Neonicotinoid pesticides exert metabolic effects on avian pollinators

All experimental procedures were approved by the University of Toronto animal care committee (Animal Use Protocol number 20012112) and conformed to guidelines prescribed by the Canadian Council on Animal Care.

Animal capture and husbandry

Wild male ruby-throated hummingbirds (Archilochus colubris; (n = 23); mass range during experimental period: 2.59 g to 4.52 g), were caught on the University of Toronto Scarborough Campus (({43.7838}^{circ }hbox {N}), ({79.1875}^{circ }hbox {W})) or the University of Western Ontario campus (({43.0096}^{circ }hbox {N}), ({81.2737}^{circ }hbox {W})) using box traps modified with hook-and-loop fastener tape on a drop door containing hummingbird feeders. Birds were trapped between 06:00 h and 12:00 h during the months of May through September of 2017, 2018, and 2019. Pilot trials were conducted in 03/2018. Birds in the pilot study were on wintering/migratory seasonality with a 12 h daylight schedule. Subsequent trials were conducted between 04/2019 and 01/2020. In 04/2019, birds were under breeding seasonality (14 h daylight) and in 01/2020, birds were under wintering/migratory seasonality (12 h daylight) during experimental trials. The daylight schedule approximated the photoperiod encountered as part of annual migrations to Central America and back. Upon capture, hummingbirds were quickly transported to metal EuroCages (({50.8 times 91.5 times 53.7},hbox {cm}) ((hbox {L}times hbox {W}times hbox {H}))) at the animal care facility where they were housed individually and acclimated to feed from syringe feeders. Birds were provided an 18 % (w/v) Nektar Plus (Guenter Enderle, Tarpon Springs, FL, USA) solution (henceforth referred to as maintenance diet), which was consumed ad libitum, and syringes were replaced daily (range of average consumption of daily maintenance diet during study period was 5.4 mL13.2 mL).

Experimental design

Birds drank solutions of imidacloprid (IMI; Sigma-Aldrich Cat. No. 37894) dissolved in a 20% w/v sucrose solution and were randomly assigned to either control (({0.0},upmu hbox {g g}^{-1}cdot)BW), low (({1.0}upmu hbox {g g}^{-1})), middle (({2.0}upmu hbox {g g}^{-1})), or high dose (({2.5}upmu hbox {g g}^{-1})) groups (n = 7, 4, 8, 4, respectively). Stock solution concentrations were analytically confirmed (low: ({0.32},hbox {gL}^{-1}), middle: ({0.59},hbox {gL}^{-1}), high: ({0.78},hbox {gL}^{-1})) such that a 3 g bird dosed with ({10},upmu hbox {L}) of solution would receive the dosage rate corresponding to either the low, middle or high dose. The volume of imidacloprid stock solution used for dosage was adjusted on a body weight (BW) basis, pipetting from the stock solution into a new nectar syringe and drawing up to a final volume of ({50},upmu hbox {L}) with 20% w/v sucrose solution, ensuring that birds received the same dosage rate throughout the trial. Birds were deprived of their regular nectar solution for 10 min to 15 min consumed the entire small-volume dosing solution within 10 min of being offered the solution. The dose was considered to be delivered when there was no visible solution remaining in the transparent syringe.

Doses were established within a range spanning expected exposure in a bird drinking ({10},hbox {mLd}^{-1}) from contaminated flowers¹⁰ up to 10 % of the LD50 in canaries⁶¹ (Serinus canaria, LD50: ({25},upmu hbox {g g}^{-1}) to 50 (upmu hbox {g g}^{-1})), similarly small birds with fast metabolic rates to target a sub-lethal concentration expected to produce toxic effects³². When energy demands are high, hummingbirds may consume over three times their body weight in nectar⁶³, therefore ({10},hbox {mLd}^{-1}) is a probable figure for contaminated nectar consumption. Pooled blueberry flower samples collected about 1 year after treatment with imidacloprid contained the neonicotinoids at a concentration of ({5.16},hbox {ng g}^{-1})¹⁰. We extrapolated our very low and low dose concentrations based on these data. We stipulate that given the flower sample is a pooled sample, it was collected from flowers long after treatment, and there are different regulations on pesticide use within the ruby-throated hummingbird’s range, these doses were environmentally relevant.

We tested multiple intermediate doses which allowed us to explore dose-response relationships in observed effects⁶⁴. Pilot experiment data with control, very low (({0.2},upmu hbox {g g}^{-1})), or high dose (({2.5},upmu hbox {g g}^{-1})) (n = 3 per group) are included for cholinesterase activity and toxicokinetic elimination analyses. Other metrics including behaviour and energy expenditure were not included from the pilot study due to differences in data collection protocols and are not strictly comparable. Behavioural data collection, cloacal fluid (CF) collection, and respirometry occurred over 6 days, where pre-dose data were collected for each animal on days 1 through 3, and dosing occurred once per day at 11:00 on days 4 through 6. Body weight measurements were taken daily at 10:00. The body weights of birds on the first day of experimentation ranged from 2.70 g to 4.52 g. For simplicity, 11:00 on days 1-6 is referred to as Dose Time (DT). Terminal sampling and tissue collection occurred 24 h after the third dose was administered. Birds were sacrificed by decapitation following isoflurane overdose, and whole blood, flight muscle, liver, brain, and heart tissues were rapidly excised, flash frozen in liquid nitrogen, and stored at ({-80},^{circ }hbox {C}) until downstream analysis, except in the case of blood which was immediately used for blood smear preparation.

Respirometry

Oxygen consumption and carbon dioxide production rates were measured using open-flow chamber respirometry⁶⁵. Airflow through three metabolic chambers and one empty reference chamber was maintained at a rate of ({300},hbox {mL min}^{-1}). Excurrent air from the chambers was sub-sampled at ({100},hbox {mL min}^{-1}) sequentially starting with the reference chamber at using a Turbofox-5 (Sable Systems International Las Vegas, NV, USA). Sub-sampled air was passed through a water vapour pressure analyzer, a drying column (Indicating Drierite, W.A. Hammond Drierite, Xenia, Ohio, USA), carbon dioxide meter, and finally an oxygen analyzer (Turbofox-5, Sable Systems International). The oxygen and carbon dioxide analyzers were calibrated according to manufacturer instructions using well-mixed ambient air for the oxygen analyzer, and zero and (0.25 ,%hbox {CO}_{2}) reference gases for the (hbox {CO}_2) analyzer. Respirometry data were recorded at a frequency of 1 Hz using Expedata software (v. 1.84, Sable Systems) for 5 min while sampling from the empty reference chamber, followed by three 7 min recording periods from each of the chambers holding a bird. After this 26 min period, sub-sampling was resumed from the reference chamber for another 5 min followed by another 7 min sub-sampling period from experimental chambers. A final 5 min sampling of the reference chamber concluded the respirometry data collection, and birds were returned to the cloacal fluid collection chambers approximately 60 min after initially being placed in respirometry chambers.

Behavioural data collection and processing

Video recordings of birds were collected for 2 h, starting 4 h after DT (15:00–17:00). At the start of the recording period, birds were returned to their home cages where they could feed ad libitum by hovering and tracking a syringe on a 10 cm arm oscillating through a ({90}^{circ }) range along a lateral arc at a speed of 15 RPM. Video recordings were analyzed for time spent in flight, subdivided into foraging and non-foraging flights. Foraging flights were defined as flights where the bird contacted the hover feeder with their bill. Total consumption of the maintenance diet over this 2 h period was recorded.

Heterophil/lymphocyte ratios

Approximately ({2},upmu hbox {L}) of blood was collected for blood smear preparation immediately following sacrifice. After smearing, slides were left to air dry for a minimum of 3 h before fixing with 100 % methanol and staining with Giemsa–Wright solution (Fisher Scientific Cat. No. 123869). Slides were stained by immersion in eosinophilic dye (5times 1,hbox {s}) followed by (5times {1},hbox {s}) in basophilic dye.

Cholinesterase activity assay

Brain and muscle tissues were homogenized using a sonic dismembrator ((hbox {Fisherbrand}^{mathrm{TM}}) Model 120 Sonic Dismembrator) 1:10 w:v with ice-cold 0.1 M potassium phosphate buffer (pH 7.2). Samples were centrifuged at 10,000 RPM in a Beckman Coulter microfuge 22R centrifuge held at ({4},^{circ }hbox {C}) for 5 min. Total protein concentrations in tissue homogenates were determined by the Bradford assay (Sigma-Aldrich Cat. No. B6916). Cholinesterase activity was measured by the Ellman method adapted for a microplate reader (BioTek Synergy HT)⁶⁶. Optimal assay conditions were 0.1 M potassium phosphate buffer (pH 7.2), 0.48 mM acetylcholine, 0.64 mM DTNB (Sigma-Aldrich Cat. No. D8130), 1.1 mM sodium bicarbonate. Assays were initiated through the addition of acetylcholine (Sigma-Aldrich, Cat. No. 01480) in a total volume of ({300},upmu hbox {L}). Absorbance was read at 412 nm every 2.5 min for 10 min.

Cloacal fluid

Collection

Cloacal fluid was collected for 1 h at 3 time points each day according to one of two schedules: starting (1) 1 h, 6 h, and 23 h, or (2) 2.5 h, 6.5 h, and 23 h after DT. Cloacal fluid was collected according to schedule (1) in pilot experiments, and (2) in the subsequent trials. A watch glass was placed beneath birds perching in 10 cm W (times) 12 cm H glass cylinder enclosures stopped with 19-gauge galvanized 1 cm hardware mesh openings in order to obtain cloacal fluid. To encourage greater cloacal fluid production, and to simulate the regular feeding behaviour of wild birds, individuals fed ad libitum from a syringe containing a 20% (w/v) sucrose solutions every 5 min to 10 min for the duration of cloacal fluid collection, which took place over 1 h as described under Sect. 4.2. After the collection period, cloacal fluid samples were stored at ({-20},^{circ }hbox {C}) until pooling and refreezing prior to chemical analysis.

Chemical analyses

Cloacal fluid samples and dosing solutions were analyzed for IMI by HPLC-ESI-MS/MS by Laboratory Services, NWRC (National Wildlife Research Centre, Ottawa, ON, Canada). Cloacal fluid samples were pooled by time point across dosing days by individual to reach the necessary minimum volume of ({100},upmu hbox {L}).

Cloacal fluid sample pools from 2018 trials were thawed at room temperature. Each pool was diluted 4 (times) with DI water (({25},upmu hbox {L}) cloacal fluid + ({75},upmu hbox {L}) DI water). The resulting ({100},upmu hbox {L}) diluted samples were then spiked with ({100},upmu hbox {L}) of internal standard (IS) solution. Spiked samples were filtered directly into ({300},upmu hbox {L}) glass inserts using 4 mm PVDF ({0.45},upmu text {m}) Millex filters and ({50},upmu hbox {L}) aliquots were injected. For the 2018 analyses, the minimum detection limit (MDL) and minimum reporting limit (MRL) were ({0.204},hbox {ng}hbox { mL}^{-1}) and ({0.616},hbox {ng}hbox { mL}^{-1}) respectively.

Cloacal fluid sample pools from 2019 trials were thawed at room temperature and ({50},upmu hbox {L}) of IS solution was added to (200,upmu text {L}) of pooled cloacal fluid. In cases where the sample volume was too small, volumes were adjusted: ({100},upmu hbox {L}) cloacal fluid + ({25},upmu hbox {L}) IS or ({80},upmu hbox {L}) cloacal fluid + ({20},upmu hbox {L}) IS as required. In these cases, duplicate injections of ({50},upmu hbox {L}) were not possible. All samples were filtered with 4 mm PVDF ({0.45},upmu text {m}) Millex filters prior to injection. For the 2019 analysis, the MDL and MRL were ({0.051},hbox {ng}hbox { mL}^{-1}) and ({0.154},hbox {ng}hbox { mL}^{-1}) respectively.

Cloacal fluid sample pools and dosing solutions were analyzed according to modifications to the methods of Main et al.⁹. Briefly, IMI in a cloacal fluid or DI water matrix was quantified by the internal standard method using the API5000 Triple Quadropole Mass Spectrometer (AB Sciex) and the TurboSpray ion source in positive polarity. The calibration curve was constructed from 8 concentrations ranging from ({0.1},hbox {ng}hbox { mL}^{-1}) to ({20},hbox {ng}hbox { mL}^{-1}) yielding an R greater than 0.995 (linear regression, no weighting). Injection cross-contamination was monitored by injecting solvent blanks (water:acetonitrile 80:20) before and after each set of samples. Contamination was also monitored by using a DI water sample blank spiked at ({20},hbox {ng}hbox { mL}^{-1}) IMI. In all cases, no IMI above MDLs was detected. Method precision was evaluated by duplicate injections and/or duplicate dilutions: the RPDs (relative percent differences) were all less than 15 %, demonstrating good method precision. Method accuracy was evaluated by analyzing a ({20},hbox {ng}hbox { mL}^{-1}) QC spike per set: the recoveries ranged between 96 % and 106 %, demonstrating good method accuracy.

Statistical analyses

All statistical analyses were conducted in R version 3.5.2⁶⁷. Birds exhibiting weight loss outside the lower bound of the 95 % confidence interval (CI) of 20 % over the study period ((n=3)) were omitted and data were reanalyzed. Birds exhibiting extreme weight loss across the pre-dose and post-dose conditions were in the control group ((n=2)) and the low dose group ((n=1)) suggesting a adverse response to the experimental period rather than the treatment itself. Data are presented as mean ± standard error. Significance ((p<0.05)) was determined in the DRC package by comparison of the fitted model to a simple linear regression with a slope of 0. In cases where a dose-response model was not a better fit to the data than a simple linear regression, as determined by the lack-of-fit test in the DRC package in R, a linear model was used. In cases where a linear model was used, omega-squared estimates of effect sizes and their confidence intervals where calculated using the effectsize package in R⁶⁸.

Analyses and modeling of metabolic rate data

Instantaneous (hbox {O}_2) consumption rate (({dot{V}}_{hbox {O}_{2}}), in (hbox {mL}cdot hbox {O}_2cdot hbox {min}^{-1})) and the respiratory exchange ratio of birds (RER; defined as the ratio between ({dot{V}}_{text {CO}_{2}}) and ({dot{V}}_{text {O}_{2}})) were determined across the 7 min dwell from chamber air sampled at 1 Hz with a flow rate of ({300},hbox {mL min}^{-1}), using standard equations. Data were then converted to a metabolic rate in (hbox {J min}^{-1}) by applying the following oxyjoule equivalency⁶⁹:

Instantaneous metabolic rate data (in (hbox {J min}^{-1})) were integrated over the duration of the dwell to calculate total energy expended during the dwell (in J). Assuming that hummingbirds were more likely to have been stressed immediately after handling and placement into respirometry chambers, we discarded first dwell data (collected within 30 min; Sect. 4.3) and included second dwells (collected between 30 min to 60 min; Sect. 4.3) for comparison only. Energy expenditure for each bird was normalized by dividing mean energy expenditure across post-dose days by mean energy expenditure across pre-dose days. Normalized mean energy expenditure in J was modeled across dosing groups by an asymmetric, alternative parameterization of the 3 parameter Weibull dose-response model (Weibull type 2)^70,71. The model was fitted using the DRC package with the general form⁷¹:

where it is assumed that (lim _{xrightarrow infty } f(x)=0). The biological interpretation of this theoretical limit assumes that as the concentration of IMI increases indefinitely, the metabolic rates of the birds would approach 0, as they would be dead. Parameter a characterizes the mean energy expenditure of birds unexposed to IMI (dose(= 0)). Parameter b relates to the LC50 and characterizes location and steepness of the upper shoulder of the dose-response. The sigmoidal model is asymmetric about the inflection point characterized by parameter c. Significance ((p<0.05)) was determined in the DRC package by comparison of the fitted model to a simple linear regression with a slope of 0, indicating no effect of dose on energy expenditure.

Analyses of behavioural assay

The number of flight instances, amount of time feeding, and whether feeding occurred during a flight instance were processed from each 2 h set of videos. The change in mean time in flight on each day for each individual between pre-dose and post-dose conditions was calculated for foraging and non-foraging flights. Linear models of foraging and non-foraging flight time as a function of dose were analyzed by one-way ANOVA, with an alpha level set at 0.05. The total number of foraging and non-foraging flight, average consumption of maintenance diet per instance of flight where birds engaged in feeding, and average duration of foraging and non-foraging flight were also compared among groups in this manner.

Analyses of heterophil/lymphocyte ratios

Manual 100-cell differential leukocyte counts were conducted on smears under (1000times) magnification. Fields of view were excluded from differential counts if cells did not form a monolayer or if thrombocyte aggregates were present. Differential leukocyte counts of blood smears were duplicated for each slide by 2 individuals where heterophils, eosinophils, monocytes and lymphocytes were tallied up to 100. Both individuals were blind to the dose groupings. The ratio of heterophils to lymphocytes was then calculated and the mean of the two duplicate ratios was used for analyses. Intraclass correlations for duplicate readings was calculated at 0.75 with upper and lower 95% confidence bounds of 0.90 and 0.45 using the DescTools package⁷². A linear model of heterophil/lymphocyte ratios as a function of dose were analyzed by one-way ANOVA, with an alpha level set at 0.05.

Analyses of cholinesterase activity

The specific activity of cholinesterase was modeled by linear regression in brain and muscle tissues. (Delta) (difference in) absorbance values were calculated from absorbance 5 min to 10 min after the reaction was initiated. Linear models of cholinesterase activity as a function of dose were analyzed by one-way ANOVA, with an alpha level set at 0.05.

Toxicokinetic modeling

For the general first-order toxicokinetic excretion model,

where [U] is the concentration of IMI in cloacal fluid at time x in hours, model parameters a and k represent the excretion coefficient and elimination rate constant of unmetabolized IMI in cloacal fluid, respectively. Convergence on parameter estimates was achieved by nonlinear least-squares regression, and 95% CIs for excretion models were determined by bootstrap resampling in the nlstools package⁴⁵. Elimination half-life of imidacloprid was calculated according to the following equation:

Source: Ecology - nature.com