Effects of Coal Fly Ash on Tree Swallow Reproduction in Watts Bar Reservoir, Tennessee

Transcription

1 Integrated Environmental Assessment and Management Volume 11, Number 1 pp SETAC Effects of Coal Fly Ash on Tree Swallow Reproduction in Watts Bar Reservoir, Tennessee Suzanne J Walls,*y Carolyn B Meyer,z Jacqueline Iannuzzi, and Tamar H Schlekatk yarcadis, Knoxville, Tennessee, USA zarcadis, Lakewood, Colorado, USA ARCADIS, Annapolis, Maryland, USA karcadis, Durham, North Carolina, USA (Submitted 3 March 2014; Returned for Revision 5 May 2014; Accepted 18 August 2014) Special Series EDITOR'S NOTE: The Baseline Ecological Risk Assessment (BERA) of residual coal fly ash in Watts Bar Reservoir was conducted following a 2008 spill from the Tennessee Valley Authority Kingston Fossil Plant (Roane County, TN). Results of the BERA were used to focus the long term management strategy for the impacted river system. This article is among 7 peer reviewed articles in the special series, Ecological Risk Assessment for Residual Coal Fly Ash at Watts Bar Reservoir, Tennessee. The series includes articles presenting the following: problem formulation for the river system; sediment toxicity test procedures; benthic community analysis techniques; methods used to evaluate risk to fish and riparian and aquatic wildlife; an extensive analysis characterizing risk to the insectivorous tree swallow; and finally how the BERA results influenced management decisions. ABSTRACT Coal fly ash was released in unprecedented amounts ( m 3 ) into the Emory River from the Tennessee Valley Authority Kingston Fossil Plant on Watts Bar Reservoir in Tennessee. Tree swallows were exposed to ash related constituents at the ash release via their diet of emergent aquatic insects, whose larval forms can accumulate constituents from submerged river sediments. Reproduction of tree swallow colonies was assessed over a 2 year period by evaluating whether 1) ash constituent concentrations were elevated in egg, eggshell, and nestling tissues at colonies near ash impacted river reaches compared to reference colonies, 2) production of fledglings per nesting female was significantly lower in ash impacted colonies versus reference colonies, and 3) ash constituent concentrations or diet concentrations were correlated with nest productivity measures (clutch size, hatching success, and nestling survival, and fledglings produced per nest). Of the 26 ash constituents evaluated, 4 (Se, Sr, Cu, and Hg) were significantly elevated in tissues potentially from the ash, and 3 (Se, Sr, and Cu) in tissues or in swallow diet items were weakly correlated to at least one nest productivity measure or egg weight. Tree swallow hatching success was significantly reduced by 12%, but fledgling production per nest was unaffected due to larger clutch sizes in the impacted than reference colonies. Bioconcentration from the ash to insects in the diet to tree swallow eggs appears to be low. Overall, adverse impacts of the ash on tree swallow reproduction were not observed, but monitoring is continuing to further ensure Se from the residual ash does not adversely affect tree swallow reproduction over time. Integr Environ Assess Manag 2015;11: SETAC Keywords: Bioaccumulation Metals Reproductive success Se Sr Tachycineta bicolor Tree swallow INTRODUCTION The Tennessee Valley Authority Kingston Fossil Plant accidentally released an unprecedented amount (4.1 million m 3 ) of coal fly ash into the Emory River in Watts Bar Reservoir in Tennessee in December The ash, containing metals and metalloids, was distributed upstream approximately 3 river miles and downstream 15 river miles to Tennessee River mile (TRM) 560 (Walls et al. 2015, this issue). Following this release, concerned ecologists predicted significant ecological decline within Watts Bar Reservoir, based on the threat of Se and other ash related metals accumulating in fish and benthic invertebrate tissues, even after initial dredging removed 90% of All Supplemental Data may be found in the online version of this article. * To whom correspondence may be addressed: suzy.walls@arcadis-us.com Published online 25 October 2014 in Wiley Online Library (wileyonlinelibrary.com). DOI: /ieam.1585 the ash (Vig 2009; Lemly and Skorupa 2012). Few studies, however, have evaluated the potential uptake and effects of metals from ash settling basins or release sites on wildlife. Although insectivorous passerine species, such as the tree swallow (Tachycineta bicolor), have commonly been used to study uptake of environmental contaminants, serving as indicators of local contamination of metals and other persistent compounds (Custer et al. 2003; Brasso and Cristol 2008), studies which document the effects on reproduction of ash related metals uptake into passerine species are limited. Potential effects on the tree swallow resulting from a coal fly ash release from a failed dredge disposal cell were studied as part of an ecological risk assessment (ARCADIS 2012). The ash was released into the Emory River near mile 2.2, which is upstream from its confluence with the Clinch River near Kingston, Tennessee (Figure 1). Details of the release and recovery effort are discussed in Walls et al. (2015, this issue). The residual fly ash, highest in the Emory River, followed by the Clinch River, and lowest in the Tennessee River (Walls et al. 2015, this issue; Stojak et al. 2015, this issue), was

2 Coal Fly Ash Effects on Tree Swallow Reproduction Integr Environ Assess Manag 11, primarily composed of fine silica particles very similar to sand, but it also contained some constituents of potential ecological concern (COPEC) such as As, Cr, Cu, Pb, Hg, Ni, Se, Sr, Tl, V, Zn, and other metals that occur in coal (Walls et al. 2015, this issue). If ingested by wildlife, these constituents cause adverse effects at elevated concentrations. The tree swallow was selected for this study because of its aquatic insect focused diet when nesting near water bodies (Quinney and Ankney 1985; Blancher and McNicol 1991) and its use of artificial nest boxes for reproduction (Robinson 1990; Nicholson 1997). Its use of nest boxes makes the tree swallow an easily studied indicator species for evaluating temporal and spatial bioconcentration trends of metals and other persistent compounds (Custer et al. 2003). Its choice of insect prey exposes these birds to ash related contaminants in the river system because they nest near the water and forage within 100 to 200 m from the nest, ingesting insects that have bioaccumulated the contaminants before those aquatic insects emerge (US Geological Survey 2003). The degree of exposure to ash related contaminants can be evaluated by assessing COPEC concentrations in tissues of tree swallow eggs or nestlings; and possible effects of such exposure can be quantified by studying nest productivity. Some of the ashrelated contaminants maternally transfer to bird eggs or nestlings (e.g., Se [Heinz 1996], Hg [Brasso and Cristol 2008], and Sr [Mora 2003]), which in turn may reduce hatching success and nestling survival (Janz et al. 2010). Alternatively, some of these contaminants in the diet may directly reduce clutch size (e.g., Cu [Jackson and Stevenson 1981]). The objective of this study was to evaluate possible effects of ash related contaminants on tree swallow reproduction. This study evaluated 1) whether ash related contaminant concentrations in tree swallow tissue (i.e., egg, eggshell, and nestling) were higher in colonies with birds foraging along the ashimpacted river than in reference colonies, 2) whether numbers of fledglings produced per nest were lower for ash impacted colonies compared to reference colonies, and if so, 3) whether concentrations of ash related contaminants were correlated to fledgling production, implicating ash as a factor affecting reproduction. These findings will aid in the plans for long term management at the site and in the ongoing debate of Coal Combustion Residuals disposal regulations that have been proposed following the 2008 Kingston ash release (Lemly and Skorupa 2012; DeForest et al. 2013). METHODS Study area The study area was part of the shoreline of the Emory, Clinch, and Tennessee rivers in Roane County, Tennessee affected by the coal ash release. Ash released from the failed dredge cell impacted several miles of the Emory River and adjacent embayments, and moved downstream into the Clinch and Tennessee rivers. Dredging to remove the ash from the Emory River proceeded for over a year from March 2009 to the end of May No dredging was conducted in the Clinch or Tennessee rivers. Approximately m 3 of ash remained in the river system when dredging was completed (mostly in the Emory and Clinch rivers). The residual ash intermixed with and was capped by natural river sediments, leaving heterogeneous distributions of ash in all 3 rivers. Tree swallow boxes (n ¼ 471) sampled for egg or nestling tissue were placed along impacted and reference river shorelines at 4 locations in 2009 (3 impacted, 1 nonimpacted) and 9 locations in 2010 (6 impacted, 3 nonimpacted). Impacted colonies were labeled by river mile as Emory River mile (ERM) 3.0, ERM 3.5, Clinch River mile (CRM) 1.0, CRM 2.5, and Tennessee River mile (TRM) 566, and east, west, and north embayments (hereafter composited into embayments ) (Figure 1). After dredging was completed, the estimated volume of ash in sediment with greater than 15 cm of ash was approximately m 3 in the Emory River, m 3 in the Clinch River, and no sediment with ash that deep in the Tennessee River (Table 1 in Walls et al. 2015, this issue). Reference colonies were established at Fort Loudoun Dam (FLD), Tellico Dam (TLD), and islands on the Tennessee River (islands at TRM ), all upstream of influence from the ash release (Figure 1; Walls et al. 2015, this issue). FLD was the single reference colony in FLD, TLD, and the island colonies were composited to form one single, composite reference colony in Nest boxes were placed in close proximity to the rivers (within approximately 100 m). Terrestrial habitats near the nest boxes were open with grasses as the dominant vegetation type, and were generally similar among all sites (ARCADIS 2012). Field collections and laboratory analysis As is typical for responses to accidental releases, the sampling response and study designs were organized quickly in 2009, and thus had less data collection than in 2010, when sampling was intensified. Because no predredging data for sediment or tree swallow eggs were available, reference colonies were monitored to evaluate ash impacts. Only 4 tree swallow colonies were monitored in 2009 (56 boxes at ERM3.0, ERM3.5, CRM2.5, and FLD), which was increased to 7 colonies in Of these, 40 boxes (30 impacted, 10 reference) at the 4 original colonies were monitored for fledglings produced per nesting female, and only in 2010 (with only 1 reference colony, FLD). Monitoring and collections were performed during dredging efforts in both years. Multiple dredges were operating and potentially disturbing the ERM 3.0 and ERM 3.5 colony and invertebrate prey production in Emory River (only river dredged) during this period. All nest boxes were monitored at least bi weekly from April to July to evaluate occupancy and clutch size, with daily monitoring in 2010 at the 4 original 2009 sites (ERM 3.0, ERM 3.5, CRM 2.5, and FLD). In 2009, 52 of 56 boxes monitored were occupied whereas in 2010, 147 of 471 monitored boxes were occupied. One egg was randomly collected from almost every occupied nest (2009, n ¼ 43; 2010, n ¼ 137), and one 15 day old nestling was randomly collected from the same nest, if possible (2009, n ¼ 35; 2010, n ¼ 50). Eggs were weighed and measured. Nestlings were euthanized by asphyxiation using dry ice and weighed within 24 h of collection. Egg and nestling samples were frozen and then sent to a commercial laboratory for analysis. In the laboratory, homogenized egg content, unrinsed eggshells, and whole body nestlings were analyzed for the 26 inorganic ash related constituents of potential ecological concern (COPECs) identified for the site (Al, Sb, As, Ba, Be, B, Cd, Ca, Cr, Co, Cu, Fe, Pb, Mg, Mn, Hg, Mo, Ni, K, Se, Na, Sr, Ag, Tl, V, Zn). Although Hg was previously identified as a legacy constituent in different areas of Watts Bar Reservoir, it was included in this analysis given its significant correlation with

3 58 Integr Environ Assess Manag 11, 2015 SJ Walls et al. Figure 1. Locations of tree swallow colonies and nest boxes in study area. Impacted colonies located at ERM 3.5, ERM 3.0, Embayments (East, North, and West), CRM 2.5, CRM 1.0, and TRM 566. Reference colonies located at TRM , Fort Loudoun Dam, and Tellico Dam. Nest boxes with tissue collections (half shaded circle) and nest productivity measures (full shaded circle) are shown. The Emory River flows south into the Clinch River, and the Clinch River flows south into the Tennessee River. The reference colonies are located upstream on the Tennessee River, to the southeast in the figure.

4 Coal Fly Ash Effects on Tree Swallow Reproduction Integr Environ Assess Manag 11, Table 1. Mean and range (in parentheses) COPEC concentrations in sediment and mayfly tissue in corresponding tree swallow colony locations Sediment (mg/kg dw) a Mayfly adult tissue (mg/kg dw) b n Percent Ash Cu Hg Se Sr n Cu Hg Se Sr Colony ERM (9 79) ( ) ( ) ( ) ( ) ( ) 7.56 ( ) ( ) ERM (<1 79) ( ) ( ) ( ) ( ) ( ) 6.87 ( ) 1.02 ( ) CRM (1 40) ( ) 0.56 ( ) 1.71 ( ) NA ( ) ( ) 6.22 ( ) ( ) CRM (29 43) ( ) 0.70 ( ) 2.39 ( ) ( ) ( ) ( ) 5.88 ( ) ( ) TRM < ( ) 0.38 ( ) NA ( ) ( ) 3.91 ( ) ( ) Reference 3 < ( ) ( ) NA 9 22 ( ) ( ) 1.77 (1.5 2) 0.88 ( ) COPEC ¼ constituents of potential ecological concern; CRM ¼ Clinch River Mile #; ERM ¼ Emory River Mile #; NA ¼ not available; Reference ¼ upstream Tennessee River (TRM ); TRM ¼ Tennessee River Mile #. a Sediment ( meters), containing a mixture of native sediments and ash, was collected within approximately 500 m (up to 1000 m if necessary) of each tree swallow colony using vibracore samplers. Samples were collected after dredging was complete, from October 2010 through April b Mayfly adults were collected opportunistically by sweep netting within approximately 500 m (up to 1000 m if necessary) of each tree swallow colony location. Samples were collected after dredging was complete, from June through August of Indicates below Se detection limits in >85% of sediment samples (limits ranged from 1.28 to 1.92 mg/kg in Emory and Reference Reaches, 1.23 to 2.15 mg/kg in Clinch River, and 2.26 to 2.67 mg/kg in TRM 566. coal fly ash (see Meyer et al. 2015, this issue for correlations of ash percent with COPEC concentrations). Percent moisture was recorded to convert concentrations to a dry weight (dw) basis. Eggshells were randomly selected and analyzed only in 2010 (n ¼ 20) and only in 3 impacted (ERM 3.0, Embayments, TRM 566) colonies and in the composite reference colony. For the risk assessment (ARCADIS 2012), these COPECs were also sampled in 30 cm deep sediment cores collected in 2010 and 2011 (postdredging) and in adult mayflies in 2010 (collected postdredging) within approximately 500 m of the tree swallow colonies, but included samples up to 1000 m when closer samples were not available (see ARCADIS 2012 for methods). A summary of the sediment and mayfly tissue COPEC concentrations that were found at elevated levels in tree swallow eggs (discussed below) are presented in Table 1. Nest productivity Clutch size, hatching success, and 15 day nestling survival (assumed to be equivalent to fledging success) were recorded in 2010 for a subset of the nests (see sample sizes in Table 3), specifically nests at 1 reference colony (FLD) and 3 impacted colonies (ERM 3.0, ERM 3.5, and CRM 2.5). Nests were not monitored for productivity in the other reference colonies (composite that included TLD and islands at TRM ) or impacted colonies more distant from the source of the ash release (CRM 1.0 and TRM 566) due to logistical constraints. Nests at the embayments, which had low occupancy, were not included. It was assumed the ash would not increase nest predation rates and thus nests destroyed by predators were excluded from estimates of hatching and nestling survival, reducing sample size to 40 boxes with productivity estimates. To evaluate overall nest productivity, clutch size was multiplied by hatching success and nestling survival proportions and divided by 2 to estimate female fledglings produced per nesting female. Statistical analyses Statistical analyses were conducted using SAS, v Constituents detected at a low frequency of detection (<75%) or were not of concern (Na, K, Ca, Mg) were excluded from statistical analyses, with a few exceptions. Mercury was a chemical of importance for coal and was included at 50% detection rates. Arsenic, the chemical most strongly correlated with ash (Pearson r ¼ 0.80), was also included for the eggshell analysis at 55% detection rate. The nondetects were set at the full detection limit, and data were log transformed if required to meet normality of the test comparing impacted and reference areas. Thus, if nondetects skewed the distribution of the data, the data were transformed or a nonparametric test was applied. Removal of constituents for lack of detections reduced the original list of 26 COPECs to 12, which were: As, Ba, Cd, Co, Cu, Fe, Mn, Hg, Mo, Se, Sr, and Zn. Concentrations of the 12 COPECs in tree swallow eggs, eggshells, and nestlings were compared between impacted and reference colonies for each year. Furthermore, egg and nestling weight and nest productivity measures at impacted and reference locations in 2009 and 2010 were also compared. A2 way analysis of variance (ANOVA) with Dunnett s 1 tailed post hoc test was used to evaluate effects of location and year (2009 and 2010) on tissue concentrations, egg and nestling weight. Dunnett s test is the most appropriate test to compare one control to many treatments that holds the familywise error rate at or below the desired significance level

5 60 Integr Environ Assess Manag 11, 2015 SJ Walls et al. (a ¼ 0.05) (Ott and Longnecker 2010). This test compared FLD reference versus each affected colony: ERM 3.0, ERM 3.5, and CRM 2.5 in the 2 ANOVA. Because there were 5 more impacted and reference locations in 2010 when the study was expanded, a 1 way ANOVA with Dunnett s 1 tailed post hoc test was used to compare all 2010 impacted colonies (excluding 2009 data) to the 2010 composite reference (FLD, TLD, and islands at TRM ) for egg tissue concentrations. The 3 nest productivity measures (hatching success, fledgling success, and fledglings produced), egg and nestling weight, and clutch size were also evaluated in 2010 using a 1 way ANOVA with Dunnett s 1 tailed post hoc test at a reduced number of impacted sites (ERM 3.0, ERM 3.5, and CRM 2.5) compared to the 2010 reference (FLD). When raw or log transformed data failed to meet parametric test assumptions (true of all 4 nest productivity measures), a nonparametric 1 and 2 way Kruskal Wallis test was used (2 way performed with Scheirer Ray Hare extension) (Dytham 2011). Sample sizes were selected to ensure power to detect meaningful differences in COPEC concentrations in eggs and nestlings (relative to tissue benchmarks of adverse effects) was 80%, with 95% confidence (e.g., meaningful means able to detect a 2 5 mg/kg minimum significant difference [MSD] for all metals except Zn, for which the MSD was 30 mg/kg). However, power was lower for eggshells because of small sample sizes and for nest productivity measures because of small sample sizes and high variability. Thus, the Type I error rate (a) was set to 0.05 for statistical significance of differences among egg and nestling concentrations and to 0.10 for differences among eggshell concentrations and nest productivity measures. To evaluate whether constituents were related to reductions in nest productivity, concentrations of metals in egg and nestling tissue were correlated with nest productivity measures (egg and nestling weight, clutch size, hatching success, nestling survival, and female fledglings per nesting female) using Pearson (linear) and Spearman (rank correlation) coefficients. In addition, adult mayfly concentrations in Table 1, which were assumed to approximate diet exposure for the tree swallow (assuming swallows do not ingest sediment and adult mayflies represent the larger groups of insects eaten), were correlated with nest productivity measures for the 2 Emory River colonies (ERM 3.0 and 3.5) and 1 colony of the Clinch River (CRM 2.5). Only these 3 colonies had both mayfly and nest productivity data needed to develop a diet concentration response relationship. Concentrations in diet may be a better measure of exposure for correlating with nest productivity, because egg tissue concentrations can vary significantly with laying order within a clutch (Bryan et al. 2003; Van Dyke et al. 2013). RESULTS Concentrations in tissue Se, Hg, Cu, and Sr concentrations were statistically significantly higher in egg contents in at least one of the ashimpacted colonies (ERM 3.0, ERM 3.5, CRM 2.5, or CRM 1.0) than at least one of the reference colonies (FLD or composite reference), but only Sr was significantly higher in eggshells (Table 2). Se and Hg were also significantly higher in tree swallow nestlings in ash impacted colonies than in reference colonies. Except for Cu in eggs, interaction terms for location and year were not significant (p > 0.08), indicating effects of the ash were similar in 2009 and 2010 for most COPECs (thus the 2009 and 2010 data in Table 2 represent the average across years from the 2 way ANOVA). For Cu, the difference between reference (FLD) and impacted colonies was higher in 2009 than The higher Se and Hg in impacted eggs or nestlings of the impacted colonies (Table 2) reflects the higher Se and Hg concentrations in mayflies of the impacted colonies relative to reference colony (Table 1). Though Sr and Cu were elevated in eggs, they were only elevated at CRM 2.5 (Table 2). Sr was elevated in eggshells in 2 of the 3 impacted colonies sampled for eggshells (Embayments and TRM 566). Although results in 2010 for eggs in the TRM 566 colony for Ba, Fe, Mn, and Zn (mean concentrations in eggs) and for Ba in eggshells were significantly higher than those observed at the composite reference colony (Table 2), colonies closer to the release (i.e., ERM 3.0 and ERM 3.5) where more ash was deposited had much lower concentrations of these COPECs in eggs and eggshells in 2010 than the TRM 566 colony. Given that these COPEC concentrations in eggs that were more than 8 miles downstream of the release (where ash was greatly reduced) were not elevated compared to colonies close to the release (Table 2), it is unlikely that the higher Ba, Fe, Mn, and Zn egg concentrations at TRM 566 were attributable to the ash release. Rather, an unknown source (or sources) may have caused these differences. In support, the ash percentage in sediment after dredging (Table 1 and Stojak et al., this issue) was negatively and significantly correlated with distance from the source of the ash release (r ¼ 0.58, p ¼ 0.02). Nest productivity Number of female fledglings produced per nesting female did not significantly differ between ash impacted and reference colonies (p ¼ 0.186, Table 3, Figure 2). This productivity measure is an integration of clutch size, hatching success, and nestling survival. For these individual measures, mean clutch size (using larger N in Table 3) and nestling survival were not significantly different in the measured ash impacted (ERM 3.0, ERM 3.5, CRM 2.5) and reference (FLD) colonies, whereas mean hatching success in 2 impacted colonies (ERM 3.0 and ERM 3.5) was significantly lower than at the reference colony by 15% to 21% (p < 0.01; Table 3, Figure 2). When nest productivity results were combined from all 3 ash impacted colonies to increase sample size and power to detect differences, clutch size was significantly higher at impacted sites compared to the FLD reference colony (by 12%) when the subset of nets (smaller n for nestlings in Table 3) sampled for productivity were evaluated. In contrast, hatching success was significantly lower by 11% in ash impacted than reference colonies (Figure 2). The net result of the increase in clutch size and decrease in hatching success from the impacted colonies resulted in similar fledgling production in impacted colonies compared to the reference colony. A few nest productivity measures were weakly but significantly correlated (Spearman rank) with egg concentrations or adult mayfly concentrations for the COPECs that were elevated in eggs. Specifically, Sr concentrations in tree swallow eggs correlated weakly but negatively to nestling survival (r ¼ 0.28, p ¼ 0.04, n ¼ 57), mostly at low Sr concentrations (nestling survival was 100% at the 2 highest Sr concentrations). Cu and Sr correlated weakly but negatively with egg weight (r ¼ 0.24 and r ¼ 0.14, p < 0.05), but none of the COPECs that were elevated in nestlings were negatively

6 Coal Fly Ash Effects on Tree Swallow Reproduction Integr Environ Assess Manag 11, Table 2. Comparison of mean concentration of impacted and reference tissue samples for tree swallows a ERM and ERM 3.5 Impacted 2009 and CRM CRM 1.0 b 2.5 TRM 566 b Embayments b FLD 2009 and Reference Composite ref b,c 2009 and Egg content (n ¼ 40) (n ¼ 30) (n ¼ 25) (n ¼ 12) (n ¼ 15) (n ¼ 26) (n ¼ 15) (n ¼ 9) (n ¼ 10) (n ¼ 25) (n ¼ 46) Ba Co Cu Fe Mn Hg d Se 4.59 d Sr Zn Eggshell (n ¼ 0) (n ¼ 6) (n ¼ 0) (n ¼ 0) (n ¼ 0) (n ¼ 0) (n ¼ 0) (n ¼ 3) (n ¼ 4) (n ¼ 0) (n ¼ 4) As d Ba Cu Mn Hg d Se Sr Zn d Nestling (n ¼ 24) (n ¼ 15) (n ¼ 21) (n ¼ 10) (n ¼ 0) (n ¼ 27) (n ¼ 15) (n ¼ 0) (n ¼ 0) (n ¼ 13) (n ¼ 10) e Ba Cd Cu Fe Mn Hg Mo Se Sr Zn CRM ¼ Clinch River Mile; ERM ¼ Emory River Mile; FLD ¼ Fort Loudoun Dam; TRM ¼ Tennessee River Mile. a Two way ANOVA comparing year and location was applied to 2009 and 2010 data for 4 colonies sampled both years (ERM3.0, ERM3.5, CRM2.5, FLD). Because neither year nor the interaction term was significant (with the exception of Cu in eggs), the mean from the ANOVA across years is shown in this table. The mean for each colony sampled in 2010 is also shown to include colonies with data only in The means of the 6 colonies sampled in 2010 were compared with 1 way ANOVA in If the 1 or 2 way ANOVA was significant, significant differences from reference were determined using post hoc Dunnett's 1 tailed test ( p < 0.05, p < 0.01, p < 0.001, p < ). In that test, 2009 and 2010 impacted colony means were compared to FLD 2009 and 2010 means, and the 2010 impacted colony means were compared to the 2010 Composite Reference mean composed of FLD, Tellico Dam, and upper Tennessee River colonies. Power to detect eggshell differences was low so alpha was set to 0.10 ( < 0.1), whereas it was 0.05 for the other tissues. b Location only sampled in c Composite reference includes samples from Fort Loudon Dam, Tellico Dam, and upstream Tennessee River (TRM ). d Constituent had a low (<75%) rate of detection. Nondetects were set at full detection limits. e Nestlings were all collected from Fort Loudon Dam in 2010.

7 62 Integr Environ Assess Manag 11, 2015 SJ Walls et al. Table 3. Comparison of reproductive metrics of impacted and reference tree swallow colonies a Impacted Reference ERM 3.0 ERM 3.5 CRM 2.5 Average impacted colonies FLD Egg (n ¼ 15) (n ¼ 12) (n ¼ 15) (n ¼ 91) (n ¼ 10) Egg weight (g) Clutch size (n) Nestling (n ¼ 15) (n ¼ 10) (n ¼ 15) (n ¼ 40) (n ¼ 10) Nestling weight (g) Hatching success (%) Nestling survival to near fledging (%) Female fledglings (n) per nesting female Only colonies monitored for productivity are presented here, including Clinch River mile (CRM), Emory River mile (ERM), and Fort Loudon Dam (FLD). a Significant differences from reference using 1 way ANOVA and post hoc Dunnett's 1 tailed test ( p < 0.05, p < 0.01). correlated with nestling weight. Notably, no combination of egg COPECs in a multiple regression was predictive of hatching success, which was the only productivity measure significantly reduced by the ash. Concentrations in mayflies (representing concentrations in swallow diet) of the COPECs elevated in tissue were not significantly correlated to clutch size, hatching success, or nestling survival. The Se concentration in mayflies was significantly and negatively correlated to female fledglings produced per nestling female; however, the linear correlation was weak due to high variability in fledglings produced and a small sample size of Se concentrations of mayflies (n ¼ 3) around each of the colonies (ERM 3.0, ERM 3.5, CRM 2.5) (Pearson correlation, r ¼ 0.35, p ¼ 0.03, Figure 3). Strontium mayfly concentrations exhibited a similar relationship with fledglings produced (r ¼ 0.34, p ¼ 0.03; Figure 3). Because nest productivity measures were not monitored in any Tennessee River locations, these correlations only apply to the Clinch and Emory rivers. The correlations suggest some risk to hatching or fledgling success in the Emory River due to Se or Sr, or both. However, the correlations are weak, and the differences between the Emory and Clinch river colonies in mayfly diet concentrations are small for Se and Sr (Figure 3). The difference in nest productivity between the 2 rivers may instead be due to the disturbance from dredging that took place on the Emory River near colonies ERM 3.0 and ERM 3.5. DISCUSSION In the ash impacted colonies, concentrations of 8 of 26 ashrelated COPECs were higher in tree swallow eggs, eggshells, or nestlings than in the reference colonies. However, 4 of those constituents (Ba, Fe, Mn, and Zn) were higher only in the colony eggs at the most downstream impacted Tennessee River location, where ash deposits were lowest (Table 1 in Walls et al. 2015, this issue), and those elevated concentrations were unlikely related to ash. The remaining 4 elevated COPECs in tissue (Se, Sr, Cu, and Hg) might be attributed to the ash release and were higher in at least 1 of 3 types of tissues. The reproductive study indicated no net effect of these 4 COPECs on reproduction (fledglings produced). Therefore, the mean concentrations of COPECs reported for tissues in this study for Se, Sr, Hg, and Cu (Table 2) represent a no effect effect level on reproduction in a field population of tree swallows. For comparison, the concentrations in tissues of the 4 COPECs were evaluated against background or effects concentrations in the literature for individuals or populations of birds. No known toxicity reference values (TRVs) have been established for tree swallows. Of the 4 COPECs, only Se exceeded any reported TRV for any bird species. The lowest reported valid (nonbackground) egg TRV was 7.7 mg/kg Se in mallards (Beckon et al. 2008), but egg TRVs related to hatching success ranged up to 60 mg/kg Se in American avocets (Recurvirostridae americana) (Janz et al. 2010). The 7.7 mg/kg is slightly higher than reported upper background concentration for eggs of 5 mg/kg (Ohlendorf and Heinz 2011). Of all impacted tree swallow eggs, only 2.4% (3 eggs), exceeded the lowest TRV of 7.7 mg/kg Se. The maximum Se concentration (13.04 mg/kg) was within the recommended TRV range (12 to 15 mg/kg) for mallards in Adams et al. (2003). One field study by Harding et al. (2005) observed no effects on hatching success for a red winged blackbird population averaging 22 mg/kg Se dw in eggs, another bird species that, when near water, feeds mostly on emergent aquatic insects in the air and insects on vegetation during the breeding season (Bird and Smith 1964). The maximum egg concentration in that study was 39.9 mg/kg Se, much higher than the TVA results. Not all eggs were sampled in each nest in these studies or this study so some with higher Se may have been missed, but the differences in concentration from egg to egg in a nest are usually no more than 1 mg/kg Se (Van Dyke et al. 2013). Concentrations of Se in whole body tissue in grackle nestlings feeding on coal fly ash deposits ranged from 8 to 12 mg/kg (range of grackle nestling carcass, liver, and feather concentrations; Bryan et al. 2012), but the effect on grackles in that study is unknown. No nestlings in the Kingston ashimpacted sites exceeded this range (maximum concentration of 6.6 mg/kg dw), supporting that exposure to Se for nestlings is relatively low. Moreover, nestling survival was similar between impacted and reference reaches.

8 Coal Fly Ash Effects on Tree Swallow Reproduction Integr Environ Assess Manag 11, Figure 2. Nest productivity measures (clutch size of eggs per nest, hatching success and nestling survival, and number of female fledglings produced per nesting female) at reference colony (FLD) and impacted colonies (individual data [A] and pooled data [B] for ERM 3.0, ERM 3.5, CRM 2.5). Asterisk ( ) indicates a significant difference between impacted and reference colonies (p < 0.05). When Se is evaluated by colony, only 1 of 4 ash impacted colonies sampled for nest productivity in 2010 (ERM 3.0) had eggs exceeding the 7.7 mg/kg egg TRV for Se (6.7%, or 2 of 30 eggs at and mg/kg Se). However, the other Emory River colony evaluated (ERM 3.5) had an egg exceeding this TRV (10.87 mg/kg) in 2009, the year when nest productivity was not measured (1 of 25 eggs or 4% for this colony exceeded 7.7 mg/kg over 2 years). The ERM 3.0 colony had 15% reduction in hatching success in 2010 relative to reference colonies, slightly less than the 21% reduction observed for the ERM 3.5 colony. Only the ERM 3.0 colony showed significant elevation of Se in eggs (Table 2). Clutch size was slightly higher at ERM 3.0 than the reference, compensating for the lower hatching rate relative to reference. The finding of 4% to 7% of eggs exceeding the Se TRV in colonies with reduced hatching success suggests that Se could be contributing to decreased hatching success, but is unlikely given that Se in egg or nestling tissue was not correlated to any nest productivity measure and Se was not significantly elevated in the eggs of one of the colonies with reduced success. Moreover, differences in Se of mayflies in the diet between the Emory and Clinch river colonies (with reproductive success measured) were relatively small (6.87 to 7.56 mg/kg vs 6.22 mg/kg Se, Table 1), which suggests mayflies were not responsible for the reduced hatching success at the Emory River colonies. Strontium in eggshells can interfere with Ca transfer from the eggshell into the embryo, leading to insufficient Ca for bone formation (Mora et al. 2007). No evidence supports this occurred in the impacted colonies. Though high concentrations of Sr in eggshells are known to reduce eggshell strength and increase the number of cracked eggs (Mora 2003), the concentrations at which Sr was elevated and correlated to nestling survival at the impacted colonies in the egg and eggshell had no effect on hatching success, nor on overall

9 64 Integr Environ Assess Manag 11, 2015 SJ Walls et al. Figure 3. Relationship between female fledglings produced per nesting female and (A) dietary Se concentrations in adult mayflies or (B) dietary Sr concentrations in adult mayflies. reproductive success. No significant differences were found in Ca between egg, eggshell, or nestling concentrations among reference and impacted sites (Supplemental Material, Appendix 1). Furthermore, no deformed nestlings were observed during the 2009 or 2010 monitoring. Finally, no studies have implicated Sr as decreasing reproduction at the egg concentrations observed in this study. The maximum concentration of mg/kg Sr in eggs is below the mean concentrations of normally developed embryos (66 mg/kg dw) at a chemicallyimpacted site in Schwarzbach et al. (2006) and below the median lethal dose (LD50) of Sr injected into eggs at 73 mg/kg (assuming 80% moisture in egg and 55 g egg,) (Ridgway and Karnofsky 1952). The same is true for eggshells. Background concentrations of Sr in eggshells from the literature have been noted at 176 mg/kg (Mora 2003) to 507 mg/kg (Mora et al. 2011). The maximum eggshell concentrations from impacted sites were all within or below this range (maximum of mg/kg). All of these results suggest that impacts from Sr to the tree swallow population are unlikely. Copper also was somewhat elevated in a Clinch River ashimpacted colony and was negatively correlated with egg weight, but was not correlated with any productivity measures. Cu is a micronutrient, but at high concentrations can be toxic and reduce reproduction (Eisler 1998). One outlier egg on the Emory River (ERM 3.5) in 2009 had a Cu concentration of mg/kg, but most concentrations were quite low and similar to background concentrations in eggs of 1 to 9 mg/kg (King et al. 1983; Andrews et al. 1997; Mora 2003). Means for Cu in the TVA impacted eggs ranged from 2.43 to 3.16 mg/kg for 2009 and 2010 (Table 2). Ridgway and Karnofsky (1952) provide an LD50 of 20 mg/kg based on their egg injection study, which exceeds all swallow egg Cu concentrations on the impacted reaches. No definitive tissue TRV for Cu has been published, possibly because reduced egg production can result from a high Cu diet intake without elevated concentrations of Cu in egg tissue (data in Jackson and Stevenson 1981). A high Cu concentration (16.3 mg/kg) also was observed in one 2010 egg in the TRM reference site, indicating background egg samples for Cu sometimes can be high. A lack of correlation with nest productivity and lack of elevated Cu concentrations in the Emory River colonies suggests Cu from ash is likely not affecting tree swallow populations. Concentrations of Hg, although elevated in some eggs and nestlings in the Emory or Clinch river colonies as compared to the reference colonies, did not exceed the reported egg and nestling TRVs. The maximum Hg concentration (0.37 mg/kg dw) in tree swallow eggs in the Emory and Clinch rivers was well below the estimated Hg TRV of no effects in tree swallow egg tissue (3.85 mg/kg dw, converted using 80% moisture) reported in Custer et al. (2007) as well as the Hg TRV estimated for all avian eggs (2.0 mg/kg) reported in Shore et al. (2011). Similarly, the maximum whole body nestling concentration (0.15 mg/kg dw) was well below 5 mg/kg dw Hg, which is the level reported as no observed effect on productivity (nestlings per occupied nest) in an osprey population in a Hg polluted area (Hakkinen and Hasanen 1980). Ospreys are not closely related to swallows in body type, but the comparison suggests that the very low maximum value of 0.15 mg/kg dw in swallow nestlings is not of concern. Because Hg was not correlated to productivity measures and was well below the estimated tissue TRVs, it is unlikely to have had an effect on tree swallow reproduction. To understand the reason for minimal effects on tree swallows despite exposure to COPECs in the ash, mayfly concentrations were compared to sediment and tree swallow egg concentrations. Metals and metalloids significantly (p < 0.05) elevated in adult mayflies in the impacted river reaches compared to the corresponding upstream reference reach in the same river were As, Ba, B, Cd, Co, Mn, Mo, Se, Sr, V, and Zn (see details in ARCADIS 2012); however, only Se, Sr, Cu, and Hg were significantly elevated in tree swallow eggs. Only Se and Sr show a consistent increase in concentration from one media to the next in the food chain (sediment to mayfly, mayfly to egg), resulting in elevated concentrations in the egg. Selenium was detected at very low levels in sediment, near detection limits, and was not even detected in the sediment of the most impacted Emory River reach (Table 1). However, in the Clinch River, Se was detectable in the sediment; and on average, concentrations more than doubled when incorporated into mayfly tissue (Table 1). However, Se concentrations did not change much between the mayfly and bird egg tissue (compare Table 1 to Table 2). Strontium did not bioconcentrate from sediment to mayfly tissue (much lower concentrations in tissue), but did increase by approximately 10 times between mayfly tissue and swallow egg tissue (although concentrations were still low in tissue). Therefore, both Sr and Se can increase in concentration in the food chain, though at different steps in the chain. This bioaccumulation analysis would be more accurate if concentration from all components of the diet were included and may be underestimated using only aquatic insects (mayflies). Overall, mayfly concentrations were low and bioaccumulation of ash COPECs

10 Coal Fly Ash Effects on Tree Swallow Reproduction Integr Environ Assess Manag 11, from aerial insects (mayflies) to tree swallow eggs is low for most metals or metalloids in the ash, which supports why reproduction was not significantly affected by the ash. Previous contaminant studies conducted on tree swallow reproduction focused primarily on contaminants such as polychlorinated biphenyls or Hg in river systems (Custer et al. 2003; Echols et al. 2004; Dobs et al. 2005; Brasso and Cristol 2008). Although some studies have reported concentrations of Se in tree swallow eggs or tissue for other species (White et al. 1986; Custer et al. 2003, 2006; Ratti et al. 2006), these studies found only low levels of Se, comparable to other locations in the United States. To date, there has been no study that addressed both exposure and effects of Se in the tree swallow species. A barn swallow study (King et al. 1994) reported somewhat lower concentrations of Se (mean of 2.8 mg/kg at impacted sites) than this study with no observed effects on reproduction, which aligns with the results presented here (impacted means ranged from 3.15 to 4.75 mg/kg Se). The results at the Kingston Plant release site are similar to another study area with coal fly ash, which found Se was the main constituent of concern and much higher in ash basin diets of common grackles (Quiscala quiscula) than in reference diets (Bryan et al. 2012). Selenium was elevated in grackle nestlings and eggs, although Se concentrations only in liver and a few eggs approached concentrations of concern. The reproduction results from the Kingston ash release study supplemented by the exposure results of Bryan et al. (2012) and King et al. (1994) indicate that, whereas insectivorous birds can be exposed to Se in coal fly ash through the food chain, their exposures are not significant enough to cause large reproductive losses in these species. Given the magnitude of the ash release at TVA and predicted risk of dietary exposure for Se (predicted to be above thresholds of concern in the wildlife risk assessment, see Meyer et al. 2015, this issue), these results are surprising. The study took place during the extensive ash recovery phase and not all ash was removed. As a result, these results offer valuable information that can help advise the long term management at the site. In summary, Se in ash appears to be the main constituent of concern to tree swallows because Se concentrations in eggs were elevated in impacted areas with up to 13 mg/kg Se, at levels identified of potential concern (Adams et al. 2003). However, this study shows no effect on reproduction of the tree swallow population. Monitoring of the Emory River tree swallow colonies is continuing as part of the long term monitoring program to further ensure Se from the residual ash does not adversely affect tree swallow reproduction over time. Acknowledgment The authors would like to thank the following individuals and agencies for their assistance with collecting, processing, and analyzing samples for this project: N. Carriker, T.H. Henry, W. Hopkins, W. James, A. Johnson, J. Iannuzzi, E. Burton, J. Morris, Environmental Standards, Pace Analytical Services, Kingston Parks and Recreation, US Fish and Wildlife Services, and Tennessee Wildlife Resources Agency. the Tennessee Valley Authority (TVA) funded this study. SUPPLEMENTAL DATA Appendix S1. Trace element concentrations and productivity in tree swallows: REFERENCES Adams WJ, Brix KV, Edwards M, Tear LM, DeForest DK, Fairbrother A Analysis of field and laboratory data to derived selenium toxicity thresholds for birds. Environ Toxicol Chem 22: Andrews BJ, King KA, Baker DL Environmental contaminants in fish and wildlife of Havasu National Wildlife Refuge, Arizona. US Fish and Wildlife Service. [cited 2013 September 19]. Available from: southwest/es/documents/r2es/havasu.pdf. ARCADIS River system baseline ecological risk assessment, Tennessee Valley Authority, Kingston Ash Recovery Project. EPA AO 050. May. [cited 2013 February 1]. Available from: 20Documents/BERA/Appendix%20I%20BERA/App%20I%20BERA.pdf Beckon WN, Parkins C, Maximovich A, Beckon AV A general approach to modeling biphasic relationships. Environ Sci Technol 42: Bird RD, Smith LB The food habits of the red winged blackbird, Agelaius phoeniceus, in Manitoba. Can Field Nat 78: Blancher PJ, McNicol DK Tree swallow diet in relation to wetland acidity. Can J Zool 66: Brasso RL, Cristol DA Effects of mercury exposure on the reproductive success of tree swallows (Tachycineta bicolor). Ecotoxicology 17: Bryan AL, Hopkins WA, Parikh JH, Jackson BP, Unrine JM Coal fly ash basins as an attractive nuisance to birds: parental provisioning exposes nestlings to harmful trace elements. Environ Pollut 161: Bryan AL Jr, Hopkins WA, Baionno JA, Jackson BP Maternal transfer of contaminants to eggs in common grackles (Quiscalus quiscala) nesting on coal fly ash basins. Arch Environ Contam Toxicol 45: Custer CM, Custer TW, Dummer PM, Munney KL Exposure and effects of chemical contaminants on tree swallows nesting along the Housatonic River, Berkshire County, Massachusetts, USA, Environ Toxicol Chem 22: Custer CM, Custer TW, Hill EF Mercury exposure and effects of cavitynesting birds from the Carson River, Nevada. Arch Environ Contam Toxicol 52: Custer CM, Custer TW, Warburton D, Hoffman DJ, Bickham JW, Matson CW Trace element concentrations and bioindicators responses in tree swallows from northwestern Minnesota. Environ Monit Assess 118: DeForest DK, Reash RJ, Toll JE Comment on Wildlife and the coal waste policy debate: Proposed rule for coal waste disposal ignore lessons from 45 years of wildlife poisoning. Environ Sci Tech 47: Dobs PL, Birmingham EM, Williams TD, Ikonomou MG, Bennie DT, Elliott JE Reproductive success and contaminants in tree swallows (Tachycineta bicolor) breeding at a wastewater treatment plant. Environ Toxicol Chem 24: Dytham C Choosing and using statistics: A biologist's guide. 3rd ed. Hoboken (NJ): Wiley Blackwell. 298 p. Echols KR, Tillitt DE, Nichols JW, Secord AL, McCarty JP Accumulation of PCB congeners in nestling tree swallows (Tachycineta bicolor) on the Hudson River, New York. Environ Sci Technol 38: Eisler R Copper hazards to fish, wildlife, and invertebrates: A synoptic review. Biological/Contaminant Hazard Review Report 33. January. National Biological Service, US Department of the Interior Laurel (MD). Hakkinen I, Hasanen E Mercury in eggs and nestlings of the osprey (Pandion haliaetus) in Finland and its bioaccumulation from fish. Ann Zool Fennici 17: Harding LE, Graham M, Paton D Accumulation of selenium and lack of severe effects on productivity of American dippers (Cinclus mexicanus) and spotted sandpipers (Actitis macularia). Arch Environ Contam Toxicol 48: Heinz GH Selenium in birds. In: Beyer W, Heinz G, Redmon Norwood A, editors. Environmental contaminants in wildlife: interpreting tissue concentrations. Boca Raton (FL): CRC. p Jackson N, Stevenson MH A study of the effects of dietary added cupric oxide on the laying, domestic fowl and a comparison with the effects of hydrated copper sulphate. Br J Nutr 45: Janz DM, DeForest DK, Brooks JL, Chapman PM, Gilron G, Hoff D, Hopkins WA, McIntyre CA, Mebane Palace VP., et al Chapter 6: Selenium toxicity in aquatic organisms. In: Chapman PM, Adams WJ, Brooks ML, Delos CG, Luoma SN, Maher WA, Ohlendorf HM, Presser TS, Shaw DP, editors. Ecological