Clothianidin treatment, milkweed growth, and chemical analysis

A clothianidin standard (purity 99.9%; MDL # MFCD06200753, Sigma-Aldrich, St. Louis, Missouri, United States of America) was used to make stock solutions that were then diluted with distilled water to dose commercial soil (LA4 Sunshine Loosefill, Sungro Horticulture, Massachusetts, United States of America) at concentrations of 15 ng/g of soil (hereafter “low dose”) and 25 ng/g of soil (hereafter “high dose”). The two treatment levels were chosen to represent field-realistic values recorded in Ontario, Canada and are established standard sublethal doses (Chan et al. Reference Chan, Prosser, Rodríguez-Gill and Raine2019).

Swamp milkweed (Asclepias incarnata) was grown from seed in soil from either the low-dose or high-dose treatments or from a control (no clothianidin). A total of 256 milkweed plants per treatment were grown at a density of four plants per 1.68-L pot (total 64 pots per treatment) in environmental chambers at the University of Guelph Phytotron (University of Guelph, Guelph, Ontario, Canada). Temperature was maintained at 29 °C during the day and 23 °C at night, with a light intensity of 11 914–16 280 lx (18 hours light : 6 hours dark), and 60% relative humidity, based on conditions outlined in Flockhart et al. (Reference Flockhart, Martin and Norris2012). Plants were watered daily with reverse-osmosis water and fertilised weekly with Plant-Prod Solutions fertiliser 17:5:17 NPK (Master Plant-Prod Inc., Brampton, Ontario, Canada). Amblyseius swirskii (Mesostigmata: Phytoseiidae) was introduced as a biocontrol (Bioline AgroSciences Swirskiline Biocontrol Agent and Biobest Swirskii-Breeding-System; Bioline AgroSciences, Oxnard, California, United States of America) to reduce the impact of thrips (Thysanoptera) (Flockhart et al. Reference Flockhart, Martin and Norris2012).

To examine temporal variation in clothianidin concentrations throughout the experiment, a minimum 15 g of soil (sensitivity ± 1.0 g; MyWeigh iBalance i500; HBI Technologies Canada, Vancouver, British Columbia, Canada) was collected for analysis in sterile polypropylene centrifuge tubes (VWR High-Performance Centrifuge Tubes, CAT # 89039-656; VWR International LLC, Mississauga, Ontario, Canada) when the soil was dosed, 14 days after dosing, 18 days after dosing when monarch eggs were transferred to the treatment leaves (see below), and 32 days after dosing when monarchs pupated. A single leaf was randomly selected from each milkweed plant at 18 and 32 days after the soil was dosed. These leaves were then combined to reach a minimum mass of 2.0 g required for clothianidin detection and stored in sterile polypropylene centrifuge tubes. To determine if clothianidin was metabolised before metamorphosis, a subset of fifth-instar caterpillars was randomly selected and combined ( ${\overline x}$ = 2.4 caterpillars per sample tube) to reach the minimum mass required for analysis. At the completion of the experiment, adult monarch butterflies were combined ( ${\overline x}$ = 5.2 butterflies per sample tube) and submitted for clothianidin analysis. Before analysis for clothianidin concentration, all samples were stored at –20 ºC at the University of Guelph Agriculture and Food Laboratory. Analysis was done using the QuEChERS (quick, easy, cheap, effective, rugged, and safe) method, which is appropriate for samples with high water content (Perestrelo et al. Reference Perestrelo, Silva, Porto-Figueira, Pereira, Silva, Medina and Câmara2019). In brief, a sample of the soil or tissue was extracted and placed in a solution of 1% acetic acid in acetonitrile with anhydrous sodium and magnesium sulphate. The precipitate is then diluted with methanol and 0.1 mol ammonium acetate. High-performance liquid chromatography/electrospray ionisation–tandem mass spectrometry and gas chromatography–tandem mass spectrometry were used to assess concentration (Canadian Food Inspection Agency 2008; Wang and Daniel Reference Wang and Daniel2009), which was returned in parts per billion (1 ppb = 1 ng/g; Boguski Reference Boguski2006). The limit of quantification is the lowest concentration that can be accurately quantified, whereas the limit of detection is the lowest concentration that can be distinguished from the assay background and therefore has a higher degree of error. The limit of quantification and limit of detection depend on the sample type (soil: limit of quantification/limit of detection 20 ppb/7 ppb; leaf: 30 ppb/10 ppb; monarch tissue: 2 ppb/0.7 ppb).

Monarch capture and rearing

Caterpillars and butterflies used during the experiments were reared from eggs laid by wild monarchs obtained from properties in Long Point, Ontario, Canada (42.58° N, 80.43° W; ♂ n = 2, ♀ n = 4) and the Guelph Lake Conservation Area, Guelph, Ontario (43.61° N, 80.26° W; ♂ n = 1, ♀ n = 5) that were not treated with clothianidin or any other neonicotinoid. Butterflies were held in coin envelopes (6.35 cm × 10.8 cm) inside an animal carrier kept at ambient temperature and lined with a damp cloth to maintain humidity to avoid drying of the wings and limit mortality during transport to the University of Guelph. Upon arrival, butterflies were weighed (Denver Instrument PI-602 scale; Denver Instrument, Bohemia, New York, United States of America) to the nearest 0.01 g and hand-fed a 10% honey–water solution daily until satiation. All monarchs were tagged with a small, uniquely numbered, adhesive sticker (approximately 0.65 cm) on the left hindwing near the abdomen that allowed us to visually identify individuals and to monitor body condition. Breeding took place in large mesh enclosures (60 cm height × 60 cm depth × 60 cm width) inside an incubator set at temperatures fluctuating between 29 °C and 23 °C with a light intensity of 11 914–16 280 lx (18 hours light : 6 hours dark) and 60% relative humidity to simulate natural conditions common during summer breeding. Each enclosure was outfitted with untreated milkweed (i.e., not exposed to clothianidin) and an artificial nectar source (i.e., sucrose water), changed daily, until all eggs were collected and the reproductive monarchs were released.

We removed 201 eggs (n = 67 per treatment) by gently rubbing the eggs off the milkweed leaf with a fine-tipped paintbrush onto a damp cloth before adhering them to the milkweed leaf using residual latex. Leaves with eggs were placed in large plastic containers, enclosed using a finely perforated mosquito netting (Bulk Mosquito Netting, CAT # 09A04.73, Lee Valley, Ottawa, Ontario, Canada), and cleaned daily. Caterpillars were fed experimental milkweed (i.e., control, 15 ng/g, or 25 ng/g clothianidin) ad libitum to minimise risk of nutritional stress from food limitation. Light cycle and temperature mimicked ambient conditions during the early breeding season in Guelph, Ontario (43.5° N, 80.2° W; 16 hours light : 8 hours dark; Flockhart et al. Reference Flockhart, Martin and Norris2012) at 28 °C during the day, 22 °C at night, and 60% relative humidity to reduce the risk of monarchs entering reproductive diapause (Goehring and Oberhauser Reference Goehring and Oberhauser2002; Flockhart et al. Reference Flockhart, Martin and Norris2012).

Given observed differences in the size of caterpillars reared on milkweed grown in clothianidin-treated soil, we collected body measurements at three timepoints at instar stages that could be reliably identified. Using digital calipers to the nearest 0.01 mm, we recorded body length measurements at the first, third, and fifth instars and body width measurements at the third and fifth instars. Previous methods to measure body volume involved haemolymph extraction (Lin et al. Reference Lin, Slate, Paetsch, Dorfmann and Trimmer2011). As a nonlethal measure of body volume, we calculated the volume of a cylinder (V = π × r ² × h) to estimate the body volume of the third and fifth instars, where r is half the body width and h is the body length. Monarch caterpillars were randomly sampled for clothianidin analysis (see the section, Clothianidin treatment, milkweed growth, and chemical analysis, this paper), and in the fifth instar, we recorded the mass to the nearest 0.1 g for 117 caterpillars (control: n = 41; low dose: n = 41; high dose: n = 35). Early death was recorded for 9 caterpillars (control: n = 1; low dose: n = 2; high dose: n = 6) between 18 and 26 June 2018. Chrysalids were transferred to mesh enclosures (120 cm × 120 cm × 120 cm; Popadome Plant Dome, CAT # XC515, Lee Valley, Ottawa, Ontario, Canada) within two days of pupation until eclosion, after which adult monarchs were fed ad libitum a sucrose solution from dishes placed at the bottom of the enclosures (Flockhart et al. Reference Flockhart, Martin and Norris2012). All monarchs were measured and weighed in captivity and examined for Ophryocystis elektroscirrha parasites (Neogregarinorida: Ophryocystidae) by applying clear tape to the abdomen and analysing tape for spores under a microscope at 400 × (Altizer and Oberhauser Reference Altizer and Oberhauser1999). Individuals that tested positive for O. elektroscirrha were removed from the study and euthanised due to the potential for developmental deformities (control: n = 4; high dose: n = 1). All procedures were conducted under the Ontario Ministry for Natural Resources Wildlife Scientific Collectors Permit (#1090000).

Assessing reproductive output

To assess the influence of clothianidin exposure on adult female monarch butterfly reproductive output, a subset of female monarchs reared from caterpillars was provided untreated milkweed as substrate for oviposition in mesh enclosures (40.6 cm × 63.5 cm) and fed a 10% honey–water solution. Virgin reproductive pairs (control: n = 7; low dose: n = 8; high dose: n = 7) were selected because prior matings can reduce the size of subsequent spermatophores being transferred (Oberhauser Reference Oberhauser1988). Once mated, males were removed, and females remained in the enclosures to complete oviposition. We removed female monarchs when they died or after egg deposition had ceased for a maximum of seven days (Oberhauser Reference Oberhauser1997). For all females, we counted the number of eggs deposited, and we stored dead female butterflies for approximately two weeks until we could count the number of mature oocytes remaining in the ovarioles (Oberhauser Reference Oberhauser1997). Due to the small size of the eggs, we calculated the average mass in groups of 10 to the nearest ± 0.001 mg (XP26 Micro Balance; Mettler-Toledo Inc., Mississauga, Ontario, Canada) and measured the height and width to the nearest 0.01 mm using handheld digital calipers. Damaged eggs were included in the total count of the number of eggs deposited but were not measured. The small size of monarch eggs did not permit digital measurement, so we followed the procedures outlined in García-Barros (Reference García-Barros2000) to calculate the egg size, despite deviations from spherical shape potentially inducing slight measurement errors. Therefore, the cubic root of the volume of an ellipsoid,

$${\rm{Egg \;size }} = {\rm{ }}{(0.5236 \times {d^2} \times h)^{1/3}},$$

was used to estimate monarch egg size, where d equals egg diameter and h equals egg height (García-Barros Reference García-Barros2000).

Statistical analyses

Outliers, potentially as a result of measurement or transcription error, were identified using the outliers package in R, version 3.4.1(R Core Team 2015b) and were removed from analysis. Four outliers were identified for body length (control: n = 1; low dose: n = 2; high dose: n = 1) and two for body width (control: n = 2). To test for the effect of clothianidin on caterpillar body length, body width, and body volume, we applied series of generalised linear mixed models using the nlme package in R, version 3.4.1 (Bolker et al. Reference Bolker, Brooks, Clark, Geange, Poulsen, Stevens and White2009; R Core Team 2015b). The models included instar stage (first, third, fifth for length, and third and fifth for width and volume), treatment (i.e., control, low dose, and high dose), the interaction between stage and treatment as fixed effects, and individual identity as a random effect. We then separately ran a post-hoc Tukey honestly significant difference using the stats package (R Core Team 2015a) to determine if there was a difference in caterpillar body length, body width, and body volume between treatments for each instar. Lastly, we used a general linear model to test whether treatment affected the mass, because only fifth-instar caterpillars were measured for mass.

A general linear model was used to assess whether the clothianidin treatment affected adult monarch butterfly size. Forewing length was square-root transformed to improve the normality of the distribution and was included as a response variable to estimate body size (Miller Reference Miller1977, Reference Miller1991). Date of eclosion, mass, and sex were also included as predictors. We also used a generalised linear model with Poisson distribution and log link function to determine whether treatment affected the total number of eggs laid. The total number of eggs laid was included as the response variable, with treatment, forewing length, mass, and age as predictors. Age of the male monarch in the mated pair was included as a proxy for spermatophore size and the amount of resources transferred to the female during mating that may contribute to egg production (Oberhauser Reference Oberhauser1997).

We applied a series of generalised linear mixed models to determine if early clothianidin exposure affected the size of eggs. We used the average egg size and mass as response variables in separate models, with treatment and total number of eggs laid as fixed effects. To account for potential differences in resource investment by older and larger female monarch butterflies (Oberhauser Reference Oberhauser1997), we included age, mass, and forewing length as fixed effects, as well as the age of the male monarch in the mated pair to account for the male contribution to the overall resource budget (Oberhauser Reference Oberhauser1997). Female identity was included as a random effect. For all models, each possible combination of parameters was tested in a separate model, as well as in a null model that included only the intercept. Akaike’s information criterion (AIC) or an AIC corrected for small sample sizes (AIC_c) was then used to rank models, and those with a ΔAIC or ΔAIC_c equalling or less than two were considered to have support (Burnham and Anderson Reference Burnham and Anderson2002). Support for a model was also provided by the log likelihood and Akaike weights (ω_i).