, Volume 21, Issue 8, pp 2391–2397

Relationship between blood mercury levels and components of male song in Nelson’s sparrows (Ammodramus nelsoni)


    • Department of Biological SciencesUniversity of Southern Maine
  • Christine R. Maher
    • Department of Biological SciencesUniversity of Southern Maine

DOI: 10.1007/s10646-012-0994-0

Cite this article as:
McKay, J.L. & Maher, C.R. Ecotoxicology (2012) 21: 2391. doi:10.1007/s10646-012-0994-0


Mercury (Hg) adversely affects the health and behavior of exposed wildlife; however, behavioral effects remain largely unknown. Changes in avian singing behavior may affect a male’s fitness because song reveals male quality and thus influences female mate choice and male territory-holding ability. Nelson’s sparrows (Ammodramus nelsoni) live exclusively on salt marshes and risk high levels of Hg exposure and bioaccumulation. We recorded songs of male Nelson’s sparrows at two locations with different Hg exposure to determine if total blood Hg concentration was related to song characteristics, as previously reported for other species. Males with higher blood Hg levels sang at higher maximum tonal frequency, but blood Hg and site location did not influence low tonal frequency and bout duration, contrary to predictions based on other species. Within the contaminated site, Hg levels were related to bouts per minute and gap duration, such that males at that site sang faster songs. Hg influences hormones and alters brain development, raising questions about specific effects on the brains and singing behavior of male Nelson’s sparrows.


MercuryBird songAmmodramusnelsoniSong rate


Mercury (Hg), a heavy metal, readily enters rivers and streams where biotic and abiotic processes methylate Hg, resulting in high methylmercury (MeHg) levels (Bouton et al. 1999; Eagles-Smith et al. 2009). Salt marshes make excellent habitats for Hg methylation because urban watersheds carrying Hg pollutants commonly drain into estuaries and other wetlands (Evers et al. 2007). In addition, constant changes in tidal water levels allow high rates of deposition and accumulation of Hg (Shriver et al. 2006). Waterborne Hg and MeHg in turn put salt marsh wildlife, both terrestrial and aquatic, at risk for exposure (Wolfe et al. 1998). Wildlife exposed to Hg can experience several effects, including muscle ataxia, brain damage, and death (Wolfe et al. 1998). However, behavioral changes may be more indicative of toxin exposure than traditionally measured chemical or physical criteria because of the complex development of behaviors (Hallinger et al. 2010). Specifically, Hg exposure can impair learning, contribute to memory deficits, and reduce social behavior (Wolfe et al. 1998; Bouton et al. 1999).

Exposure to Hg alters bird song in some species. Heavy metal exposure in male great tits (Parus major) results in less song production and lower repertoire sizes (Gorissen et al. 2005). Carolina wrens (Thryothorus ludovicianus), house wrens (Troglodytes aedon), and song sparrows (Melospiza melodia) exposed to Hg produce a lower diversity of note types and sing at lower frequencies; Carolina wrens and house wrens perform shorter strophes or sections of song (Hallinger et al. 2010). However, few studies specifically relate Hg exposure to bird song, and we must look at a variety of species to fully understand the effects of Hg on song and song production.

Reproduction serves as one of the most sensitive toxicological responses (Wolfe et al. 1998). Overall reproductive success in birds can decline by 35–50 % due to dietary MeHg exposure even with the absence of impairment in adults (Wolfe et al. 1998; Evers et al. 2007; Hallinger and Cristol 2011). Bird songs play a key role in reproduction because songs reliably indicate male quality and determine male attractiveness, and females use song to identify superior mates (Nowicki et al. 2002; Buchanan et al. 2003; Gorissen et al. 2005; Markman et al. 2008; Byers and Kroodsma 2009). Any disruption or alteration in song production or its perception and meaning might disrupt pair formation and successful rearing of young (Hoogesteijn et al. 2008).

Nelson’s sparrows (Ammodramus nelsoni) are classified as a conservation priority by Partners in Flight (Nocera et al. 2007). They inhabit salt marshes along the Atlantic coast from Maine into Canada and depend on these habitats for survival and reproduction (Shriver et al. 2006). Because salt marshes can accumulate high levels of Hg, birds spending their entire lives in marshes can serve as good bioindicators of Hg in the environment. Specifically, Nelson’s sparrows ingest Hg through their diet, which consists primarily of arthropods (Post and Greenlaw 2006). Hg exposure causes neurological and sensory damage that might impair song learning (Hallinger et al. 2010), and exposure also might weaken adult birds such that song performance declines (Hallinger et al. 2010).

We hypothesized that high levels of Hg in Nelson’s sparrows change characteristics of male songs. We predicted that exposure to Hg leads to lower tonal frequency of notes, decreased bout duration, and a reduced number of bouts per minute causing a decrease in song rate. Since song is costly to produce (Nowicki et al. 2002; Podos et al. 2004), exposed male Nelson’s sparrows might be unable to maintain high frequencies, and they might sing slower or shorter songs.

Materials and method

Study populations and sites

In 1995, the American Ornithologists’ Union on Classification and Nomenclature split the sharp-tailed sparrow into two species: the saltmarsh sparrow (A. caudacutus) and the Nelson’s sparrow (Hodgman et al. 2002). Their songs differ in the introductory and terminal notes, and they occupy different geographic ranges (Greenlaw 1993; Greenlaw and Rising 1994). Saltmarsh and Nelson’s sparrows overlap only in southern Maine, where the two species hybridize (Hodgman et al. 2002; Shriver et al. 2005).

Nelson’s sparrows are sexually monomorphic, exhibit natal philopatry, and do not maintain territories (Greenlaw and Rising 1994; DiQuinzio et al. 2001; Driscoll et al. 2007). Females provide sole parental care in these promiscuous birds (Greenlaw and Rising 1994; Driscoll et al. 2007). Breeding season falls between May and August, and females nest 2–3 times within a single season following flood tides of the new moon (Greenlaw and Rising 1994).

We studied Nelson’s sparrows at two locations in Maine outside of the hybrid zone. Mendall Marsh (44.59° N, 68.86° W) occupies approximately 300 ha and is located in Frankfort and Prospect, Maine, as part of the Howard Mendall Wildlife Management Area. More specifically, we studied birds in salt marshes situated along the South Branch Marsh River, a branch of the Penobscot River, 3–7 km from the Atlantic Ocean. This high salt marsh habitat consists primarily of smooth cordgrass (Spartina alterniflora), salt meadow hay cordgrass (S. patens), prairie cordgrass (S. pectinata), black grass (Juncus gerardii), and spike grass (Distichlis spicata). Invasive phragmites (Phragmites australis) and cattail (Typha angustifolia) rim the area. We collected song samples from a 200 ha area of the marsh. Previous sampling of salt marsh areas surrounding the Penobscot River indicated high levels of Hg contamination in both aquatic sediments and biota (Bodaly et al. 2009).

The other location consisted of three separate tidal, high salt marsh habitats all located on Mount Desert Island. Marshes at these study sites consist of smooth cordgrass, salt meadow hay cordgrass, prairie cordgrass, black grass, and spike grass. Bass Harbor Marsh, a 200 ha marsh, sits in Tremont, Maine (44.25° N, 68.34° W) as part of Acadia National Park. Another marsh is located on a privately owned, 10 ha section of land in Tremont, Maine (44.14° N, 68.21° W). The third marsh, 15 ha in size, is situated within Babson Creek Preserve, private property managed by Maine Coast Heritage Trust in Somesville, Maine (44.22° N, 68.19° W). Previous work revealed low levels of Hg contamination in biota (Bank et al. 2007); thus we designated them as reference sites. Salt marsh elevations for all study locations range from 0 to 3 m above sea level. We collected song samples from sections up to 100 ha within each marsh.

Data collection

At each location, Nelson’s sparrows were captured using mist nets, aged as second year (SY) or after hatching year (AHY), and 1–2 capillary tubes of blood drawn from the cutaneous ulnar vein with a sterile 26 gauge disposable needle. Capillary tubes were sealed with Critocaps®, placed in 10 cc plastic vacutainers, and stored at −20 °C for later analysis. Males were color banded for individual identification using 2.3 mm celluloid bands and metal numbered serial bands. Birds were released within 10–20 min of capture.

We identified males by their unique color bands and observed 10–15 males at each site during morning hours when they were most active and vocal (0430-1200 EST). All songs used for analysis were recorded between 9 July and 8 August 2009. We recorded song from perched males using an Optimus IMP 6000 unidirectional microphone (Model 33-3017) and a Sony Hi-MD mini disc system recorder (Model Mz-M200) from a distance of <100 m (typically 6–20 m). We opportunistically recorded each song for 10 min or until the bird stopped vocalizing or could no longer be followed. We recorded songs at the contaminated site for 100 h and at reference sites for 70 h.

Data analysis

Blood samples were analyzed within 6 months of collection using a Direct Mercury Analyzer (DMA-80 Milestone, Inc.), which uses cold vapor atomic absorbance spectroscopy to determine Hg levels in a sample. Every 20–30 samples consisted of two standard reference materials (DORM-3 and DOLT-4, National Research Council, Canada), a method blank, and a sample blank. Mean percent recoveries for total Hg of standard reference materials was 95–105 %. The factory calculated instrument detection limit (IDL) was 0.005 ng Hg.

To determine if male song differed between sites, we analyzed recordings with RAVEN 1.3 (Cornell Lab of Ornithology, Ithaca, NY), using the filter and amplify settings to produce a clean spectrogram. We analyzed specific components of song, including high and low tonal frequency (defined as the frequency at which the maximum and minimum amplitude, respectively, occurred), tonal frequency range, bout duration, gap duration between bouts, and song rate in bouts per minute. Nelson’s sparrows have a monosyllabic, discontinuous song (Marler and Isaac 1960). Because male song was not continuous, we defined a song as the total time that individual males sang, and a bout as each individual segment of song. Because Nelson’s sparrows sing only a short time compared to total performance time, we also measured gap duration between each bout. We determined bout duration as the time between the start of the song and the time immediately following the final “chip” of each segment.

To determine if Hg levels differed between contaminated and reference populations, we performed Mann–Whitney U Tests. To determine if song characteristics were related to blood Hg levels of individual males, we ran a linear mixed model with each song characteristic as the dependent variable, individual bird ID as the random effect, and log Hg, site, and the interaction of log Hg and site as fixed effects. Recording date and age were not included as fixed effects in the model. Recording date was confounded with site because song collection at the contaminated site and reference sites did not occur simultaneously. In addition, ages of adult males were classified as either SY or AHY, which allows for crossover because SY birds can fall under the AHY classification. Finally, we used linear mixed models to look at the relationship between bouts per minute, gap duration and log Hg within the reference and contaminated sites separately. We performed statistical analyses using JMP 8 (SAS Institute, Inc., Cary, NC) and SPSS version 20 (IBM Corp., Armonk, NY).


A total of 22 male Nelson’s sparrows were captured and color banded at contaminated and reference sites in Maine (n = 15 males at contaminated site, 7 males at reference site). Blood Hg levels of Nelson’s sparrows at the contaminated site were 3.9× higher than at the reference site (Mann–Whitney U test: Z = −3.66; n = 15, 7 males; p = 0.0002; Fig. 1).
Fig. 1

Median, 95 % CI and range of blood mercury levels (μg/g ww) measured in male Nelson’s sparrows from reference and contaminated sites in Maine during summer 2009

Low tonal frequency, tonal frequency range, and bout duration were not related to site or log Hg blood (p > 0.05). Log Hg blood was related to high tonal frequency, such that birds with higher blood Hg levels sang at higher maximum frequencies, whereas site and site*log Hg blood were not related with high tonal frequency (Table 1).
Table 1

Estimates of fixed effects from linear mixed models of song characteristics recorded from male Nelson’s sparrows on contaminated and reference sites in central coastal Maine during summer 2009


Song characteristic



F test


High tonal frequency




F1,58 = 1993.81





F1,58 = 1.30


 Log Hg blood



F1,58 = 5.72


 Site*Log Hg blood



F1,58 = 0.48


Bouts per minute (both sites combined)




F1,58 = 88.64





F1,58 = 2.27


 Log Hg blood



F1,58 = 0.42


 Site*Log Hg blood



F1,58 = 13.25


Bouts per minute (reference site only)




F1,14 = 14.52


 Log Hg blood



F1,14 = 3.12


Bouts per minute (contaminated site only)




F1,44 = 94.23


 Log Hg blood



F1,44 = 11.67


Gap duration (both sites combined)




F1,58 = 210.82





F1,58 = 6.08


 Log Hg blood



F1,58 = 0.19


 Site*Log Hg blood



F1,58 = 7.31


Gap duration (reference site)




F1,14 = 71.14


 Log Hg blood



F1,14 = 1.71


Gap duration (contaminated site)




F1,44 = 108.39


 Log Hg blood



F1,44 = 6.68


Each song characteristic was the dependent variable, whereas site, log Hg blood, and the interaction between site and Hg were fixed effects, and individual bird identity was a random effect in the models

Bouts per minute were not related to site and log Hg blood (p > 0.1). However, there was a significant site*log Hg blood interaction, such that the relationships trended in opposite directions (Table 1). Because all birds at reference sites may have had Hg concentrations below effects thresholds, we analyzed each site separately. We found no relationship between Hg and bouts per minute at the reference site (Table 1). However, at the contaminated site, males with higher Hg levels sang faster songs (Table 1).

Similarly, site and site*log Hg blood were significantly related to gap duration, whereas log Hg blood was not (Table 1). Males from the contaminated site had 60 % shorter gap durations than males from reference sites (contaminated: median = 5.36 s; reference: median = 9.94 s; Mann–Whitney U test: Z = 2.89, p = 0.004). Again, when we analyzed each site separately, we found no relationship between Hg and gap duration in the reference site (Table 1). Yet, males at the contaminated site that carried higher blood Hg levels sang songs with shorter gap durations (Table 1).


Nelson’s sparrows at a site contaminated with Hg had significantly higher blood total Hg levels than birds at reference sites. Because blood total Hg consists of up to 95 % MeHg in insectivorous birds, blood total Hg can serve as a proxy for MeHg levels (Rimmer et al. 2005). Adult males living on Mendall Marsh averaged 2.9 μg/g ww of Hg (range = 1.25–6.04 μg/g ww), >3× higher than males on Mount Desert Island, where all individuals had blood Hg levels of less than 0.9 μg/g ww. High sediment and biota concentrations of Hg are attributed to industrial Hg pollution from a chlor-alkali production facility (HoltraChem Manufacturing Company) operating from 1967 to 2000 along the Penobscot River (Bodaly et al. 2009). Nelson’s sparrows are philopatric and thus may return to this site yearly, where they are exposed to Hg and ingest contaminated prey (DiQuinzio et al. 2001). Since Hg bioaccumulates, blood Hg concentrations in individuals could increase each year, resulting in these high levels (Bouton et al. 1999; DiQuinzio et al. 2001; but see Thompson et al. 1991).

We predicted that Hg exposure in Nelson’s sparrows would be related to decreased tonal frequencies, bout duration, and song rate. To the contrary, blood Hg levels and site location were not related to low frequency, frequency range, and bout duration, suggesting that Hg concentrations in the blood of Nelson’s sparrows may not have affected these components of male song as predicted from a study on other species (Hallinger et al. 2010). However, individuals with higher Hg levels produced songs with higher maximum tonal frequencies. Additionally, at the contaminated site, males with higher blood Hg levels produced shorter intervals between bouts and increased the number of bouts per minute, resulting in faster songs. These results suggest a relationship between Hg levels and singing behavior in Nelson’s sparrows, but one that is manifested at higher levels of Hg. At least two physiological mechanisms could account for such relationships: (1) Hg acting as developmental stress and (2) Hg acting as an endocrine disrupting chemical (EDC).

Juvenile Nelson’s sparrows living on the polluted section of the Penobscot River may develop different song if they are exposed to Hg when song learning occurs. Later, they return to the natal, contaminated site to perform that song. Environmental stressors experienced early in life, when the brain develops, may negatively affect learning and subsequent production of song (Nowicki et al. 1998). Hg may affect development, producing observable changes in male song, because Hg penetrates the blood brain barrier and accumulates in developing brains (Wolfe et al. 1998). To our knowledge, researchers have not investigated whether Nelson’s sparrows learn their song or not. However, a congener, the grasshopper sparrow (A. savannarum), must be exposed to conspecific song for its song to develop normally, although it does not learn by imitation (Soha et al. 2009). Thus, Hg might impact normal development of song in these species.

Stress during key developmental periods leads to changes in characteristics of song such as decreased song complexity, fewer and shorter song bouts, and decreased total song duration in several species (Buchanan et al. 2003, 2004; Spencer et al. 2003; Bischoff et al. 2009). These changes appear to minimize the costs associated with song. Philopatric Carolina wrens, house wrens, and song sparrows exposed to Hg on natal sites along the Shenandoah River, Virginia USA, sang at lower frequencies and produced a lower diversity of notes than birds from reference sites (Hallinger et al. 2010). However, songs of eastern phoebes (Sayornis phobe) from a Hg contaminated site did not differ in frequency compared to songs from reference sites (Hallinger et al. 2010). Eastern phoebes do not learn their songs, but, like Nelson’s sparrows, their songs lack large repertoires and note complexity (Hallinger et al. 2010). We found that Nelson’s sparrows carrying higher Hg levels produced songs with higher maximum frequency but did not change the minimum frequency. Production of low frequency notes can signal size and condition to potential mates and rivals (Slabbekoorn and Ripmeester 2008). Perhaps Nelson’s sparrows experiencing Hg toxicity lack the necessary power to produce low frequency notes, resulting in songs sung at higher frequencies. Alternatively, another factor that we did not measure, such as subtle differences in habitat or genetics could influence song frequencies in this species. Manipulative studies could untangle causal patterns from correlational patterns.

In addition to producing relatively simple songs, Nelson’s sparrows sing for only a small percentage of the total performance time (Marler and Isaac 1960). Perhaps the difference in energy costs between producing higher or lower frequency notes or singing shorter or longer bouts of song is negligible in this species compared to birds with richer song repertoires and multiple note types. Alternatively, Hg may impact activities indirectly related to singing, such as hearing songs produced by other birds (Hallinger et al. 2010). Intervals might be shortened if males cannot hear other males’ songs.

Another explanation for changes in male song involves the effects of Hg as an EDC. Hg is categorized as an EDC because it affects thyroid and reproductive hormones, which influence the initial differentiation between male and female brains and increase HVC volume in males (Clotfelter et al. 2004; Zala and Penn 2004; Hoogesteijn et al. 2008; Markman et al. 2008). European starlings (Sturnus vulgaris) dosed with EDC spend more time singing, sing more and longer bouts, and produce a larger repertoire size compared to controls. Furthermore, their brains exhibit larger HVC volumes (Markman et al. 2008). Hg exposure in Nelson’s sparrows might affect hormones that increase HVC volume and thus increase song rate. Determining the developmental pathways that alter song production warrants further investigation.

Perhaps males sing faster due to other factors, such as population density. Marsh wrens (Cistothorus palustris) and sedge wrens (C. platensis) enlarge their song repertoires when population densities increase (Kroodsma et al. 2001). Increased densities may result in more intense interactions between males or result in fewer resources, causing stress that can alter song (Kroodsma et al. 2001). We did not measure population densities; therefore, we cannot determine if males sang faster due to higher density at the contaminated site.

Acoustic properties of the habitat, urbanization, and bill morphology can alter song production (Bermúdez-Cuamatzin et al. 2011; Slabbekoorn and Ripmeester 2008). However, factors associated with urbanization or differing habitats probably did not affect song of Nelson’s sparrows in the study populations. Both study sites were located in rural areas near two-lane roads, and neither site was located near noisy regions such as airports or major highways. Both sites contained similar vegetation and food resources (J. McKay, personal observations), although the birds may perceive differences that humans cannot detect. Both populations also had similar bill morphology (J. McKay, unpublished data); however, sample sizes at the reference site were low.

Song differences also may be due to geographic variation in song. However, the study of dialects among birds reveals conflicting information. Dialects correlate to genetic boundaries in some species (i.e., swamp sparrows; Melospiza georgiana, and song sparrows) but not in others (i.e., rufous-crowned sparrows; Aimophila ruficeps), and local dialects can be learned on natal territories or after dispersal (Slabbekoorn and Smith 2002). Birds are sensitive to detail in complex songs, and in many species females prefer local dialects; however, birds detect foreign dialects at different distances outside their home ranges (Searcy et al. 2002; Slabbekoorn and Smith 2002).

We could not find any dialect studies on birds in the Ammodramus genus. In addition, no studies focused on gap duration as a marker for dialect differences. The sites were approximately 80 km apart, which was relatively close compared to studies in other species (Searcy et al. 2002; Slabbekoorn and Smith 2002). Most importantly, the relationships between blood Hg, gap duration, and bouts per minute within the contaminated population suggest that differences between the two populations are not due to dialect.

Knowledge of the effects of Hg on song may lead to better understanding of the pollutant’s effects on behavior, in general, including broader impacts of fitness. Ultimately, we need to measure reproductive success of these males to determine Hg’s effects not only on individual fitness but also Hg’s subsequent impacts at the population level. Nelson’s sparrows living along the Penobscot watershed only occur in the largest adjoining section of marsh along the Penobscot River and show the highest blood Hg levels of several songbirds and shorebirds (Bodaly et al. 2009). Because Hg concentrations in birds living on Mendall Marsh are exceptionally high compared to other contaminated wetlands along the lower Penobscot, Mendall Marsh remains an area of principal concern, and these Hg concentrations are high enough to be a problem for both humans and wildlife (Bodaly et al. 2009).

In conclusion, song production is sensitive to Hg exposure (Hallinger et al. 2010; this study); thus, we can use bird song to assess impacts of this environmental contaminant. Results of this study raise questions about the effects of Hg exposure on brain development in males and on reproductive success. Knowledge of Hg’s sublethal effects on birds may lead to better understanding of the long-term ramifications of this pollutant in ecosystems and to more effective conservation of wildlife.


We thank J. Walker and D. Evers for their advice and direction; staff at Biodiversity Research Institute, particularly O. Lane, for field assistance, analysis of blood Hg levels, and field housing; D. Cristol for comments on a previous draft; K. O’Brien and N. Pau for insight into the birds; and E. Adams, J. Atwood and K. Poulin for help with statistics. We also thank the National Park Service and B. Connery for use of their land and equipment, field housing, and knowledge; and the Maine Coast Heritage Trust, Tremont residents, and Maine Department of Inland Fisheries and Wildlife for permission to access their lands. Funding was provided by the University of Southern Maine’s Biology Department and Biodiversity Research Institute. All procedures were approved by the University of Southern Maine’s Institutional Animal Care and Use Committee (#03808-01) and comply with current Federal laws.

Copyright information

© Springer Science+Business Media, LLC 2012