Imidacloprid exposure is detectable in over one third of wild bird samples from diverse Texas ecoregions

Abstract

Avian decline is occurring globally with neonicotinoid insecticides poised as a potentially contributing factor. Birds can be exposed to neonicotinoids through coated seeds, soil, water, and insects, and experimentally exposed birds can show varied adverse effects including mortality and disruption of immune, reproductive, and migration physiology. However, few studies have characterized exposure in wild bird communities temporally. We hypothesized that neonicotinoid exposure would vary temporally and based on avian ecological traits. Birds were banded and blood sampled at eight non-agricultural sites across four Texas counties. Plasma from 55 species across 17 avian families was analyzed for the presence of 7 neonicotinoids using high performance liquid chromatography-tandem mass spectrometry. Imidacloprid was detected in 36% of samples (n=294); this included quantifiable concentrations (12%; 10.8– 36,131 pg/mL) and concentrations that were below the limit of quantification (25%). Additionally, two birds were exposed to imidacloprid, acetamiprid (18971.3 and 6844pg/mL) and thiacloprid (7022.2 and 17,367 pg/mL), whereas no bird tested positive for clothianidin, dinotefuran, nitenpyram, or thiamethoxam, likely reflecting higher limits of detection for all compounds compared to imidacloprid. Birds sampled in spring and fall had higher incidences of exposure than those sampled in summer or winter. Subadult birds had higher incidences of exposure than adult birds. Amongst the species for which we tested more than five samples, American robin (Turdus migratorius) and red-winged blackbird (Agelaius phoeniceus) had significantly higher incidences of exposure. We found no relationships between exposure and foraging guild or avian family, suggesting birds with diverse life histories and taxonomies are at risk. Of seven birds resampled over time, six showed neonicotinoid exposure at least once with three showing exposures at multiple time points, indicating continued exposure. This study provides exposure data to inform ecological risk assessment of neonicotinoids and avian conservation efforts.

1.1. Introduction

In 1991, imidacloprid, the first of a new class of pesticides called neonicotinoids, was introduced (Kitsiou et al., 2009); neonicotinoids have since become the most common and widely used insecticides in the world (Klingelhöfer et al., 2022). Neonicotinoids are often applied in the form of seed coatings but applications also include foliar sprays, trunk injections, and soil drench applications. Neonicotinoids provide systemic protection throughout plant tissues by virtue of their high water solubility (van der Sluijs et al., 2015). Coupled with repeated applications and long half-lives, neonicotinoids can accumulate and persist in soil, water, and particular habitats like wetlands. (Bonmatin et al., 2015; Hladik et al., 2018; Huang et al., 2020; Jones et al., 2014; Main et al., 2014).

Neonicotinoids are thought to be a major driver of insect biodiversity decline and biomass loss (Barmentlo et al., 2021; Sánchez-Bayo and Wyckhuys, 2019; van der Sluijs, 2020). Neonicotinoids may also have conservation implications for two other invertebrates, earthworms (van Loon et al., 2022) and spiders (Řezáč et al., 2019), which can be important sources of food for birds (Nyffeler et al., 2018). An estimated 96% of terrestrial North American birds feed insects to their young (Tallamy and Shriver, 2021) and Losey and Vaughan (2006) found that all North American passerines they surveyed were at least partly insectivorous. Though compensatory diet-shifts have been demonstrated in insect-eating birds, these shifts may not mitigate population declines (e.g. (Sitko and Heneberg, 2020; Trevelline et al., 2018). Thus decreasing insect populations can impact many bird species that depend on insects as prey (Bowler et al., 2019; Hallmann et al., 2014; Nebel et al., 2010). Widespread use of pesticides, in particular neonicotinoids, has been linked to parallel declines in insects and birds (Benton et al., 2002; Goulson, 2014; Mineau and Whiteside, 2013; Møller, 2019), reduction of which on the landscape can have detrimental impacts on ecosystem functioning (Anderson et al., 2011). Birds may ingest neonicotinoids from the environment via insects (Botías et al., 2017), seeds (Roy et al., 2019; Roy and Coy, 2020), water (Morrissey et al., 2015), nectar (Bishop et al., 2018; Graves et al., 2019), soil (Douglas et al., 2015), dust, and through skin or feathers during preening (Gibbons et al., 2015; Goulson, 2013; Mineau, 2011; Vyas et al., 2007).

Avian decline is pervasive across many families and ecosystems globally (Gaston et al., 2003) – cumulatively, over 3 billion birds have been lost from North America alone since 1970 (Rosenberg et al., 2019). The role of pesticides in this ongoing avian decline has not been fully explored. In addition to a loss of available prey, birds also face direct mortality from ingestion of neonicotinoids (Millot et al., 2017) and a number of adverse effects that may ultimately decrease fitness or survival. In experimental studies with red-legged partridges (Alectoris rufa), dosing birds with sublethal quantities of imidacloprid-coated seeds produced distinct blood chemistry changes, carotenoid-based color changes, delayed egg laying, decreased clutch sizes, depressed T-cell immunity of chicks, and decreased chick survival (Lopez-Antia et al., 2015). Imidacloprid toxicity has also been shown to have adverse effects on the humoral immune system (Kammon et al., 2012) and liver (Balani et al., 2011) in chickens. Some hypothesize that widespread immunosuppression from environmental exposure to neonicotinoids could be contributing to increased incidences of disease outbreaks in wildlife (Mason et al., 2013). A study of captive white-crowned sparrows (Zonotrichia leucophrys) found that ingestion of imidacloprid-coated seeds caused severe loss of body mass and temporarily impaired orientation during migration (Eng et al., 2017). In a later study of geo-tagged white-crowned sparrows experimentally dosed with imidacloprid during migration, birds showed decreased food consumption, decreased mass and fat, and delayed departure timing (Eng et al., 2019). Given that birds must rapidly gain mass and fat in order to complete the sustained flights required during migration (Bairlein and Gwinner, 1994; Jenni and Jenni-Eiermann, 1998), clinical effects of neonicotinoids may be especially impactful for migratory birds, though any decline of immune, reproductive, and neurological function are problematic for resident and migrant birds alike. A number of studies have detected neonicotinoids in focal species of wild-living birds (Byholm et al., 2018; Hao et al., 2018; Lennon et al., 2020; Taliansky-Chamudis et al., 2017), but few studies have explored the exposure of broad communities of wild birds across time, including assessments of individuals over time. This is a critical precursor for identifying the birds most at risk of neonicotinoid-related fitness or survival effects. The objective of this study was to measure acute neonicotinoid exposure longitudinally in wild bird communities across four Texas counties. We hypothesized that there would be variation in avian exposure to neonicotinoids across bird families, species, and ages; across seasons; and across avian life history traits. We predicted that spring and summer would be peak times for neonicotinoid exposure in wild birds due to increased agricultural activity and crop planting at these times (“Crop Information- Planting & Harvesting,” 2022). We predicted that insectivores and granivores would have increased exposure to neonicotinoids through their diets (Li et al., 2020; Roy et al., 2019), and that migrant birds would be at higher risk for exposure than residents given that they must consume large quantities of food in unfamiliar habitats during stopover (Klaassen et al., 2012) at times that coincide with crop planting (“Crop Information- Planting & Harvesting,” 2022). Finally, we predicted that neonicotinoid-exposed birds would be in poorer body condition than unexposed birds, since weight loss can be an effect of sublethal neonicotinoid exposure (Eng et al., 2019; Lopez-Antia et al., 2015).

2. Materials and Methods

2.1. Bird Capture and Sampling

Field studies were carried out at eight sites in four counties in Texas (Fig. 1): Blanco, Brazos, Cameron, and Walker. Each county is in a different ecoregion of the state- Edwards Plateau (Blanco), Oak Woods and Prairies (Brazos), South Texas Brush Country (Cameron), and Piney Woods (Walker) (Wilkins et al., 2003). Brazos and Cameron Counties each contained three sites. In Brazos County: Ecological and Natural Resource Teaching Area (30.5710575, −96.3672611), private yard, and Biodiversity Research and Teaching Collections (30.18083, −98.493333). In Cameron County: Cactus Creek Ranch (26.21599, −97.4298142), Las Palomas Wildlife Management Unit- Arroyo Colorado Unit- hereafter ACU (26.3173093, −97.524518), and a private yard. Blanco and Walker had one site each (Blanco: Bamberger Ranch Preserve (30.180834, −98.493333); Walker: Sam Houston National Forest Overflow Campground (30.5662596, −95.633105)). Each of the three sites in Cameron County lies within the Arroyo Colorado watershed, where agriculture comprises nearly half of land use (Berthold, 2014); this is reflected in imidacloprid use estimates (Fig 1). As such two of our sites, ACU and Cactus Creek Ranch, are within natural areas but abut agricultural fields in at least one place. Similarly, the private residence in Cameron County does not directly abut any agriculture but agricultural lands predominate in this region of south Texas. Outside of Cameron County, none of the sampling sites was bordered closely by agriculture. From December 2020 to February 2022, passerines (birds in the order Passeriformes) and closely related birds (near-passerines) were sampled during each season (spring, summer, winter, fall), using 12 by 2.6m, 30-mm mesh mist nets (Avinet, Portland, ME). Nets were placed in wooded, edge, and grassland habitat, and were opened half an hour before sunrise, with net effort varying based on weather conditions. Upon capture, birds were banded with U.S. Geological Survey bands, and measures of wing chord (unflattened wing length from carpal joint to wingtip, mm), tarsus length (mm), and mass to the nearest 0.1 g were taken (Pyle, 1997); when possible, sex and age were assigned based on species accounts in the Identification Guide to North American Birds (Pyle, 1997). Fat was scored on a 0–5 scale with zero indicating no abdominal or furcular fat seen, and 5 indicating large, mounded fat deposits in both sites (Rogers, 1991). Brachial venipuncture was used to blood sample each bird when conditions allowed, using 28-gauge needles, and heparinized 70 μL capillary tubes (Fisher Scientific, Pittsburgh, PA) to collect up to 1% of a bird’s body mass per collection using techniques outlined in the Guidelines for the Use of Wild Birds in Research (Fair et al., 2010; and Owen, 2011). Bleeding was stopped with gentle pressure from a cotton ball. Recaptured birds had another blood sample taken if they were re-caught at least ten days after the first capture, at which time no more than 1% of the bird’s body mass was again sampled (Owen, 2011). No acute mortality occurred during processing or before release.

Figure 1.

A) The estimated agricultural use for imidacloprid, the most widely-used neonicotinoid, in the Texas based on 2019 data (lowest estimate; USGS, 2021). notably, this map and others created after 2015 do not include estimates for seed treatment applications. White circles indicate counties of sampling. B) Four counties of sample collection (white circles) atop the Texas Ecoregions (Wilkins et al., 2003).

Blood was expelled from microcapillary tubes, centrifuged, and plasma was stored at −80° until use. Animal research was approved by the Texas A&M University Institutional Committee on Animal Use and Care under protocol number IACUC 2021–0124, the Texas Parks and Wildlife Department, and the U.S. Geological Survey Bird Banding Laboratory (Federal Bird Banding Permit #: 23789).

2.2. Sample Pooling

In this manuscript, “sample” refers to an aliquot of 50 μL of avian plasma; this may come from one bird, or more than one if 50 μL couldn’t be obtained. For small-bodied birds from which the minimum plasma volume requirement of 50 μL could not be obtained, plasma samples were pooled across two to four individual birds of the same species, age, sex, and location that were collected no more than 1 month apart. Additionally, some pools were created by combining plasma from an individual with unknown detection status with residual plasma from a large volume sample in which neonicotinoids were not detected (ND). When plasma from an individual bird was tested more than once (i.e., once alone, and subsequently in a pool), data from the initial testing of the individual was used for analysis. The sample pooling process is detailed in Fig 2. Inadvertently, four pools were created using plasma from an individual of unknown status and residual plasma from an individual in which neonicotinoid concentrations were detected but below the limit of quantification (i.e., a positive sample); in all these 4 cases the result of the pooled sample result was ND. In all cases of pooled samples, negative pool results are interpreted with the understanding that each individual sample is diluted, and low neonicotinoid concentrations may not be detected. Positive pool results are interpreted with the understanding that a single or all individuals in the pool may have been positive without the ability to assign individual-level results, thus percentages of exposure in any category should all be interpreted as a minimum.

Fig 2:

Sample pooling was necessary in some cases to reach the required volume of 50 μL for this method. In such cases, and as shown in the schematic above, plasma from birds that met the pooling criteria were pooled into one sample. In some cases, we used extra plasma from a bird that had previously been tested individually and assigned an unexposed (ND) status to pool with a bird of unknown status, thereby diluting the unknown bird plasma with known negative plasma. In other cases, plasma from two birds of unknown status were pooled together (Created with BioRender).

2.3. Chemicals

Eleven pesticides and analytical internal standards (Sigma Aldrich, St Louis, MO) were used to determine the concentration of seven neonicotinoids in avian plasma using high performance liquid chromatography tandem mass spectrometry (HPLC-MS/MS) (Supplementary Table 2).

2.4. Determination of Neonicotinoid Concentrations in Plasma

Fifty microliters of plasma per sample (individual bird or pooled birds) were spiked into 100 μL of chilled acetonitrile containing 0.1 μM mixture of internal standards. Next, samples were vortexed briefly then centrifuged for 10 min at 12,000 x g. The supernatant was collected and transferred to a 1.5 mL centrifuge tube, dried under vacuum with a SpeedVac (Savant SPD1010, Thermo Scientific, Waltham, MA) and reconstituted with 50 μL aqueous mobile phase prior to HPLC-MS/MS analyses. Errors in pipetting technique resulted in twenty samples having proportionately higher volumes of plasma and reagents used; because the ratios of sample and reagents were still correct these samples were retained for analysis.

2.5. HPLC-MS/MS Analyses

A Kinetex Biphenyl column (4.6 × 50 mm, 2.6 μm, 00B-4622-E0; Phenomenex, Torrance, CA) with a SecurityGuard cartridge (KJ0–4282; Phenomenex) was used for chromatography of samples (10uL) using the 1290 Infinity II LC system and a 6470 triple quadrupole mass spectrometer (Agilent Technologies, Santa Clara, CA). Column temperature and the LC flow rate were set at 40°C and 0.6 mL/min, respectively. Initial chromatographic condition started at 60% mobile phase A (2 mM ammonium acetate in water with 0.1% formic acid, v/v) and 40% mobile phase B (2 mM ammonium acetate in methanol with 0.1% formic acid, v/v) then increased to 80% B by 9 min, and then returned to initial condition at 10 min until 12 min for sufficient equilibrium. Neonicotinoid levels in plasma were quantified by using the peak area ratios of standards to isotopically-labeled internal standards in a 12-point calibration curve (0, 1, 2.5, 5, 10, 25, 50, 100, 250, 500, 1000, 2500 pg/mL). Samples that had neonicotinoid concentrations greater than 2500 pg/mL were re-analyzed using a calibration curve ranging from 500 to 50000 pg/mL. MS/MS analyses were performed in positive ion mode with an electrospray ionization (ESI) source. The capillary voltage was set at 4000 V. The nebulizer gas pressure and gas temperature were set at 55 psi and 260°C, respectively. A chromatogram of each compound at 1 micromolar is seen in Supplemental Figure 1. The levels of detection and quantification for each compound can be found in Supplementary Table 2.

2.6. Outcome

The analysis of samples had three possible outcomes: ND (no compounds were detected); <LOQ (one or more neonicotinoids were detected but were below the level of quantification of this method; See Supplementary Table 2), or a quantifiable concentration of a neonicotinoid was reported (See Fig 3).

Figure 3:

Quantified concentrations of each of three compounds (acetamiprid, n=2; imidacloprid, n=35; thiacloprid, n=2) detected in avian plasma. Inset shows proportion of total samples (n=294) in each outcome; ND is no detection, <LOQ is less than the quantifiable concentration (See Supplementary Table 2). Note the LOD for thiacloprid, imidacloprid, and acetamiprid are 750 pg/mL, 5 pg/mL, and 500 pg/mL respectively, making the detection and quantification of imidacloprid more probable than other compounds.

2.7. Statistical analysis

We tested the hypothesis that bird exposure to neonicotinoids was influenced by avian ecologic and demographic factors, including sex, age, migratory status, family, season, foraging guild, and body condition. Exposure was treated as a binomial outcome, with the designations <LOQ and quantifiable neonicotinoid concentrations considered an exposure, and ND considered not exposed. We categorized species by what they eat seasonally (Billerman et al., 2022; De Graaf et al., 1985) within the following foraging guilds: frugivores (including three frugivore/insectivore species), granivore, insectivore (including a gummivore/insectivore species), and omnivore. Age was binned into categories of subadult (including local, hatch year and second year birds) and adult (including after hatch year, after second year, and older) and unknown. We created a parameter called ‘migratory status’, in which birds were coded as non-resident if they did not reside year-round in the location they were caught and sampled, residents if they are present throughout the year, or “unknown” if this could not be determined based on time of year and presence of breeding characters (Billerman et al., 2022; eBird, 2021; Pyle, 1997). To test the effect of season, we binned sampling dates into one of four seasons by day of the year (in pools containing individuals sampled on different dates, the average was used); Spring: March 1- May 31, Summer: June 1- August 31, Fall: September 1- November 30, and Winter: December 1- February 28. We analyzed species level relationships to exposure among species for which we had more than 5 samples run and calculated confidence intervals for incidence values for each of 13 species using GLMs; confidence intervals were unable to be estimated for swamp sparrow (Melospiza georgiana) and bronzed cowbird (Molothrus aeneus) as all samples were negative. Using wing chord and mass measurements (Ashton, 2002), we calculated scaled mass index (SMI) (Peig and Green, 2009) and used SMI and fat scores (average) as proxies of body condition. These associations were tested using averages from all samples (Table 1), and within species using samples from the two most numerous species in the sample set, northern cardinal (Cardinalis cardinalis) and red-winged blackbird.

Table 1.

Bivariable Results: Factors influencing neonicotinoid exposure in wild birds collected across Texas, 2020–2022, in which exposure includes the designations <LOQ and quantifiable neonicotinoid concentrations. All exposed birds had detections of imidacloprid with two individual birds simultaneously with detections of acetamiprid, and thiacloprid.

Factor	Variable	No. of samples1	Proportion of samples pooled	No. of individual birds2	Exposed samples (%)	Unexposed samples (%)	Factor P Value
Season	Spring	98	26/98	121	44 (45%)	54 (55%)	<0.001 ***
	Summer	50	5/50	53	11 (22%)	39 (78%)
	Winter	95	21/95	116	24 (25%)	71 (75%)
	Fall	51	10/51	59	28 (55%)	23 (45%)
Family	Bombycillidae	1	0/1	1	1 (100%)	0 (0%)	0.15
	Cardinalidae	55	8/55	60	22 (40%)	33 (60%)
	Corvidae	9	1/9	9	4 (44%)	5 (56%)
	Cuculidae	5	0/5	5	2 (40%)	3 (60%)
	Fringillidae	2	2/2	4	2 (100%)	0 (0%)
	Icteridae	66	4/66	69	25 (38%)	41 (62%)
	Icteriidae	4	1/4	5	0 (0%)	4 (100%)
	Mimidae	15	1/15	15	7 (47%)	8 (53%)
	Paridae	8	2/8	11	1 (13%)	7 (88%)
	Parulidae	11	6/11	17	1 (9%)	10 (91%)
	Passerellidae	61	25/61	87	18 (30%)	43 (71%)
	Passeridae	9	0/9	9	4 (44%)	5 (56%)
	Picidae	8	0/8	8	4 (50%)	4 (50%)
	Troglodytidae	12	6/12	16	6 (50%)	6 (50%)
	Turdidae	13	1/13	14	6 (46%)	7 (54%)
	Tyrannidae	11	4/11	15	3 (27%)	8 (73%)
	Vireonidae	4	0/4	4	1 (25%)	3 (75%)
Foraging Guild	Frugivoreb	9	3/9	11	5 (56%)	4 (44%)	0.31
	Granivore	113	28/113	142	46 (41%)	67 (59%)
	Insectivorea	53	14/53	65	16 (30%)	37 (70%)
	Omnivore	119	17/119	131	40 (34%)	79 (6%)
Age	Adult	162	40/162	200	49 (30%)	113 (70%)	0.05 *
	Subadult	119	21/119	135	52 (44%)	67 (56%)
	Unknown	13	1/13	14	6 (46%)	7 (54%)
Sex	Female	49	9/49	56	17 (35%)	32 (65%)	0.57
	Male	104	9/104	111	42 (4%)	62 (60%)
	Unknown	141	44/141	182	48 (34%)	93 (67%)
County	Blanco	10	0/10	10	3 (30%)	7 (70%)	0.37
	Brazos	172	53/172	220	62 (36%)	110 (64%)
	Cameron	109	8/109	115	42 (39%)	67 (62%)
	Montgomery	3	1/3	4	0 (0%)	3 (100%)
Migration Status	Non-resident	111	37/111	149	32 (29%)	79 (71%)	0.10
	Resident	105	16/105	115	42 (40%)	63 (60%)
	Unknown	78	9/78	85	33 (42%)	45 (58%)
Body Condition	Scaled Mass Index	289	61/289	342	average 38.35 g adjusted	average 34.50 g adjusted	0.10
Body Condition	Fat score	294	62/294	349	average fat score = 0.434	average fat score = 0.54	0.26
Total		294	62/294	349	107 (36%)	187 (64%)

¹A sample is comprised of plasma from either one individual or up to 4 pooled individuals

²The sample size of birds within each group is indicated; re-captured individuals are counted more than once.

^aOne gummivore/insectivore was included in insectivore guild for the analysis of foraging guilds.

^bThree frugivore/insectivores were included in frugivore guild for the analysis of foraging guilds.

Some categories add up to less than the total number of samples tested due to incomplete data.

We tested for associations between exposure and the factor-based predictive variables (season, family, species, foraging guild, age, sex, county, and migratory status) using binomial generalized linear models and ANOVAs (Chi Square test) to interpret resultant outputs (binomial model, logit link) (Popovic, 2016); we used a cutoff of p≤0.05 to denote factor significance. We tested for associations between exposure as a binary variable against continuous variables (Fat, SMI) using ANOVAs of linear models. Each factor was tested for associations with exposure within separate models. We plotted incidence of exposure and associated 95% confidence intervals based on model estimates and standard error. We used the program R for all statistical analyses (version 4.2.0) (R Core Team, 2020).

3. Results

Between December 2020 and February 2022, we performed 349 avian blood draws (including recaptured birds), amounting to 294 total samples for analysis (including pools). Samples represented 55 species across 17 avian families. We tested samples from frugivores (n=6), frugivore/insectivore (n=3), granivores (n=113), insectivores (n=52), omnivores (n=119), and one gummivore/insectivore. Overall, neonicotinoid exposure was detected in 36% of avian samples comprised predominantly of imidacloprid (for which the limit of detection was lowest) with rare detection of both acetamiprid or thiacloprid. No bird tested positive for clothianidin, dinotefuran, nitenpyram, or thiamethoxam. Of these, 11.8% of samples showed quantifiable concentrations of imidacloprid, 24.7% of samples were below the limit of quantification (<LOQ) for imidacloprid, and 63.5% of samples showed no detection (ND). Imidacloprid detected in quantifiable concentrations ranged from 10.8 pg/mL to 36,131 pg/mL, with a standard deviation of 177.7 pg/mL (Fig 3). Two individuals, a yellow-rumped warbler (Setophaga coronata ssp. coronata) from Brazos County (sampled 11/24/2021) and a brown-crested flycatcher (Myiarchus tyrannulus) from Cameron County (sampled 9/5/2021) each had quantifiable concentrations of 3 compounds in their plasma samples- imidacloprid, acetamiprid, and thiacloprid; other than these two samples, we did not detect any acetamiprid or thiacloprid (quantifiable or <LOQ) in the sample set. Acetamiprid concentrations in plasma of a yellow-rumped warbler, and a brown-crested flycatcher were 18971.3 pg/mL and 6844pg/mL, respectively; thiacloprid concentrations in the same birds were 7022.2 pg/mL and 17,367 pg/mL, respectively.

Avian family, foraging guild, and sex were not significant predictors of exposure (P=0.15; P=0.36, P=0.57; respectively). Age was a significant predictor of exposure (P=0.05), in which subadults had a significantly higher probability of exposure (43.4%, P=0.02, See Fig. 4) than adults (29.7%).

Figure 4:

Incidence of exposure to neonicotinoids by season, sex, age, foraging guild, and migratory status. Data points represent incidence of exposure at each level. Lines represent a 95% confidence interval estimated from a predictive model. “*” denotes a significance at p value ≤0.05, “**” at P<0.01, and “***” at P<0.001. Sp, Su, Fa, Wi signify spring, summer, fall, and winter respectively; F, M, U stand for female, male, and unknown sexes; Ad, Subad, and U refer to adult, subadult, and unknown ages; Frug, Gran, Insect, and Omni denote the foraging guilds frugivore, granivore, insectivore, and omnivore; Mig, Res, U stand for migratory, resident, and unknown migratory statuses.

Exposure did not differ among counties (P=0.35), but differed significantly across seasons, with birds sampled in fall and spring each having a significantly higher probability of exposure (fall: 53.8%, spring: 44.5%) to at least one neonicotinoid compared to birds sampled in summer (22%) or winter (24.7%) (P <0.001; Fig.4).

Neither measure of body condition (SMI and fat) was associated with the exposure outcome in average values of the total sample set or average values within each of the two most sample-rich species (northern cardinal SMI P= 0.44, Fat P= 0.51; red-winged blackbird SMI P=0.15, Fat P=0.16). Similarly, migratory status was not associated with the exposure outcome.

Of the sixteen species for which we tested more than five samples, American robin and red-winged blackbird had a significantly higher incidence of exposure than the overall exposure average (P=0.015 and 0.012, respectively; Fig. 5 and SM Table 1).

Figure 5:

Incidence of exposure for all avian species for which more than 5 samples were tested. Data points represent incidence of exposure per species with lines representing the 95% confidence interval estimated using a predictive model; for two species, swamp sparrow and bronzed cowbird, confidence intervals could not be estimated, as all samples tested were negative (See Supplemental Materials “SM” Table 1). The vertical line represents the average percent exposure across all samples. Species with significant relationships (red-winged blackbird and American robin) are designated with species icons “*” denotes species significance compared to the mean positivity rate of all sampled species and refers to a p value <0.05; “**” denotes a p value < 0.01.

3.1. Recaptured birds

Seven individual birds- one swamp sparrow, three northern cardinals, and three Carolina wrens (Thryothorus ludovicianus)- had initial and recapture blood samples tested. Two Carolina wrens and the swamp sparrow were analyzed as part of pooled samples at a single timepoint; all others were analyzed individually. Six of the seven birds were positive during at least one time point, whereas one bird was consistently negative (Fig. 6). Three birds showed positivity at two time points. First, a northern cardinal was positive in March 2021, negative over three months later in June, and positive again by the end of July of the same year. Second, a Carolina wren was positive at both its sampling points over an approximate one-month period in July- August 2021. Finally, a northern cardinal was positive at two sequential time points in fall of 2021 between September and October, and negative at the third time point in November.

Figure 6:

Neonicotinoid exposure over time in individual birds that were recaptured at a Brazos County site, Ecological and Natural Resource Teaching Area. Individuals are shown in different colors over time, and neonicotinoid exposure is shown as a binomial on the y axis; the only compound detected was imidacloprid. †Two Carolina wrens and one swamp sparrow which were analyzed as part of a pool at one time point.

4. Discussion

Based on field sampling from non-agricultural field sites in Texas, we detected neonicotinoids in 36% of samples from diverse wild bird communities with exposed birds found across the calendar year and across all avian foraging guilds and most avian families that were tested. Prior studies have often focused on agricultural sites and/or on focal birds species. For example, Lennon et al. (2020) found that nearly 51% of avian plasma samples post-sowing tested positive for clothianidin, compared with 11% of pre-sowing controls; they also noted overall exposure in 10/11 species tested. In a study of white-crowned sparrows, imidacloprid was above the method detection limit in 78% (28/36 birds) of birds, with thiamethoxam, acetamiprid, and thiacloprid detected commonly but in lower numbers (Hao et al., 2018). A study of House Sparrow feather samples (n=146) from a mix of conventional and organic farms showed that all samples were positive for neonicotinoids (Humann-Guilleminot et al., 2019).

Of the seven neonicotinoids for which we tested birds, nearly all exposures were to imidacloprid with two individual birds co-exposed to imidacloprid, acetamiprid and thiacloprid. For years, the highest proportion of the neonicotinoid market globally was imidacloprid, followed by thiamethoxam, clothianidin, acetamiprid, and thiacloprid in that order (Jeschke et al., 2011; Simon-Delso et al., 2015), but newer trends suggest imidacloprid use is decreasing, globally and in the U.S., in favor of other compounds like thiamethoxam and clothianidin (Simon-Delso et al., 2015). Neonicotinoid use in Texas has mirrored the large rise in popularity across the rest of the continent (Ertl et al., 2018; USGS, 2021). Beginning in 2015, the NAWQA estimate use data no longer includes estimates for seed treatment applications, the most common and widespread application by far, so 2013–2014 data was used for this comparison though it is now a decade old. The average estimated use from 2013–2014 in these four counties showed that they followed the national trend at that time with imidacloprid used most heavily, followed by thiamethoxam, clothianidin, then acetamiprid. Estimated 2013–2014 use data for the most commonly bought insecticides- acetamiprid, imidacloprid, thiamethoxam, and clothianidin- shows that Cameron County had the highest estimated use (in kg) of all four compounds in both years, with the exception of acetamiprid in 2013, in which Brazos County had higher estimated use (Wieben, 2019). The rare detection of both acetamiprid or thiacloprid and lack of detection of clothianidin, dinotefuran, nitenpyram, and thiamethoxam may reflect either that birds did not encounter them within the ~24 hours prior to capture (Bean et al., 2019) or that birds were exposed to levels that were below the method detection limit. Therefore, non-detections (or rare detections in the case of acetamiprid and thiacloprid) of compounds other than imidacloprid should not be interpreted as a complete reflection of neonicotinoid residues on the landscape, but should be interpreted in the context of our detection limits and the resulting method sensitivity for each compound. Given the short window of detection for neonicotinoid parent compounds in plasma (Bean et al., 2019), these data indicate that recent exposures to neonicotinoids (e.g., within the 24 hours prior to sampling) occur across most avian families and ecoregions sampled.

4. 1. Foraging guilds and taxonomic families

Granivorous birds were thought to be at highest risk of neonicotinoid exposure because of their potential ingestion of treated seeds (Lennon et al., 2020; Roy and Coy, 2020). Aside from the fraction of a percentage occupied by organic production, nearly every crop seed planted in North America is coated with a neonicotinoid (Simon-Delso et al., 2015). Each seed is coated in concentrations such that ingesting even a small number of seeds could cause sublethal effects or be lethal to a bird (Gibbons et al., 2015; Goulson, 2013). Insectivorous birds may also be at risk through the consumption of neonicotinoid-exposed insects (Hallmann et al., 2014; Li et al., 2020). However our study suggests a much broader array of birds are at risk, as foraging guild and avian family were not significant predictors of exposure. For example, we found imidacloprid exposure in a cedar waxwing (Bombycilla cedrorum), a frugivore/insectivore (De Graaf et al., 1985) that typically wouldn’t be thought of as high-risk for neonicotinoid ingestion, except perhaps in an orchard environment. We found that a yellow-rumped warbler had been exposed to three compounds. This bird typically forages in trees and bushes, not on the ground, which literature suggests is a more important dietary exposure route (Eng et al., 2017; Roy and Coy, 2020) (though they are known to be versatile in feeding style and height) (Hunt and Flaspohler, 2020)). Our results support a number of studies which have described neonicotinoid exposure in non-granivorous birds, including detection in free-living Eurasian eagle owls (Bubo bubo) (Taliansky-Chamudis et al., 2017), hummingbirds (Bishop et al., 2020; Graves et al., 2019), several non-granivorous farmland species (Lennon et al., 2020), and European honey buzzards (Pernis apivorous) (Byholm et al., 2018). Foraging guild as it relates to our results is also mentioned in section 4.5, in exploring species-specific exposure. The diverse assemblage of exposed birds found in this study and others suggests that environmental routes of exposure other than diet may be important, including water, sediment, or a yet to be identified environmental source (Bonmatin et al., 2019; Hladik et al., 2018; Hladik and Kolpin, 2015; Huang et al., 2020).

4. 2. Site

Avian exposure to neonicotinoids was detected in all sites from which we had 10 or more samples; there was no statistical difference in exposure incidence by county. All sites were non-agricultural, though two in Cameron County have adjacent agricultural fields. Although birds may have been exposed in agricultural settings before moving to the sites where we sampled them, the life history of many of the exposed birds suggests that they were likely exposed on site. We commonly sampled Carolina wrens and northern cardinals, both of which are non-migratory and have home ranges small enough that it is unlikely they were exposed off-site (Haggerty and Morton, 2020; Halkin et al., 2021); additionally, many of these birds were recaptured within and across seasons at the same field site. Previous work has described neonicotinoid exposure in wild living birds captured outside of agricultural settings (Hao et al., 2018) and has emphasized the importance of non-agricultural sources of exposure in honey bees (David et al., 2016; Long and Krupke, 2016). Our new data further underscore the importance of non-agricultural sites of exposure for wild birds.

4. 3. Season

Birds sampled in spring and fall had a significantly higher probability of exposure than those sampled in winter or summer (P<0.001). According to agricultural data for Texas, the usual planting day for the most commonly planted crops across the state occurs between March 1- May 1 in the Spring, and early September in the fall (“Crop Information- Planting & Harvesting,” 2022). The planting of neonicotinoid-coated seeds increases environmental residues (Morrissey et al., 2015; Samson-Robert et al., 2014) and could account for the effect of season on exposure we see at these times, even in non-agricultural areas. At sites in Blanco, Brazos, and Montgomery Counties April, May, and October were the months of highest rainfall, while sites in Cameron typically see the most rain in September and October based on measurements in each site’s closest cities (For Brazos, Cameron, Walker, and Blanco they are College Station, Harlingen, Huntsville and Johnson City, respectively; “Climate Data Texas,” 2022). We know rainfall during or just after crop planting is a main mechanism for neonicotinoid movement into bodies of water across the landscape (Radolinski et al., 2019; Wood and Goulson, 2017), so rainfall patterns could also help explain trends we saw in avian plasma during these seasons. Previous temporal studies have found elevated environmental pesticide residues associated with crop growing seasons of the study region. Sites throughout the Fraser Valley in British Columbia, Canada exhibited widespread and chronic pesticide contamination from spring to early summer in honey bee nectar and honey, plants, and hummingbirds (Bishop et al., 2022). A study of pesticides in bee pollen found reduced concentrations of neonicotinoids after the blooming of oilseed rape flowers, presumably because bees less frequently visit this crop and residues degraded throughout the summer (David et al., 2016). Yet other researchers commonly detected neonicotinoids in bee pollen samples collected after the crop planting season, concluding that environmental residues in dust, soil, or water are important sources of exposure risk for bees into late summer (Long and Krupke, 2016). Our data aligns with existing studies, but also expands our knowledge of temporal exposure in birds, as exposure in our sample areas increased, apparently in unison, with spring and fall growing seasons in the state.

The overlap of planting times and peak bird migration (Cohen et al., 2015; Gallinat et al., 2015) suggests that migratory birds may be exposed to increased neonicotinoid concentrations when they are most physiologically active and vulnerable. To assuage the physiologic toll of migration, migrant birds must refuel quickly and efficiently, often exploiting novel food and water sources in unfamiliar stopover sites (Bairlein and Gwinner, 1994; Jenni and Jenni-Eiermann, 1998). For this reason, migrant birds may be more susceptible to ingesting and suffering effects from neonicotinoids among other environmental toxicants (Klaassen et al., 2012). Because of the overlap in migration and crop planting timelines, we attempted to determine whether the influx of migratory birds in spring and fall may be associated with the likelihood of neonicotinoid exposure in birds, but there were no significant differences in exposure between migrants and non-migrants in our sample set.

4. 4. Age

Younger birds had a significantly higher probability of being exposed to imidacloprid than older birds in our sample set. There is evidence that subordinate, younger birds of some species may occupy lower quality habitat than older conspecifics (Lozano et al., 1996; Mettke-Hofmann et al., 2015), though there is no direct evidence that birds use neonicotinoid presence as a factor in habitat choice. Additionally, some aviary studies have demonstrated that avoidance of neonicotinoids may not be innate due to any smell or appearance of the compounds but conditioned in some situations, possibly due to post-ingestion upset (Lopez-Antia et al., 2014); in other cases avoidance may have been innate, but varied with food source and did not necessarily prevent mortality (Addy-Orduna et al., 2022). Accordingly, younger, naive birds may occupy habitat or use food and water sources with greater risk of neonicotinoid exposure since they have not yet encountered and learned to avoid these compounds, but these explanations need further study in field settings.

4. 5. Species specific exposure patterns

Among the 15 species for which we tested more than 5 samples, American robin (Turdus migratorius) and red-winged blackbird (Agelaius phoeniceus) had a significantly higher probability of exposure in comparison to the average of all samples. All the red-winged blackbirds we sampled were captured at a single site (Cactus Creek Ranch) in March 2021 and in February 2022. Red-winged blackbirds are partial migrants, with northern populations tending to winter in the southern part of the species’ range, while southern populations are believed to be nonmigratory (Yasukawa and Searcy, 2020)- the migratory status of the individuals in our study is therefore unknown. Red-winged blackbirds are known to form large flocks in the fall and throughout the winter; foraging in fields, they are often considered crop pests (Dolbeer, 1990). Indeed, exposure of red-winged blackbirds to neonicotinoids has been documented in agricultural settings (Roy and Coy, 2020). American robins also forage on the ground, which may be a risk factor for exposure (Eng et al., 2017; Roy and Coy, 2020). American robins are facultative migrants (Vanderhoff et al., 2020) and as such may be at risk of exposure to neonicotinoids as they move to find and follow food sources. Further, both species are synanthropic (Burns et al., 2012; Rodewald et al., 2013) which may afford them opportunities for encountering neonicotinoids in proximity to human dwellings. A limitation of comparing samples by species is that due to method blood volume requirements, larger bodied bird species are over-represented in individual-level (i.e., non-pooled) analyses.

4.6. Neonicotinoid levels in Recaptured Birds

We provide the first evidence of repeated exposure of wild passerine birds to imidacloprid residues in nature based on our recapture data (but see Bishop et al., 2020). Among seven re-sampled birds, imidacloprid exposure was detected in six, although three samples associated with the recaptured birds were assayed in pools and therefore have a lowered sensitivity for detecting neonicotinoid residues (See Fig. 6). Among the exposed birds, three had multiple positive timepoints separated by 1–4 months. Given the window of imidacloprid detection in avian plasma is likely under 24 hours (Bean et al., 2019; English et al., 2021), each positive sample came from a bird which was likely exposed to imidacloprid no more than a day prior. The chronically exposed species were Carolina wrens and northern cardinals; both are resident species with fairly small home ranges (Haggerty and Morton, 2020; Halkin et al., 2021). Though cardinals are known to forage in larger areas during the non-breeding season (Halkin et al., 2021), given the placement of mist nets (approximately 127m from property edge at the least), documented re-exposure during the breeding season, and the speed with which parent compounds are metabolized in avian plasma, it is likely these birds were exposed on site.

4. 7. Body Condition

Based on experimental data, we predicted individuals with lower body condition would be more likely to be associated with neonicotinoid exposure, yet we did not observe a relationship between exposure and two metrics of body condition - fat and scaled mass index – in the field. We have no data on the concentration of neonicotinoids to which birds were exposed prior to sampling, and we have only plasma concentrations from one time point, which are likely reduced from peak concentrations; accordingly, it is not possible to determine how the natural levels of environmental exposure may compare to the concentrations that were shown to cause weight loss in controlled aviary studies (among other clinical signs) (Eng et al., 2019; Lopez-Antia et al., 2015).

4. 8. Quantified concentrations in plasma

The quantified concentrations of neonicotinoids detected in avian plasma represent a fraction of the original dose to which the bird was exposed. Bean et al. (2019) found that recovery of imidacloprid is greatest in plasma one hour after exposure. Concentrations of imidacloprid (parent compound) detectable in Japanese quail (Coturnix japonica) plasma one hour after exposure represent an average of 0.25% of single low dose, or 0.2% of the dose after ten repetitions (low dose = 0.904 mg/kg) of imidacloprid to which a bird was exposed. At the same time interval after exposure, concentrations of imidacloprid in plasma were an average of 0.16% of a single high dose, and 1.6% the dose after ten repeated high doses (high dose=2.712 mg/kg). Assuming that imidacloprid metabolism in the wild birds of our study is comparable to a captive quail, we can estimate the environmental doses to which birds we sampled may have been exposed. For example, the median quantified imidacloprid concentration we detected was 214.1 pg/mL from the plasma sample of a 71.3 g Golden-fronted Woodpecker (Melanerpes aurifrons). Assuming 6−11 mL of whole blood/100 g of body weight of which between 35–55% is plasma (Fair et al., 2007; Samour, 2006) this amounts to 1.5–4.3 mLs of plasma for this woodpecker with the total amount of parent compound in the plasma compartment estimated to be 321.2–920.6 pg. In the unlikely scenario that we sampled this bird under the highest recovery conditions (mean of 1.6% of dose recovered, 1 hour after the last of ten repeated high doses) (Bean et al., 2019), the concentration to which this bird was exposed would have been 13,381.3 pg/mL, equivalent to a dose of 1.3e⁻² mg/kg (assuming 1 ml=1 g). The highest concentration of imidacloprid we found in wild bird plasma was 36,131 pg/mL and was from a 25.2g house sparrow; house sparrow blood is between 52.7–60.6% plasma (Puerta et al., 1995). By the same math and assumptions above, sampling at an hour after one low dose (0.25% dose recovery) and an hour after ten high doses (1.6% dose recovery), the estimated range of imidacloprid in the plasma was 28,543.5 pg- 61,422.7 pg. The environmental concentration encountered by this bird would be between 2,258,187.5 pg/mL and 14,452,400 pg/mL, making the dose between 2.3 mg/kg and 14.5 mg/kg. Imidacloprid is considered “highly toxic” to house sparrows (LD50= 41 mg/kg) among other sensitive species, prompting the Ecological Effects Branch of EPA to state that levels of concern within the agency were exceeded for songbirds (endangered and nonendangered alike) (Federoff et al., 2008; Lateulere, 1993). These calculations regarding a woodpecker and an old-world sparrow were made based on the metabolism of captive quail and assume the bird was sampled at the time of highest plasma neonicotinoid concentration but give us a rough idea of the residues to which they may have been exposed. Under these conditions, the highest quantified concentration we found in avian plasma may equate to higher than the oral doses given in some aviary studies to cause various sublethal effects in birds (Eng et al., 2019; Goulson, 2013); in a study of house sparrows specifically, a dose of 6 mg/kg caused house sparrows to exhibit incoordination, inability to fly, and unresponsiveness (Cox, 2001). This imidacloprid concentration is also high relative to some concentrations found in other wild birds. For comparison, a study of Eurasian eagle owls found one individual to have a blood concentration of 3.28 ng/mL (Taliansky-Chamudis et al., 2017). Studies in hummingbirds have found concentrations ranging from 3.1–3.4 μg/kg (Graves et al., 2019), and 3.63 ng/mL (ppb) of imidacloprid, clothianidin, and thiamethoxam in pooled cloacal fluid samples (Bishop et al., 2018). In contrast, detections of clothianidin in farmland birds were often orders of magnitude higher in concentration (median clothianidin residue in all positive House Sparrow samples was 7,120 ng/mL) than the concentrations we detected (Lennon et al., 2020). The concentrations we detected in the plasma of birds living in non-agricultural areas may be much lower than some of those detected in bird species sampled on farmland, but even the lower end of our dose calculations may be sufficient to cause high toxicity in bees (Pisa et al., 2015) and magnitudes higher than concentrations found to affect sensitive aquatic invertebrate populations over time (Morrissey et al., 2015; Stoughton et al., 2008).

4. 9. Limitations

Factors limiting our study include higher than desired method detection limits for parent compounds aside from imidacloprid. Imidacloprid was the most commonly detected compound in wild birds in our study which is likely explained in part because it has the lowest minimum detection limit of 5 pg/mL, though it is also the oldest and one of the most commonly used neonicotinoids (Goulson, 2013; Jeschke and Nauen, 2008; Simon-Delso et al., 2015). Another major limitation was the plasma volume requirement of 50uL for our method. Given we could safely obtain a blood volume of no more than 1% of a bird’s body weight, many smaller sized birds were not able to be tested individually. Accordingly, we pooled samples from smaller birds, leading to the dilution of each individual’s plasma and a corresponding reduction in sensitivity for detecting neonicotinoids in individual birds. Further, in the case of a positive pool, we assumed only one bird of the pool was positive and so the overall prevalence of exposure should be interpreted as a minimum in terms of bird number. Due to site-specific limitations and sampling practicality, two of our sites have low numbers of birds sampled (Blanco and Walker Counties). To better understand interactions between body condition and neonicotinoid exposure, future testing should consider sampling more actively migrating birds, whose body conditions can vary widely (Jenni and Jenni-Eiermann, 1998); additionally, body condition is best compared between individuals of the same species, so this metric may be easier to interpret in a species-focused study. Due to the short window of detection for neonicotinoid parent compounds in plasma, we had an acute view of exposure in birds, and testing of excrement (which may retain the parent compounds for longer) or detection of metabolites would provide information over a longer retrospective time window (Bean et al., 2019). As such, testing plasma for parent neonicotinoid compounds in plasma should not alone be used as a definitive risk assessment tool for species or locations. Additionally, birds given high doses of imidacloprid are known to avoid flight (Eng et al., 2019), so sampling via mist-net will inherently bias capture to reflect only those birds healthy enough to fly, precluding us from capturing and sampling birds that have encountered the highest, most debilitating environmental concentrations; therefore, those with affected mobility, those that were unable to migrate, and of course those that died are not reflected in this sample set

5.0. Broad Impacts

Neonicotinoids are widely used across the world, subject to inconsistent regulation by location and crop in some areas, and are used in part prophylactically; the ecological risks of neonicotinoids, especially co-exposures with multiple chemicals, their effects over time, and interactions with other threats remain poorly understood (van der Sluijs et al., 2015). Neonicotinoids can persist in soil months or years after application (Jones et al., 2014), and are found to accumulate in surface water (Hladik and Kolpin, 2015) and marshes (Main et al., 2014). Neonicotinoid compounds are also commonly found in humans; the CDC’s National Health and Nutrition Examination Survey (NHANES) detected a 4.8%, 0.54%, and 0.12% prevalence of imidacloprid, acetamiprid and thiacloprid in human urine, respectively (NHANES Neonicotinoids-Urine-Surplus, 2019). Likewise, Baker et al., (2019) reported detection frequencies of parent compounds of 30% imidacloprid, 2% acetamiprid, and 0% thiacloprid, in 60 human urine samples from adults in Atlanta, GA with no history of exposure to the compounds. Though neonicotinoids have relatively low toxicity in humans and other mammals, neurologic, hepatic, immunologic, reproductive, and genetic toxic effects of neonicotinoids have been observed (Han et al., 2018). Birds serve as sensitive indicators of ecosystem health and change, and can be monitored easily (Morrison, 1986; Rosenberg et al., 2019); studying neonicotinoid exposure in birds can be beneficial to understanding risk to humans and their shared ecosystems. Our study adds to the data available on temporal neonicotinoid exposure in wild bird communities. Identifying key drivers of exposure, such as season, species, and life stage, could inform future ecological risk assessments and risk management practices to reduce the potential deleterious impacts of neonicotinoid pesticides on bird populations and the ecosystems to which they belong.

Highlights.

1. Birds from non-agricultural Texas sites were tested for 7 neonicotinoids over time

2. Imidacloprid in 36% of samples; quantified concentrations from 10.8– 36,131 pg/mL

3. Higher limits of detection likely account for rare and undetected other compounds

4. Spring and fall had highest exposure rates; re-sampling showed continued exposure

5. Repeated and seasonal imidacloprid exposures are important conservation concerns