Pollen traps were emptied and the content was transferred to a polyethylene bag and kept at − 20 °C. Sub-samples (2 g) of pollen (2 sub-samples per site × 4 site × 9 samplings = 72 total) were sorted in the laboratory using a binocular microscope and pollen grains collected directly from the tassels of the maize plants as a reference for appearance, colour and size. Active foragers (~100) were captured in front of each hive, using a collector's net, four times: July 25, August 9, August 27 and September 11. The bees were transferred in glass jars (1 L) and immediately euthanized by placing them on dry ice for 4 min. The frozen bees were then put in pre-identified Whirl-pak® plastic bags and kept at −80 °C until analysis of all parameters as well as metal content. For metal analysis, pollen was harvested randomly from shedding maize plants by shaking the tassels back and forth above a large brand new paper bag. Up to 15–20 g of maize pollen per site were collected, sifted in a row with a sieve of 1.5 mm and 0.75 mm, weighed and stored in plastic bags at −20 °C. 2.4. Metal content analysis in soil, pollen and bees Metal concentrations in all samples (soil, pollen and honey bees) were measured by chemical analysis by the Centre d'expertise en analyse environnementale du Québec (CEAEQ, QC, Canada). Soil samples were dried overnight at 104 °C and kept in the desiccator until extraction. Homogenized and dried soil samples (≈1 g) were extracted with 4 mL of nitric acid (50%) (V/V) and 10 mL of hydrochloric acid (20%) and heated at reflux for 30 min. The extracts were then filtered (0.45 μm) into a 100-mL volumetric flask, rinsed with water and transferred to a plastic bottle. All samples were analyzed in duplicates for their correspondence to blank reactions and reference material (ERA metal in soil). The metal quantifications of the soil samples were conducted in duplicates using inductively-coupled plasma equipped with mass spectrometer equipment (ICP-MS Agilent 7700× Model). Metals in bees (one pool of 6 bees per hive from the last sampling) and pollen (0.1 g per site) were measured following the method described in Jumarie et al. (2017). 2.5. Preparation of tissues Tissues were prepared according to Jumarie et al. (2017), using three pools of 8 whole bees per hive (15 pools per site) and collection date (15 pools × 4 sites × 4 dates), the hives being the experimental units. 2.6. Analysis of carotenoids, retinoids and α-tocopherol For the analysis of carotenoids and tocopherol, we relied on methods that we have previously developed for honey bees (Gauthier et al., 2016; Hedrei Helmer et al., 2014). Retinoids were detected and quantified using the method described in Jumarie et al. (2017). Carotenoids, α-tocopherol and retinoids compounds are presented in the text as percentages of total lipophilic antioxidants (carotenoids and α-tocopherol) or total retinoids. Means and standard deviation values are available as supplementary material (see STable 1). 2.7. Metallothionein-like proteins (MTLPs) Analysis of the metallothionein was carried out exactly as described in the publication by Gauthier et al. (2016) without modification. The MTLP contents in the samples are expressed as MTLPs (ng/g of tissue).

2.8. TBARS analysis (lipid peroxidation) The oxidative stress was measured as presented in Hedrei Helmer et al. (2014). Lipid peroxidation was expressed in mole equivalents of MDA and is reported in the text as TBARS/g tissue. 2.9. Measurement of protein The protein content determination in the samples was performed following the first centrifugation of the homogenate using the Bradford method (Bradford, 1976). 2.10. Statistical analysis Differences for carotenoid, retinoid, α-tocopherol, TBARS (lipid peroxidation), MTLP and protein (mean value 8 to 15 pools of 8 bees) contents were compared using two-way repeated measures ANOVA, taking into account the time (4 times = 4 sampling dates), the site (4 sites: A, B, C and D and the interaction term time*site). In case of noncompliance with the sphericity (homogeneity of variance in repeated measures), the Greenhouse-Geisser correction was used to adjust the degree of freedom. When significant, the models were tested with a multiple comparison test (Bonferroni) to detect differences between sampling dates or sites. Metal contents in honey bees were compared between sites with GLM procedure followed by Tukey test. Relationships between variables including bees' metal content were explored with non-parametric Spearman correlations. A discriminant analysis (derivation of linear combinations of at least two factors that can discriminate pre-established groups) was conducted to determine whether the variables measured in honey bees predicted their association to a given maize condition. The stepwise discriminant model had similar log determinants for each group (site) and covariance matrices were equals as stated by Box'M (p N 0.001). The selection of predictor variables was based on minimum Wilk's lambda value and significant mean differences (F N 3.84; p b 0.05). Prior to analysis, the data were log10-transformed to reduce skewness and stabilize the variances. Discriminant scores from the significant functions' equations (functions 1 and 2) were used to produce box plots illustrating the effectiveness of the model. Pairwise comparisons were performed on the mean of the discriminant scores with one-way analysis of variance (ANOVA) followed by Tukey test. All statistical analyses were carried out with SPSS 20.0 software (IBM, Amonk, NY, USA). 3. Results and discussion 3.1. Collection of pollen by bees and protein contents The honey bees foraging in maize fields were exposed to a mixture of compounds (herbicides, insecticides, metals, etc.), several of which are known to induce oxidative stress (Valko et al., 2005) or disrupt the retinoid system (Defo et al., 2014) in vertebrates. Using the collected pollen samples, we examined the inputs of the maize pollen to the hives equipped with pollen traps. The graph in Fig. 1 clearly demonstrates that from the beginning of flowering, maize pollen reaches 35% or more in the traps of the sites B and C. In the maize fields A and D, percentages do not exceed 15%. Slight fluctuations were observed in most sites from August 10 to the end of this month. For example, in the field A, maize pollen accounted for 10% of the total found in the trap at the end of August. While foragers do not consume large amounts of pollen, larvae and young bees (5–6 days) are dependent on it for protein and amino acids to ensure their growth and development (Malerbo-Souza, 2011). Corn pollen as a food source for honey bees is currently subject to debate. In a study by Höcherl et al. (2012), reduced longevity and brood production were observed in bees fed with corn pollen. Thus, without

M. Boily et al. / Science of the Total Environment 601–602 (2017) 1522–1532

1525

Fig. 1. Chronological collection of maize pollen (%) and honey bees (black arrows) under the four maize field sites (A, B, C and D).

claiming that maize pollen is necessarily harmful to bees, the latter is considered a poor source of protein with 15% (Malerbo-Souza, 2011). According to Pernal and Currie (2001), bees are not able to distinguish a good source of protein from a bad source, and according to Keller et al. (2005), they are more attracted to maize pollen than to other plants that would be much better sources. Variations were observed for proteins measured in bees (D7.9, 132.3 = 2.26 p b 0.05), but without demonstrating stable trends for either site as a function of time (Fig. 4I). Protein content ranged between 211 and 282 mg/g of tissue, which was similar to the means obtained in our previous studies with caged bees, namely 253 ± 23.3 mg/g of tissue (Gauthier et al., 2016) and 248 ± 44.2 mg/g of tissue (Jumarie et al., 2017). 3.2. Lipophilic antioxidants For honey bees, the pollen is the source of fatty acids, carotenoids and lipophilic vitamins such as vitamins A (retinol) and E (tocopherol). Up to 1.53 mg/100 g of β-carotene and 6.21 mg/100 g of tocopherol may be present in corn pollen (Chantarudee et al., 2012). According to Kayser (1982), carotenoids in adult honey bees are stored in adipose tissues like fat bodies. When needed, carotenoids are mobilized in the gastrointestinal tract and converted into retinoids and later into retinaldehyde compounds, essential for vision (Arshavsky, 2010; von Lintig, 2012). Variations of carotenoids and xanthophylls during the field exposure in connection with the seasonal succession of pollen sources were expected, as shown in Fig. 2. In all sites, β-carotene (D9,159 = 8.76 p b 0.001) tended to increase with time and emerged as the most abundant carotenoid measured in honey bees, followed by α-carotene, β-cryptoxanthin, α-cryptoxanthin, zeaxanthin and lutein. Most lipophilic antioxidants fluctuated depending on the sampling times and sites: α-cryptoxanthin (D9,159 = 6.32 p b 0.001), lutein (D9, 156 = 12.5 p b 0.001), zeaxanthin (D9,156 = 7.26 p b 0.001) and α-tocopherol (D9,156 = 7.26 p b 0.001). The concentration of βcryptoxanthin increased steadily over the experiment in all sites except for B and starting at time 3, values at site A were higher than that of all other sites (D9,159 = 7.30 p b 0.001) (see STable 1 for details). In Jumarie et al. (2017), the range values for zeaxanthin (27.1 to 67.1 ng/g of tissue) and α-tocopherol (131 to 214 ng/g of tissue) were similar to the concentration observed in the present study (18.1 to 111 ng/g of tissue for zeaxanthin and 76.6 to 249 ng/g of tissue for αtocopherol) while lower concentrations were found for α-carotene (31.3 to 45.7 vs 60.7 to 103 ng/g of tissue) and β-carotene (83.9 to

145 vs 116 to 299 ng/g of tissue). The fact that higher carotene values were measured in bees that had access to a source of pollen (this study) was not surprising. On the other hand, that these bees had similar concentrations of zeaxanthin and α-tocopherol than bees kept for 10 days on a sucrose-syrup diet for any food is rather surprising. It must therefore be concluded that these compounds are part of a reserve in the tissues of the bee. 3.3. Retinoid compounds The cleavage of carotenoid compounds leads to the formation of retinaldehyde and subsequent polar retinoids. In our field experiment, RALD tended to increase with time (D3.132 = 29.8 p b 0.001) without showing any differences between sites (Fig. 3). The fact that higher concentrations were measured with time could be linked to higher contents of carotenoids. In a study conducted on the moth by Oberhauser et al. (2008), RALD can be metabolized not only from β-carotene but also from cryptoxanthin, zeaxanthin, α-carotene or lutein. In our study, positive correlations (p ≤ 0.001; n = 224) were found between RALD and α-carotene (r = 0.283), β-carotene (r = 0.357) β-cryptoxanthin (r = 0.380) and lutein (r = 0.400), showing a possible link between RALD and these compounds. To get a better idea on the formation of RALD from carotenoid precursors, the activity of NinaB and the formation of RALD could be measured in vitro using bee homogenates and different substrates of carotenes or xanthophylls. Although the cleavage of carotenoids and xanthophylls can lead to greater amounts of RALD, this reactive intermediate compound is unlikely to be accumulated in the tissues. Instead, it will tend to be metabolized into ROH or retinoic acid (RA). Among retinoid compounds, the statistical model associated with retinol was significant (D6.89,108 = 2.12 p b 0.05) without demonstrating a clear pattern for either time or site (Fig. 3 and STable 1). However, different results were observed for retinoic acids and their metabolites. At time 1, bees from the site C site had significantly higher concentration of 13-cis-RA than those from sites A and B (Fig. 3). From time 2, values of 13-cis RA had decreased by at least 50% and no significant difference between fields were found at this time and at time 3. However, at the last sampling, bees from the site C had significantly higher concentrations of 13-cis-RA than those existing in other sites (D5.9,77.9 = 4.89 p b 0.001). Compared to 13-cis-RA, concentrations of 9-cis-RA were not very high. In fact, in most fields, the concentration in the bees did not exceed 25 ng/g of tissue (means + SD, see STable 1 for details). Nevertheless, significantly higher values were found in bees from site C at time 4 (D6.6,89.1 = 1.98 p b 0.05). Higher concentrations of 13-cis-4-oxo-RA

1526

M. Boily et al. / Science of the Total Environment 601–602 (2017) 1522–1532

Carotenes

Xanthophylls

-Carotene

A

B

C

D

1

-Cryptoxanthin

Lutein Zeaxanthin

-tocopherol

11.6%

11.3%

11.6%

10.9%

25.4%

27.7%

30.3%

36.4%

10.4% 11.2% 8.7% 14.5%

7.3% 18.6% 7.7% 8.1%

5.9% 18.1% 5.4% 5.8%

6.5% 16.2% 7.3% 6.5%

18.1%

19.3%

22.9%

16.2%

12.4%

11.3%

12.9%

12.2%

24.4%

34.5%

33.4%

35.0%

9.3% 11.3% 7.5% 19.1%

5.2% 14.2% 7.2% 9.0%

5.8% 11.3% 5.5% 5.7%

16.1%

18.7%

25.4%

7.6% 12.6% 8.0% 7.5% 17.1%

14.5%

13.5%

11.5%

11.0%

29.8%

33.6%

32.9%

37.0%

10.2% 11.0% 8.2% 4.3%

6.1% 11.1% 6.0% 8.0%

5.9% 9.0% 5.9% 5.9%

7.5% 9.6% 8.8% 7.2%

22.0%

21.6%

14.6%

15.0%

11.6%

13.1%

27.1%

31.0%

35.1%

33.5%

9.9% 8.9% 9.5% 12.1%

7.9% 12.8% 8.2% 7.5%

6.0% 11.4% 7.0% 5.8%

7.6% 11.3% 8.3% 6.5%

17.9%

17.6%

23.1%

19.7%

2

Times

28.8%

3

18.9%

4

Fig. 2. Carotenoids and α-tocopherol (% of total lipophilic antioxydants) measured in bees exposed in four maize fields (A, B, C and D) and sampled four times between July 21 and September 11, 2013.

were measured in bees existing in sites B (times 1 and 2) and C (times 2 and 4) (D7.3,88.3 = 3.60 p b 0.01) while higher values of the metabolite all-trans-4-oxo-RA were found in bees from the field C (times 1 and 4) (D6.3,71.1 = 2.54 p b 0.05). Since the balance between the different forms of retinoids is ensured by reduction-oxidation processes, one may assume that these reactions

were altered in the bees existing in the fields B and C, as higher metabolite values were observed in bees from these sites. The loss of a retinoid equilibrium could be due to i) exposure to oxidative contaminants or ii) a lack of endogenous antioxidants (enzymatic or not) to neutralize excess free radicals. According to our analyses, potential antioxidants were present in all sites during the field exposure. At this point,

M. Boily et al. / Science of the Total Environment 601–602 (2017) 1522–1532

1527

Fig. 3. Retinoid compounds (% of total retinoids) measured in bees exposed in four maize fields (A, B, C and D) and sampled four times between July 21 and September 11, 2013.

however, we do not know whether carotenoids, α-tocopherol and to a lesser extent retinoids can play an effective role as antioxidants for the bee. These results will require further investigation if only for the βcarotene-retinol-retinoic acid relationship and the possible implications for vision. A study by von Lintig et al. (2001) on Drosophila revealed that a mutation of the gene coding for the enzyme NinaB rendered the flies blind. Retinoic acid is also of interest as this compound is very important during the development and growth of insects. Altered retinoic concentrations led to deformities in the tick (Rhodnius prolixus) (Nakamura et al., 2007) and Drosophila melanogaster (Halme et al., 2010) as well as complications during metamorphosis in the firebug (Pyrrhocoris

apterus), the red cotton stainer (Decipifus cingulatus) and the mealworm (Tenebrio molitor) (Němec et al., 1993). 3.4. TBARS, MTLPs and exposure of bees to metals Lipid peroxidation as TBARS is a quick and general manifestation of oxidative damage. At time 1, significantly higher values were observed in site B when compared with sites A and D (D6.9123.3 = 4.18 p b 0.001) and at time 3, the bees from the A site had significantly lower TBARS content compared to others (Fig. 4II). High values of TBARS in time 1 may be explained, at least in part, with the installation of the hives in the fields. Indeed, moving the hives to a new environment

1528

M. Boily et al. / Science of the Total Environment 601–602 (2017) 1522–1532

Fig. 4. Means and SD of protein content (I), TBARS (II) and MTLPs (III) measured in bees exposed to four maize fields (A, B, C and D) and sampled four times between July 21 and September 11, 2013. Means were compared between times (capital letter) and between sites within each week (lower case) using two-way repeated measures ANOVA. Means not sharing the same letters are significantly different.

may have disoriented the foragers for a few days (Riddell Pearce et al., 2013). The fact that beehives were moved from the organic beekeeper to agricultural areas, where foraging resources varied in abundance and quality, must also be taken into consideration. These factors, identified as stressful for bees (Nelson and Jay, 1989; Ratnieks and Carreck, 2010) could have led to a general stress, an increased metabolism and consequently an increase in free radicals in tissues. In our study, the oxidative injury observed in bees could be limited by the presence of antioxidants, as negative relationships were found between TBARS and lutein

(Rho = −0.294), α-tocopherol (Rho = − 0.248), β-cryptoxanthin (Rho = −0.354) and β-carotene (Rho = −0.334) (p b 0.01; n = 236). Most parameters that we measured in bees varied with time or site. With respect to carotenoids, time differences may be associated with the succession of floral resources in the fields during the hive exposure period. If one relies on the maize pollen percentage found in hives equipped with traps (Fig. 1), α-carotene and β-carotene compounds could be linked to this source under at least two sites, B and C, where the highest maize pollen percentages during the first two times was observed, which also met the highest α-carotene and β-carotene values. However, the maize pollen found in the traps cannot explain the variations observed for all the factors analyzed. The same applies to the cultivar used. The cultivar of the organic site was, of course, different from that of the other sites which all had the same cultivar. While the flowering of maize and the production of tassel were synchronous at all sites, we do not know if the quality of the pollen produced could be same between sites. The comparison of the sites for each time revealed significant differences for the parameters related to the metabolism of the bees. Higher concentrations of 13-cis-RA were found in the sites C (time 1), D (time 1) and B (time 2). The differences were most pronounced for the metabolites of retinoic acid, 13-cis-4 oxo-RA and alltrans-4-oxo-RA. These compounds are derived from metabolic oxidation-reduction reactions. In addition, TBARS values were also higher in bees from the sites B (time 1 and 3) and C and D (time 3). All these findings led us to consider the phytosanitary treatments that were practiced in the maize fields of our study as fertilizers may bring a non negligible inputs of metals in soils, and consequently, in plants and pollen. The maize fields were then analyzed for their content of metals in the soil, pollen and bees (Table 2). While the organic condition had the highest concentrations of metals in the pollen, the values measured in the bees were equal to or less than those observed under the other sites. The concentrations of metals measured in the soil were not always linked to the metal contents measured in the maize pollen or bees. For example, in the soil A, the Cu concentration was 27 mg/kg, against 14 mg/kg in the soil D. However, the concentration of Cu in the pollen (4.3 mg/kg) from site D was 1.4 times higher than that found in the site A. We also observed that lack of proportionality with metals accumulated by bees. In the field C, Al was less concentrated compared to that of the site A: 6.5 times for the soil and 16 times for the pollen. However, the bees from the field C site had twice the concentration of Al in their bodies compared to the bees from the site A after several weeks of exposure. The lack of relationship between the metals detected in soil and in pollen can be attributed to i) atmospheric deposition of metals, ii) drifts during application of fertilizers or pesticides, or iii) the proportion of clay, sand and silt in the soil, which can affect the metals' bioavailability and consequently their uptake by maize plants. Soil pH was not markedly different between sites (5.6 to 5.8) but the composition of the soil varied substantially and could explain the lack of proportionality between metal contents in soil and in pollen. Indeed, the soil in A was dominated by clay, which tends to “retain” the metals in the soil, whereas the soil in site C was mostly sand, which may have favoured the uptake of soluble metals by the maize plants. As for the site B, the soil had a clay/silt mixture, while the clay and sand were soil characteristics of the field D. However, estimation of the transfer ratios (i.e. metal content in pollen/metal content in soil) did not allow to distinguish the influence of soil characteristics on metal accumulation in maize pollen for none of the metal studied. Nevertheless it is interesting to note that, regardless of the location, the transfer ratios from soil to pollen varied considerably between metals (ranging from b0.001 up to 2.6) and was as followed: Co = Al = Fe b Cr = Ba b Pb b Sr b Cu b Ni b Mn b Zn. Although Al content in soil was relatively high, this metal would not be available to the pollen. In contrast, Zn seemed bioconcentrated in maize pollen. Most organisms exposed to metals have the ability to induce the synthesis of specialized proteins, which, by binding to metals, limit the level of the free metal ion and, consequently, its toxicity. While the

M. Boily et al. / Science of the Total Environment 601–602 (2017) 1522–1532

1529

Table 2 Metal contents (mg/kg) measured in soil, maize pollen and honey bees from the four maize fields (A, B, C and D). Metals

Aluminum Baryum Cadmium Chromium Cobalt Copper Iron Lead Manganese Molybdenum Nickel Selenium Strontium Zinc

A

B

C

D

Soil

Pollen

Bees⁎

Soil

Pollen

Bees

Soil

Pollen

Bees

Soil

Pollen

Bees

34,700 154 b0.25 97 23 27 46,500 13 826 b0.5 49 b0.7 41 95

89 0.78 0.02 1.4 0.04 3.0 110 0.18 18 0.58 1.5 b0.05 1.6 74

12.58 ± 3.50a 2.89 ± 0.74ab 0.23 ± 0.02a 0.11 ± 0.02ab 0.10 ± 0.02 26.4 ± 2.30 174 ± 18a 0.09 ± 0.03a 77.3 ± 19.3 0.58 ± 0.05a 0.16 ± 0.05 0.35 ± 0.06 5.55 ± 1.45a 134 ± 11a

23,300 121 b0.25 69 17 19 29,700 10 552 b0.5 35 b0.7 31 78

11 0.51 0.04 0.20 0.01 2.7 32 0.03 17 0.33 0.39 b0.05 1.3 76

12.13 ± 3.74a 3.79 ± 0.86b 0.22 ± 0.04a 0.09 ± 0.0a 0.11 ± 0.02 29.65 ± 3.01 194 ± 14ab 0.09 ± 0.02a 84.3 ± 20.3 0.59 ± 0.06a 0.16 ± 0.04 0.30 ± 0.04 6.16 ± 1.13a 148 ± 13ab

5360 46 b0.25 12 3 b7 8620 4 128 b0.5 6 b0.7 19 24

5.3 1.8 0.02 0.08 0.01 2.2 34 0.02 15 0.71 0.32 b0.05 3.0 63

27.84 ± 0.04b 7.44 ± 1.00c 0.31 ± 0.05b 0.14 ± 0.03b 0.13 ± 0.01 28.02 ± 2.12 230 ± 22b 0.16 ± 0.06b 91.2 ± 9.0 0.71 ± 0.09b 0.20 ± 0.04 0.32 ± 0.05 17.55 ± 4.05b 154 ± 7b

19,800 93 b0.25 53 17 14 29,100 10 663 b0.5 25 b0.7 24 68

38 0.85 0.02 1.3 0.03 4.3 77 0.36 30 0.19 2.0 b0.05 2.1 68

15.41 ± 6.39a 2.30 ± 0.44a 0.25 ± 0.04a 0.09 ± 0.02a 0.11 ± 0.02 25.80 ± 2.27 211 ± 30ab 0.11 ± 0.03ab 75.8 ± 16.5 0.65 ± 0.06ab 0.18 ± 0.02 0.34 ± 0.09 5.28 ± 2.05a 148 ± 15ab

b = Values under LD (Limits of detection). ⁎ Metal in honey bees (n = 5 pools of 6 bees sampled 11/09/13) were compared with GLM procedure. Significant models were followed by the Tukey test. Values not sharing the same letters are statistically different.

statistical model for MTLPs showed only a tendency (D9.159 = 1.86 p b 0.1), mean values were higher in sites B and C at time 1 (Fig. 4III). If one refers to Fig. 1, the sites B and C were those with the highest maize pollen percentages measured in the pollen traps. In our previous experiment, we observed a significant induction of MTLPs in honey bees exposed to Cd-contaminated syrup (0.01 mg/L) when compared to bees fed with syrup alone (Gauthier et al., 2016). Cd was present in the maize pollen (Table 2) as well as other metals known to induce the synthesis of metallothionein: Cr, Co, Fe, Mn and Ni. Several metals were measured in bees at time 4 and positive correlations (Rho = 1.0; p b 0.001; n = 4) were found between the mean levels of MTLPs and of Al, Fe, Mo, Zn, and Pb (Rho = 0.949; p ≤ 0.05; n = 4) as well as TBARS mean values and Pb (Rho = 0.949; p ≤ 0.05; n = 4), Cd and Ni (Rho = 1.0; p ≤ 0.001; n = 4). Also, the last sampling of our study revealed significant negative relationships between the bees' concentration of Al, Mo and Zn and the RALD compound (r = −1.0; p b 0.01; n = 4). It should be noted, also, that metals were not the only prooxidant contaminants present in maize fields. In all sites except the organic one, producers used synthetic herbicides such as atrazine, metolachlor and glyphosate known to induce oxidative damage in vertebrates (Nwani et al., 2010; Thornton et al., 2010) and to alter the retinoid system (Hedrei Helmer et al., 2014). Added to this were the neonicotinoid insecticides from the coated seeds used in sites B, C and D. According to recent studies, neonicotinoids have a prooxidizing power as shown in birds (LopezAntia et al., 2015), fish (Ge et al., 2015) and honey bees (BadiouBénéteau et al., 2012).

In most of the models generated by discriminant analysis, the function 1 proved to be more suitable in distinguishing sites (Table 3 and Fig. 6I, 6IV, 6VII and 6X). At time 1, this function was characterized by coefficients attributed to β-cryptoxanthin and lutein, potential antioxidants. Conversely, there were coefficients associated with MTLPs and 13-cis-4-oxo-RA, products derived from physiological reactions linked to exposure to metals (MTLPs) and oxidation of retinoic acid. The highest classification was obtained under the site A (100%), and this distinction is well depicted by the box plots of discriminant scores (Fig. 6II), while the site D was better isolated with function 2 (Fig. 6III). The discriminant analysis results of the time 2 rather reflected the lipophilic antioxidant resources available for the bees under the different maize fields. When considering the score from the function 1 (Fig. 6V), the sites B and D could not be distinguished from each other; these sites were better discriminated by the function 2 (Fig. 6VI). No variables associated with oxidative processes were included in the equations generated by the model at time 2 suggesting that the oxidative damages observed during time 1 were no longer important enough to influence the distinction between the sites. This reduction in oxidative damages could be attributed to the habituation of bees to their respective sites—in the event that the installation of bees in the fields was the main cause of the oxidative injuries—or the neutralization of oxidative

3.5. Distinguishing sites with discriminant analysis Although significant results were obtained with the two-way repeated measures ANOVA, it was difficult to distinguish the influence of the sites on variables tested. In order to identify the variables that best distinguish the four fields in which the bees were exposed, we used the discriminant analysis). Given that several parameters measured in bees varied with times, we first tested this factor. The stepwise discriminant analysis based on times retained 10 variables out of 16 with significant values: Wilks lambda (0.415 b λ b 0.035; 32.2 b F b 76.1 and p ≤ 0.001) (see Fig. 5 and STable 2 in supplementary material for details). With this model, 94.1% of the cross-referenced cases were well classified: time 1, 100%; time 2, 87.8%; time 3, 88.1%; and time 4, 100%. This result led us to investigate the influence of parameters to discriminate the conditions within each time. Furthermore, this exercise was also aimed at reducing the number of explanatory variables, an asset of discriminant analysis.

Fig. 5. Scatterplots of canonical discriminant functions based on selected variables from stepwise discriminant analysis. Black symbols are for centroid value within each week.

1530

M. Boily et al. / Science of the Total Environment 601–602 (2017) 1522–1532

Table 3 Discriminant analysis parameters (model statistics, canonical functions of selected variables and forecast predictions) for the four sites (A, B, C and D) and the four sampling times (1, 2, 3 and 4). Times

Model Box'M test p Cross validation (%) a Predicted group membership A B C D Canonical functions Wilk's lambda χ2 p Variability explained (%) Selected variables/Coefficientsb Constant (D) α-carotene β-carotene β-cryptoxanthin Lutein α-tocopherol Retinol 13-cis-4-oxo-RA all-trans-4-oxo-RA TBARS MTLPs a b

1

2

3

4

68.06 0.003 80.4

62.58 0.427 86.7

50.74 0.730 85.2

36.25 0.433 71.2

73.3 64.3 100 84.6 1 0.083 99.5 0.000 (66.9%)

2 0.342 42.91 0.000 (26.5%)

86.6 80.0 80.0 100 1 0.074 90.04 0.000 (53.7%)

2 0.253 47.4 0.000 (34.2%)

100 69.2 84.6 86.7 1 0.111 84.5 0.000 (82.1%)

2 0.517 25.4 0.001 (11.6%)

66.7 71.4 76.9 69.2 1 0.262 62.88 0.000 (70.1%)

2 0.612 23.07 0.001 (21.8%)

−5.47

−16.35

−2.14

−3.71

−8.92

10.1 −0.54

−16.2 −8.77 10.6 4.45 2.74 2.06

−8.91

3.72 −5.41

−7.73 6.09 4.04 −7.94 −7.70 7.22

−9.09 4.57 6.07

−1.40 4.51 −4.08

−9.21 9.77

2.76 1.76

1.09

-3.01

3.86 −0.99

3.26 −3.17

−2.11 3.96 6.05

2.40 2.03

3.17

Cross validated grouped cases correctly classified. Each case was classified by the functions derived from all cases other than that case. Unstandardized canonical function coefficients (structure matrix).

damages through the increased concentration of antioxidants, noticeable in bees as early as time 2 (α-carotene, β-carotene and βcryptoxanthin, Fig. 2). At time 3, the model comprised coefficients associated to oxidation processes: peroxidation of the lipids as TBARS and the retinoic acid metabolite, all-trans-4-oxo-RA. When the discriminant scores from function 1 were plotted for each group (Fig. 6VIII), we clearly observed that the site A was well isolated from the three other groups, among which the sites B and D had similar distributions. Function 2 from this model explained only 11.6% of the total variability and poorly discriminated between the sites. Only the site D proved to be significantly different from the others (Fig. 6IX). The model estimates were slightly lower for the fourth and final sampling time. The model retained the variables β-cryptoxanthin, lutein and 13-cis-4-oxo-RA in addition to the retinol (Table 3). According to the discriminating scores of function 1, all sites would be distinct from each other (Fig. 6XI), while function 2 was better able to discriminate the field B from all others (Fig. 6XII). Except for time 2, all discriminant analyses indicated that what distinguished the sites over time was related to the oxidation processes observed in bees with parameters such as retinol, TBARS, all-trans-4-oxoRA, MTLPs and 13-cis-4-oxo-RA. These “responses” were mainly associated with two xanthophylls present in all equations computed by the discriminant analysis models: β-cryptoxanthin and lutein. According to the illustration of function 1 (Fig. 6II, V, VIII and XI), the site C differed from all others. The discriminant analysis produced made it possible to distinguish the sites A and C, whereas the sites B and D were not as clearly separated. Given that the maize fields differed on the basis of a dominant (not exclusive) use of fertilizers but not of phytosanitary treatments, we believe that it was still remarkable that in 9 cases out of 16 (4 sites × 4 times, Table 3), 80% to 100% of data based on the new canonical variables were correctly reclassified into their original site. Moreover, considering that some sites were separated by a relatively short

distance, for example the sites A and B (1.04 km), differences were still found between these sites by the function 1 of the discriminant models at times 2, 3 and 4. 4. Conclusion Our study showed that bees who foraged in maize crops are also exposed to several metals, in addition to herbicides and insecticides, through the collection of pollen, contaminated from the soil (and the uptake by the plants) or by the repeated fertilizer applications. The bee, like all living organisms, needs essential metals such as zinc, calcium and iron. The excess of metals is what constitutes a risk for the bee as most metals lead to oxidative reactions. However, under the four sites of our study and at each sampling, we measured antioxidants (carotenoids and α-tocopherol) and metal-binding proteins (MTLPs) in bees. This arsenal, combined with other antioxidants that we have not measured (e.g. vitamin C) must be efficient enough to neutralize free radicals and limit oxidative damage. We believe that this balance was encountered in the organic site but not necessarily in sites of conventional agricultural practices for which values for TBARS and oxidation products of retinoic acid were higher than in the organic site. Also higher consumption of maize pollen would enhance metal uptake and could challenge bees antioxidant system. Indeed in the site A, according to the examination of pollen traps, bees harvested little maize pollen, and at the end of the experiment had accumulated fewer metals than bees from the C site. Moreover, bees in sites B and C had a higher percentage of maize pollen and, while the discriminant analysis failed to isolate the bees existing in site B from other sites (except at time 4), high values for 13-cis-4-oxo-RA were observed at times 1 and 2. Oxidative stress was not found to be a cause of mortality of the bees, at least not directly, because we did not detect any abnormal mortality under any of the sites. The mortality for each hive was between 7 and 14 individuals (data not shown), which is considered normal (http://www.

M. Boily et al. / Science of the Total Environment 601–602 (2017) 1522–1532

1531

Fig. 6. Scatterplots of canonical discriminant functions based on selected parameters from stepwise discriminant analysis for the four sampling times as follows: I) Time 1, IV) Time 2, VII) Time 3 and X) Time 4. Black symbols are for centroid value. Box plots II, V, VIII and XI represent the distribution of discriminant scores for times 1 to 4 according to Function 1, and box plots III, VI, IX and XII represent the distribution of discriminant scores for times 1 to 4 according to Function 2.

cari.be/medias/actuapi/actuapi48.pdf, 15/05/17). However, oxidative stress over a prolonged period can weaken colonies and affect survival after wintering. Our study was limited to the analysis of foraging bees that consume little pollen. We have no knowledge or understanding of the prooxidant/antioxidant effects that prevail for larvae, fed mainly with pollen or other wild insect pollinators. In order to limit oxidative damage to bees, fertilizers should be applied according to the soil trace metal contents accumulated over time. In order to acquire various proteins and antioxidants, the bee must have access to a good diversity of flowers, which is sorely lacking in the majority of the land where maize is grown. Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.scitotenv.2017.06.014. Acknowledgements The authors would like to thank the Direction des matières dangereuses et des pesticides, Ministère du Développement Durable, de l'Environnement et de la Lutte contre les Changements Climatiques

(MDDELCC) for the metal analysis. They also thank Nathalie Dassylva from CEAEQ (MDDELCC) for her valuable cooperation. We recognize the collaboration of Philip Lavoie and Donna Clark (Club AgriEnvironmental d'Argenteuil Inc.) and Yveline Martin (Club Conseil Bio-action) for facilitating contacts with maize producers in the Laurentides region. We thank Stephanie Hedrei Helmer for fieldwork and Isabelle Lecavalier-Mondor for technical assistance. This study was supported by a grant from the Programme de Soutien à l'Innovation en Agroalimentaire (PSIA) from the Ministère de l'Agriculture, des Pêcheries et de l'Alimentation du Québec (MAPAQ), awarded to M. Boily (#811175). References Alvarez, S.M., Gómez, N.N., Scardapane, L., Fornés, M.W., Giménez, M.S., 2006. Effects of chronic exposure to cadmium on prostate lipids and morphology. Biometals 20: 727. http://dx.doi.org/10.1007/s10534-006-9036-9. Arshavsky, V.Y., 2010. Vision: the retinoid cycle in drosophila. Curr. Biol. http://dx.doi.org/ 10.1016/j.cub.2009.12.039. Badiou-Bénéteau, A., Carvalho, S.M., Brunet, J.L., Carvalho, G.A., Buleté, A., Giroud, B., Belzunces, L.P., 2012. Development of biomarkers of exposure to xenobiotics in the

1532

M. Boily et al. / Science of the Total Environment 601–602 (2017) 1522–1532

honey bee Apis mellifera: application to the systemic insecticide thiamethoxam. Ecotoxicol. Environ. Saf. 82:22–31. http://dx.doi.org/10.1016/j.ecoenv.2012.05.005. Blomhoff, R., Blomhoff, H.K., 2006. Overview of retinoid metabolism and function. J. Neurobiol. http://dx.doi.org/10.1002/neu.20242. Bradford, M.M., 1976. A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal. Biochem. 72:248–254. http://dx.doi.org/10.1016/0003-2697(76)90527-3. Chantarudee, A., Phuwapraisirisan, P., Kimura, K., Okuyama, M., Mori, H., Kimura, A., Chanchao, C., 2012. Chemical constituents and free radical scavenging activity of corn pollen collected from Apis mellifera hives compared to floral corn pollen at Nan, Thailand. BMC Complement. Altern. Med. 12:45. http://dx.doi.org/10.1186/ 1472-6882-12-45. Defo, M.A., Spear, P.A., Couture, P., 2014. Consequences of metal exposure on retinoid metabolism in vertebrates: a review. Toxicol. Lett. 225:1–11. http://dx.doi.org/10.1016/j. toxlet.2013.11.024. Gauthier, M., Aras, P., Jumarie, C., Boily, M., 2016. Low dietary levels of Al, Pb and Cd may affect the non-enzymatic antioxidant capacity in caged honey bees (Apis mellifera). Chemosphere 144:848–854. http://dx.doi.org/10.1016/j.chemosphere.2015.09.057. Ge, W., Yan, S., Wang, J., Zhu, L., Chen, A., Wang, J., 2015. Oxidative stress and DNA damage induced by imidacloprid in zebrafish (Danio rerio). J. Agric. Food Chem. 63: 1856–1862. http://dx.doi.org/10.1021/jf504895h. Halme, A., Cheng, M., Hariharan, I.K., 2010. Retinoids regulate a developmental checkpoint for tissue regeneration in drosophila. Curr. Biol. 20:458–463. http://dx.doi. org/10.1016/j.cub.2010.01.038. Hébert, M., 2015. Guide sur le recyclage des matières résiduelles fertilisantes: Critères de référence et normes réglementaires – Édition 2015. Québec. ISBN- 978-2-550-729549, (216 pages). Hedrei Helmer, S., Kerbaol, A., Aras, P., Jumarie, C., Boily, M., 2014. Effects of realistic doses of atrazine, metolachlor, and glyphosate on lipid peroxidation and diet-derived antioxidants in caged honey bees (Apis mellifera). Environ. Sci. Pollut. Res. 22:8010–8021. http://dx.doi.org/10.1007/s11356-014-2879-7. Höcherl, N., Siede, R., Illies, I., Gätschenberger, H., Tautz, J., 2012. Evaluation of the nutritive value of maize for honey bees. J. Insect Physiol. 58:278–285. http://dx.doi.org/ 10.1016/j.jinsphys.2011.12.001. Jumarie, C., Aras, P., Boily, M., 2017. Mixtures of herbicides and metals affect the redox system of honey bees. Chemosphere 168:163–170. http://dx.doi.org/10.1016/j. chemosphere.2016.10.056. Kayser, H., 1982. Carotenoids in insects. In: Britton, G., Goodwin, T.W. (Eds.), Carotenoid Chemistry and Biochemistry. Pergamon Press, Toronto http://dx.doi.org/10.1016/ B978-0-08-026224-6.50030-9. Keller, I., Fluri, P., Imdorf, A., 2005. Pollen nutrition and colony development in honey bees: part I. Bee World 86:3–10. http://dx.doi.org/10.1080/0005772X.2005. 11099641. von Lintig, J., 2012. Metabolism of carotenoids and retinoids related to vision. J. Biol. Chem. 287:1627–1634. http://dx.doi.org/10.1074/jbc.R111.303990.

von Lintig, J., Dreher, A., Kiefer, C., Wernet, M.F., Vogt, K., 2001. Analysis of the blind drosophila mutant NinaB identifies the gene encoding the key enzyme for vitamin A formation in vivo. Proc. Natl. Acad. Sci. U. S. A. 98:1130–1135. http://dx.doi.org/10.1073/ pnas.031576398. Lopez-Antia, A., Ortiz-Santaliestra, M.E., Mougeot, F., Mateo, R., 2015. Imidaclopridtreated seed ingestion has lethal effect on adult partridges and reduces both breeding investment and offspring immunity. Environ. Res. 136:97–107. http://dx.doi.org/10. 1016/j.envres.2014.10.023. Malerbo-Souza, D.T., 2011. The corn pollen as food source for honey bees. Acta Sci. Agron. 33:701–704. http://dx.doi.org/10.4025/actasciagron.v33i4.10533. MAPAQ, 2014. Portrait de l'industrie bioalimentaire des Laurentides. Ministère de l'Agriculture, des Pêcheries et de l'Alimentation du Québec (31 pp.). Nakamura, A., Stiebler, R., Fantappié, M.R., Fialho, E., Masuda, H., Oliveira, M.F., 2007. Effects of retinoids and juvenoids on moult and on phenoloxidase activity in the blood-sucking insect Rhodnius prolixus. Acta Trop. 103:222–230. http://dx.doi.org/ 10.1016/j.actatropica.2007.06.009. Nelson, D.L., Jay, S.C., 1989. The effect of colony relocation on loss and disorientation of honey bees. Apidologie 20:245–250. http://dx.doi.org/10.1051/apido:19890306. Němec, V., Kodrík, D., Matolín, S., Laufer, H., 1993. Juvenile hormone-like effects of retinoic acid in insect metamorphosis, embryogenesis and reproduction. J. Insect Physiol. 39: 1083–1093. http://dx.doi.org/10.1016/0022-1910(93)90132-B. Nwani, C.D., Lakra, W.S., Nagpure, N.S., Kumar, R., Kushwaha, B., Kumar, S., 2010. Toxicity of the herbicide atrazine: effects on lipid peroxidation and activities of antioxidant enzymes in the freshwater fish Channa punctatus (Bloch). Int. J. Environ. Res. Public Health 7:3298–3312. http://dx.doi.org/10.3390/ijerph7083298. Oberhauser, V., Voolstra, O., Bangert, A., von Lintig, J., Vogt, K., 2008. NinaB combines carotenoid oxygenase and retinoid isomerase activity in a single polypeptide. Proc. Natl. Acad. Sci. U. S. A. 105:19000–19005. http://dx.doi.org/10.1073/pnas.0807805105. Perisse, E., Raymond-Delpech, V., Néant, I., Matsumoto, Y., Leclerc, C., Moreau, M., Sandoz, J.-C., 2009. Early calcium increase triggers the formation of olfactory long-term memory in honey bees. BMC Biol. 7:30. http://dx.doi.org/10.1186/1741-7007-7-30. Pernal, S.F., Currie, R.W., 2001. The influence of pollen quality on foraging behavior in honey bees (Apis mellifera L.). Behav. Ecol. Sociobiol. 51:53–68. http://dx.doi.org/10. 1007/s002650100412. Ratnieks, F.L.W., Carreck, N.L., 2010. Ecology. Clarity on honey bee collapse? Science 327: 152–153. http://dx.doi.org/10.1126/science.1185563. Riddell Pearce, F.C., Couvillon, M.J., Ratnieks, F.L.W., 2013. Hive relocation does not adversely affect honey bee (hymenoptera: Apidae) foraging. Psyche (New York) http://dx.doi.org/10.1155/2013/693856. Thornton, B.J., Elthon, T.E., Cerny, R.L., Siegfried, B.D., 2010. Proteomic analysis of atrazine exposure in Drosophila melanogaster (Diptera: Drosophilidae). Chemosphere 81: 235–241. http://dx.doi.org/10.1016/j.chemosphere.2010.06.032. Valko, M., Morris, H., Cronin, M.T.D., 2005. Metals, toxicity and oxidative stress. Curr. Top. Med. Chem. 12:1161–1208. http://dx.doi.org/10.2174/0929867053764635.