Chemosphere 119 (2015) 163–170

Contents lists available at ScienceDirect

Chemosphere

journal homepage: www.elsevier .com/locate /chemosphere

An evaluation of the use of reptile dermal scutes as a non-invasivemethod to monitor mercury concentrations in the environment

http://dx.doi.org/10.1016/j.chemosphere.2014.05.0650045-6535/� 2014 Elsevier Ltd. All rights reserved.

⇑ Corresponding author. Tel.: +61 429088813.E-mail address: laris.schneider@gmail.com (L. Schneider).

Larissa Schneider a,⇑, Sam Eggins a, William Maher a, Richard C. Vogt b, Frank Krikowa a, Les Kinsley c,Stephen M. Eggins c, Ronis Da Silveira d

a University of Canberra, Institute for Applied Ecology, Kirinary St., Canberra, ACT, Australiab Instituto Nacional de Pesquisas da Amazônia, Coordenacao de Biodiversidade, Av. André Araujo 2223, Aleixo, CEP: 69080-971 Manaus, AM, Brazilc Australian National University, Research School of Earth Sciences, Canberra 0200, Australiad Universidade Federal do Amazonas, Instituto de Ciências Biológicas, Rua General Rodrigo Otávio Num. 3000, Mini-Campus Coroado, CEP: 66077070 Manaus, AM, Brazil

h i g h l i g h t s

� Keratin is a better predictor of exposure to Hg than muscle and bone.� Keratin layer only should be used to study Hg in epidermis of reptiles.� Disagreement between previous studies is attributed to inconsistency in methods.� Carapace growth rings are useful for estimating Hg bioaccumulation over time.

a r t i c l e i n f o

Article history:Received 6 March 2014Received in revised form 3 May 2014Accepted 5 May 2014

Handling Editor: I. Cousins

Keywords:Hg historyTurtleCaimanCarapaceLaser ablationAmazon

a b s t r a c t

Reptiles are ideal organisms for the non-invasive monitoring of mercury (Hg) contamination. We haveinvestigated Hg bioaccumulation in tissue layers of reptile dermis as a basis for establishing a standard-ized collection method for Hg analysis. Tissue samples from freshwater turtle species Podocnemis unifilisand Podocnemis expansa and caiman species Melanosuchus niger and Caiman crocodilus, all from theAmazonian region, were analysed in this study. We first tested the relationships between Hg concentra-tions in keratin and bone to Hg concentrations in muscle to determine the best predictor of Hg concen-tration in muscle tissue. We then investigated the potential for measuring Hg concentrations across turtlecarapace growth rings as an indicator of longer term changes in Hg concentration in the environment. Hgconcentrations were significantly lower in bone (120 ng g�1 caimans and 1 ng g�1 turtles) than keratin(3600 ng g�1 caimans and 2200 ng g�1 turtles). Keratin was found to be a better predictor of exposureto Hg than muscle and bone tissues for both turtles and caimans and also to be a reliable non-invasivetissue for Hg analysis in turtles. Measurement of Hg in carapace growth rings has significant potentialfor estimating Hg bioaccumulation by turtles over time, but full quantification awaits developmentand use of a matrix-matched reference material for laser ablation ICPMS analysis of Hg concentrationsin keratin. Realising this potential would make a valuable advance to the study of the history of contam-ination in mining and industrial sites, which have until now relied on the analysis of Hg concentrations insediments.

� 2014 Elsevier Ltd. All rights reserved.

1. Introduction

Hg toxicity to reptiles is of particular concern because manyreptilian species are already experiencing population declines(Congdon et al., 1994; Vogt, 2008; Schneider et al., 2011b, 2013).Turtles and caimans have been proposed as ideal organisms for

monitoring Hg contamination in aquatic ecosystems because oftheir ecological and life-history attributes (Heaton-Jones et al.,1997; Golet and Haines, 2001; Schneider et al., 2009, 2010,2011a; Bezerra et al., 2013). Many turtle and caiman species arelong-lived, occupy a wide-range of habitats, occur in high densitiesin a variety of aquatic habitats, and feed at the higher trophic levels(Da Silveira et al., 1997; Da Silveira and Magnusson, 1999; DaSilveira et al., 2008, 2013; Schneider et al., 2010). In addition, thedermis of most reptiles is hard, coarse, and scaled (Toni et al.,

164 L. Schneider et al. / Chemosphere 119 (2015) 163–170

2007), features that make it suitable for non-invasive Hg studies(Schneider et al., 2011a; Bezerra et al., 2012). In this study the car-apace of turtles and skin of caimans are referred to collectively asdermis and the outermost keratin layer as the epiderm.

Reptile dermis have been shown to be useful for the reliablenon-lethal sampling and monitoring of Hg concentrations in theenvironment (Rainwater et al., 2007; Schneider et al., 2011a), how-ever, care is required to sample the appropriate layer of dermis foranalysis. Turtles and caiman dermis are composed of an exoskele-ton which consists of bony scutes that underlie the epidermalscales of the dorsal surface, which is composed mainly of beta-keratins (Alibardi, 2003; Toni et al., 2007). Hg concentrations rarelyhave been analysed in the bones of reptiles, but when analysed, Hgconcentrations were comparatively low (Grillitsch and Schiesari,2010). Hg localization in turtle and crocodilian dermis is, therefore,expected to be in the epidermal scales as a result of Hg affinity forthe sulfhydril groups of proteins (Appelquist et al., 1984).

Differences in reported Hg concentrations in exoskeleton andepidermal scales of turtle carapace and caiman skin still need tobe clarified as previous studies have used different layers of reptiledermis for Hg analysis. This may have resulted in inappropriatecomparisons and incorrect conclusions. Accordingly, this studyhas two main objectives: (1) to investigate Hg bioaccumulation indifferent tissue layers of caiman and turtle dermis as a basis forestablishing a standardized method for sampling and Hg analysis;and (2) to assess the relationships between Hg concentrations inkeratin and bone to those in muscle tissues to test which of thesenon-invasive sampled tissues is the better predictor of Hg concen-tration in muscle (the tissue usually used for Hg analysis). We alsotested the potential for measuring Hg concentrations in turtle car-apace growth rings to gain an indication of Hg concentrationchanges in the environment over longer (historical) timescales.Perspectives are also presented to help design future studies ofHg concentrations in turtle and caiman epidermis.

2. Methods

2.1. Study Species

Tissue samples from two freshwater turtle species (Podocnemisunifilis and Podocnemis expansa) and two species of caimans(Melanosuchus niger and Caiman crocodilus) were analysed in thisstudy. The turtle species belong to the Podocnemididae familyand are opportunistic omnivores that feed almost exclusively onplant material (Vogt, 2008). The two caiman species are carnivo-rous and top predators (Da Silveira and Magnusson, 1999). Allfour species are listed in Appendix II of the Convention on Inter-national Trade in Endangered Species (CITES) and on The IUCNRed List of Threatened Species (IUCN, 2013). P. unifilis is listedas vulnerable (VU) and P. expansa, C. crocodilus and M. niger asleast concern (LC).

2.2. Sampling sites

All specimens were collected in the Rio Purus of the AmazonBasin (4�45.6020 S, 62�11.3850 W, WGS84; Fig. 1). The Rio Purusis approximately 3000 km long with headwaters located in thePeruvian Andes that drain into the Rio Amazonas, 186 kmupstream of Manaus, the capital of the Amazonas State, Brazil.Most of the lower Rio Purus Basin is occupied by the Piagaçú-PurusSustainable Development Reserve and two Indian Territories cov-ering 1,008,167 ha (De Deus et al., 2003). This region incorporatesa large interdigitated mosaic of unflooded (terra firme) and annu-ally flooded forests that are inundated by white (varzea) or blackwater (igapo). The study area merges a large confluence of

floodplain forests under the influence of both the Rio Amazonasand Rio Purus.

Identification of Hg sources in the Purus River, whether fromnatural origin or from illegal gold mining, has been difficult dueto the lack of studies on the origins of Hg. As for many parts ofthe Amazon, the link between Hg contamination and gold miningis not always clear and has been the subject of ongoing debate(Fadini and Jardim, 2001).

2.3. Animal collections

Animals were collected in March 2008. Turtles were trappedusing trammel nets (Vogt, 2008) and caimans were collected atnight using locking cable snares and ketch-all animal restrainingpoles (Da Silveira et al., 1997). All individuals were measured,sexed and then euthanized prior to collection of tissue samples.

Turtles were sexed visually using sexually dimorphic patterningand morphological features. Podocnemis spp. males have a distinc-tive bright red or reddish orange pattern on the head (Vogt, 2008).Straight carapace length (SCL), the distance from the outside edgeof the nuchal at the front middle of the carapace to suture of the12th marginal scutes, was measured as an index of specimen sizeusing vernier callipers or a ruler (Vogt, 2008).

The snout-vent length (SVL) of caimans was measured as anindex of size. They were sexed by inspection of the protrusion ofmale genital organs from the cloaca and compared to the similarfemale clitoris (Ziegler and Olbort, 2007).

2.4. Tissue sampling protocols

Turtle carapace samples were collected by removing the eighthright marginal scute (dermis and epidermal scale). The choice ofscute to be collected was based on the scute location of previousstudies for comparison purposes. As in this study we aimed to col-lect a carapace sample containing bony and keratin tissues of deadanimals, we used a stainless steel hand saw to obtain a thick cara-pace sample (Supplementary Fig. 1). When collecting the keratinlayer only, a stainless steel biopsy tool was used to scrape the epi-dermis layer from the outermost edge of scutes without damagingthe animal.

A standardized scute area should be determined for sample col-lection as scutes have different patterns of growth and the possibil-ity of different patterns of mercury bioaccumulation (especiallymarginal and coastal scutes). The advantage of using marginalscutes is that they appeared to have keratin of sufficient thicknessand texture to provide sufficient sample mass. It is important toavoid scraping too deeply to prevent contaminating the samplewith untargeted tissues and causing injury to the turtle. Carapacesamples were cleaned using a plastic scrubbing pad, rinsing liber-ally with high purity deionised water.

Caiman skin samples were taken from the dorsal section of thebase of the tail. Three triangles (2 cm in size) of dermis (bone andepidermal scale) were collected using a stainless steel biopsy tool.No hand saw was needed for skin collection of caimans as this is asofter material than turtle carapace (Supplementary Fig. 2)

Dermis samples of turtles and caimans were sealed in labelledplastic bags. The samples were then frozen and transported toINPA (Manaus, Brazil), where they remained frozen in sealedcontainers until transported to the University of Canberra (ACT,Australia). In the laboratory, keratin fragments of both turtlesand caiman samples were washed with deionised water for15 min in an ultra-sonic bath to remove algae and other incrustingresidues and sediment grains, then freeze-dried for 48 h (LabconcoFreezone Plus 6). Freeze drying was required to allow the separa-tion of the keratin layer from the bone part as the drying processdisaggregates the keratin from the bone.

Fig. 1. Location of the sample site in the Amazon Basin (arrow point) and within Brazil (right bottom corner).

L. Schneider et al. / Chemosphere 119 (2015) 163–170 165

2.5. Mercury analysis

Each dermis sample (carapace of turtles and skin of caimans)was divided into three subsamples: keratin layer, bone layerand a mixture of ground keratin and bone. These subsampleswere digested following the procedure of Baldwin et al. (1994);tissue was weighed (0.07 g) and placed into 7 ml polytetrafluroac-etate (PFA) digestion vessels (A.I.Scientific, Australia), to which1.0 ml of concentrated nitric acid (Suprapur, Merck KGaA, Ger-many) was added. Samples were digested in a microwave oven(CEM MDS 2100, USA) at 6000 W for 2 min and 314 W for45 min. Hg analysis was undertaken by inductively coupledplasma mass spectrometry (ICPMS) using an Elan DRCe (Maheret al., 2001).

Procedural blanks and certified reference material were alsoanalysed with each sample batch (2 per 20 samples). Measuredvalues for the NRCC (National Research Council of Canada) DORM2 (4420 ± 330 ng g�1) and TORT 2 (256 ± 65 ng g�1) were in agree-ment with certified values (4640 ± 260 and 270 ± 60 ng g�1

respectively).Analysis of Hg in growth rings of turtle scutes (keratin layer)

was made by laser ablation inductively coupled plasma mass spec-trometry (LA-ICPMS), using an ArF (k = 193 nm), 20 ns pulse widthLambda Physik LPFpro 205 excimer laser and custom-built Helexsample cell and transport system (for more details refer to Egginset al., 1998). Laser sampling was performed in a He atmosphereat a repetition rate of 5 pulses s�1 using a 100 lm diameter sizespot, which was scanned across the sample at the rate of50 lm s�1. The laser energy density at the sample surface wasapproximately 5 J cm�2, which removes approximately 0.5 lmthick layer per pulse from the sample surface.

A pre-analysis laser ablation run was undertaken to clean thesample surface prior to analysis. This procedure involved scanninga 240 lm diameter spot along the analysis track at 1000 lm s�1

using a laser repetition rate of 10 pulses s�1. This procedure wasrepeated and resulted in the removal of 3–4 lm from the samplesurface.

The standard reference material NIST 612 glass was used forsemiquantitative estimation of a Hg sensitivity factor based onsensitivity factors for elements of similar mass and ionization

potential. It should be note that the NIST 612 glass does notpossess a certified value for Hg, and is not an appropriate matrix-matched reference material nor does it contain a suitable internalstandard element for analysis of a keratin matrix (e.g. C). As a con-sequence the NIST 612 glass was only used herein to provide aqualitative measure of the recovery and changes in relative con-centrations of Hg and other elements. There is a need to developand employ an appropriate calibration standard for the analysisof Hg concentrations in keratin material and this is discussed latterin this manuscript.

2.6. Statistical analysis

Data were analysed using the statistical package R (RDevelopment Core Team, 2013) with p 6 0.05 employed as thelevel of statistical significance.

A one-way ANOVA was used to test for significant differences oflog-transformed Hg concentrations between tissues (keratin, boneand mixed). Post-hoc Tukey – HSD tests were performed to identifythe tissues in which Hg concentration differed significantly(p < 0.05).

A linear regression model was used to determine the relation-ship between animal size ratio and log transformed Hg concentra-tions. Size ratio was used to allow inclusion of both species in thesame regression for each group (turtle and caiman) eliminating theeffect of size. The size ratio for each individual of each species wascalculated by dividing the sampled animal length by the maximumreported straight carapace length (SCL) for turtle and snout-ventlength (SVL) for alligators, multiplying by 100.

A pairwise correlation was performed to test if log10 Hg concen-trations in keratin and/or bone were related to Hg concentrationsin muscle. The same analysis was also performed to test which ofthe three tissues (keratin, bone or muscle) provided the best rela-tionship with animal size.

For all analyses, no difference in Hg concentration betweensexes was assumed. This assumption could not be tested in thisstudy because of unequal sample sizes between sexes, but hasbeen tested previously with no difference in Hg concentrationfound between genders (Schneider et al., 2009, 2010, 2012;Eggins, 2012).

166 L. Schneider et al. / Chemosphere 119 (2015) 163–170

3. Results

3.1. Hg concentrations in different dermal layers

Number of specimens, size and Hg concentrations are given inTable 1.

All four species show a significant difference in log10-trans-formed Hg concentrations at the p < 0.05 level among the three tis-sues: F(2, 18) = 23.4, p < 0.001 for C. crocodilus; F(2, 36) = 115.1,p < 0.001 for M. niger; F(2, 12) = 15.51, p < 0.001 for P. expansa;and F(2, 27) = 50.7, p < 0.001 for P. unifilis.

Post hoc comparisons using the Tukey HSD test indicated thatthe mean log-transformed Hg concentration was significantly dif-ferent between all tissues and for all species, but not between boneand mixed samples in C. crocodilus (Fig. 2, Supplementary Table 1).

3.2. Relationships of Hg concentrations in keratin, bone and muscletissues

The pairwise correlation indicated that there was no significantdifferences of log10 Hg concentration among different caiman tis-sues, with p > 0.05 for all pairs (Supplementary Fig. 3).

Pairwise correlations for turtles, gave a significant relationshipbetween log10 Hg concentration in keratin and muscle r(15) = 0.65, p < 0.05) (Supplementary Fig. 4). For all other tissuepairs the correlations were not significant (p > 0.05).

3.3. Relationship of Hg concentrations in keratin and muscle tissues toanimal size ratio

Size ratio predicted Hg concentrations in caiman keratin;b = 29.48, t(18) = 2.16, p < 0.05. Size ratio also explained a signifi-cant proportion of variance in Hg concentration of caiman keratin;R2 = 0.20, F(1, 18) = 4.65, p < 0.05 (Fig. 3A).

Size ratio, however, did not predict Hg concentrations in muscleof caimans; b = 35.46, t(12) = 1.64, p > 0.05 (Fig. 3B).

Size ratio predicted Hg concentrations in turtle carapace kera-tin; b = 17.22, t(13) = 2.50, p < 0.05. Size ratio also explains a signif-icant proportion of variance in Hg concentrations of keratin;R2 = 0.27, F(1, 13) = 6.29, p < 0.05 (Fig. 4A).

Size ratio, however, did not predict Hg concentrations in muscleof turtles; b = 10.81, t(10) = 1.64, p > 0.05 (Fig. 4B).

3.4. Analysis of Hg concentration variation across turtle keratingrowth rings

The lack of a certified reference material to ablate keratin andanalyse Hg in this study made possible only the generation of semiquantitative data (Fig. 5). The decrease in relative Hg concentrationobserved in the 30 mm distance ablated corresponds to �15 years(Fig. 5).

Table 1Average size (cm) and Hg concentration x ± SD (min – max) (ng g�1 dry mass) in keratin, bBasin in Brazil.

Scientificname

Common name N Size (cm) Hg muscle (lg

Caimancrocodilus

Spectacled caiman 7 75 ± 10 (66–94) 234 ± 144 (132

Melanosuchusniger

Black caiman 13 106 ± 28 (87–191) 177 ± 102 (56–

Podocnemisexpansa

Giant Amazon RiverTurtle

5 48 ± 10 (37–61) 100 ± 48 (45–1

Podocnemisunifilis

Yellow-spottedSideneck Turtle

10 37 ± 5.2 (30–45) 13 ± 12 (4–43

4. Discussion

4.1. Hg concentrations in different dermal layers

A significant difference in Hg concentrations for bony scutesand epidermal scales was found for all four species. Bone Hg con-centrations were mostly too low for precise analysis and not corre-lated to other tissue Hg concentrations (Fig. 2). It follows thatkeratin should be used in preference to bone for study of Hg con-centrations in dermis of reptiles.

Comparative data on Hg concentrations in bones of turtles andcaimans is largely lacking, with the only other study on Hg in thebone of reptiles (Sakai et al., 2000) reporting low Hg concentrationsin bone (0.6 ng g�1 in the bones of Caretta caretta and 2 ng g�1 inChelonia mydas).

4.2. Relationship between Hg concentrations in keratin, bone andmuscle and the potential for non-invasive sample collection

4.2.1. CaimansOur results show that the use of scutes as a non-lethal sampling

method to predict Hg concentrations in muscle is not reliable for C.crocodilus and M. niger. Neither keratin nor bone Hg concentrationswere good predictors of Hg concentrations in muscle tissues ofthese two species (Supplementary Fig. 3). If a significant correla-tion had been found, a small amount of dorsal scute could havebeen collected without harming the animal, an important consid-eration when studying endangered species or populations.

It is not clear why Hg concentrations in keratin and muscle tis-sues do not show a significant correlation. A review of the literaturereveals significant disagreement between published results withsome studies reporting significant positive correlations for Hg con-centrations in the scales and muscle tissue of Alligator mississippien-sis (e.g. Burger et al., 2000), while others find scale analysis is not areliable nonlethal monitoring technique for Hg concentrations incrocodilians (Heaton-Jones et al., 1997 and this study). A reasonfor this inconsistency might be the use of different layers of thescute for Hg analysis. Burger et al. (2000) used a mixture of boneand keratin, while other studies have not specified the layer used.This precludes any definitive conclusions being made at this timeas to why different Hg concentration relationships were found.

Further studies are needed to investigate the feasibility of usingcrocodilian dermis (following the method proposed in this study ofonly using keratin layer) as a feasible non-invasive method to char-acterise Hg concentrations in these animals. We also call uponfuture studies to collect keratin tissue and specify the layer ana-lysed in their methods.

4.2.2. TurtlesOur results show that the use of keratin from scutes of turtle

carapaces is a reliable non-invasive method to study Hg concentra-tions in turtles. Keratin was found to be a good predictor of Hg

one and mixed samples of turtle and caiman dermis in the Rio Purus of the Amazon

g�1) Hg keratin (lg g�1) Hg bone (lg g�1) Hg mixed tissues(lg g�1)

–447) 3350 ± 2143 (500–7150) 153 ± 121 (40–370) 348 ± 298 (70–770)

371) 3846 ± 2815 (1100 10400) 80 ± 93 (20–380) 205 ± 158 (70–570)

36) 2782 ± 1683 (100–4200) 2 ± 2.7 (BDL - 5) 434 ± 352 (50–920)

) 1630 ± 1438 (100–4800) 0.5 ± 1.6 (BDL - 5) 270 ± 271 (0–700)

Fig. 2. Log10 Hg concentration (ng g�1 dry mass) in keratin layer, bone layer and mixed (keratin + bone layer) samples of turtle carapace and caiman dermis from the PurusRiver, Amazonas, Brazil. Bar whiskers show the 10th and 90th percentiles, the box defines the 25th and 75th percentiles and the solid line in the box shows the median Hgconcentration. The black circles indicate values that are more extreme than the 1.5 inter-quartile range.

Fig. 3. Relationship between log10 Hg concentration (ng g�1) and size ratio of caimans from the Purus River, distinguishing between (A) keratin and (B) muscle.

Fig. 4. Relationship between log10 Hg concentration (ng g�1) and size ratio of turtles from the Purus River, distinguishing between (A) keratin and (B) muscle. Note the log10

scale on the Y axis.

L. Schneider et al. / Chemosphere 119 (2015) 163–170 167

concentrations in muscle tissues (Supplementary Fig. 4), whereasbone Hg concentrations were not. The lack of a significant relation-ship between Hg concentrations in bone and muscle tissues wasexpected given Hg concentrations in bone are low (mostly belowdetection limits).

Our results corroborate the findings of previous studies that havereported a significant positive relationship between Hg concentra-

tions in carapace scutes and muscle tissues (Sakai et al., 2000;Golet and Haines, 2001; Day et al., 2005; Turnquist et al., 2011).Scute is a suitable material for Hg monitoring in turtles as this pro-teinaceous material contains high Hg concentrations that are highlycorrelated with muscle Hg concentrations in the same animal.

The absence of a significant correlation in Hg concentrationsbetween turtle scute and muscle tissues was reported by

Fig. 5. Measured 202Hg counts as a function of distance in turtle carapace usinglaser ablation ICP–MS. The laser ablation track was conducted from the oldestgrowth ring to the youngest.

168 L. Schneider et al. / Chemosphere 119 (2015) 163–170

Schneider et al. (2010) and Green et al. (2010), however, theseresults likely reflected the use of different carapace layers for anal-ysis; one analysed Hg in a mixture of bone and keratin tissues(Schneider et al., 2011a) and the other did not specify the layerused (Green et al., 2010).

4.3. Relationships of Hg concentrations of keratin and muscle tissueswith animal size ratio

We found a positive relationship between size ratio and Hg con-centrations in keratin of scutes for both caiman and turtles, but notfor muscle Hg concentrations. It follows that keratin Hg concentra-tions is a better predictor of exposure to Hg than Hg concentrationsin muscle tissues. Since beta-keratins from caimans and turtleepidermis contain high amounts of amino acids (Alibardi, 2003,2004; Toni et al., 2007), there is an abundance of potential bindingsites for Hg within cells. As a result of the affinity of Hg to keratin,Hg is immobilized in the keratin layer and better reflects Hg expo-sure over time than muscle tissue, even though muscle tissueshave been traditionally used to monitor Hg concentrations in theenvironment.

Previous studies have reported a significant positive correlationbetween Hg concentration in scute and size in A. mississippiensis(Yanochko et al. 1997; Burger et al., 2000) but not in Crocodylusmoreletii or Crocodylus acutus (Rainwater et al., 2007). The lack ofrelationship between size and Hg concentrations in some crocodilespecies may be a result of feeding behaviour and different stages ofcaiman life history (Schneider et al., 2012). Again, the layer of tis-sue used from the scute has not been specified, and precludes anyfurther definitive comparison of results between studies.

Few studies have examined the relationship between size andHg concentration in keratin, as the most often analysed tissue forthis assessment has been muscle. A study of Chelydra serpentinahas shown that as turtle size increased, Hg concentrationsincreased in scutes. However, the same relationship was not found

for turtles in water bodies with low Hg concentrations (Turnquistet al., 2011). A significant negative correlation between the sizeof the sea turtle Chelonia mydas and Hg concentrations in muscletissues was reported by Bezerra et al. (2012) and the authors notedthe probable cause to be the different diets of juveniles and sub-adults/adults.

4.4. Evaluations of previous studies on Hg concentrations in dermis ofturtles and crocodilians

Our results suggest that it is necessary only to collect keratinlayers of turtles and crocodilians to analyse Hg concentrations fordetermining environmental mercury inputs. Previous studies onturtle carapace and crocodilian dermis that did not specify thelayer used for Hg analysis should not be used for comparison ofHg concentrations.

A literature review revealed 31 species of turtles had their car-apace Hg concentration measured (Supplementary Table 2). Ofthese, only nine species had the keratin layer collected separatelyand clearly specified in the manuscript. Six species had their cara-pace Hg concentration measured, but the entire carapace sectionwas used, mixing bone and keratin layers. For the other 16 species,results were published without specifying the carapace layer usedfor Hg analysis.

The lack of specified sampling parameters is even greater forcrocodilians (Supplementary Table 3). Of the seven studies thathave measured Hg concentrations in dermis, only one specifiedthe tissue layer used for analysis, in that case being a mixture ofkeratin and bone.

It is concluded that the inconsistent sampling methods used formeasuring Hg concentrations in reptiles dermis have generatedresults that make comparisons difficult. We call on researchers toseparate and specify the tissues analysed.

4.5. Analysis of Hg concentrations through time in turtle keratingrowth rings

The qualitative data generated from the scute of an individualP. unifilis shows a decrease in Hg concentration over a 30 mmdistance.

Growth rings vary in size in direct relationship to the size andgrowth rate of the turtle. Immature turtles generally have largergrowth rings, 4–6 mm, whereas after maturity growth slows suchthat rings are only 1–2 mm thick (Legler and Vogt, 2013). Consid-ering this change in growth ring thickness and assuming thatgrowth rings are annual, the turtle carapace distance analysed bylaser ablation here would correspond to �15 years (Fig. 5).

These results are broadly consistent with the history of Hg con-tamination in the region. The first data on Hg contamination in theBrazilian Amazonia were reported in the 1980s, following a decadeof small-scale artisanal gold mining (Martinelli et al., 1988). Goldproduction resulted in atmospheric and aquatic Hg contaminationas metallic Hg was directly discharged in the river during the amal-gamation process and thereafter, the amalgam was burned torecover gold (Pfeiffer et al., 1993). Artisanal gold mining and asso-ciated Hg contamination has since decreased and this is expectedto be reflected in organisms of the region.

No definitive conclusions about the magnitude of Hg concentra-tions, however, can be drawn from these results as they are onlyqualitative. Our aim was to explore the potential of using laserablation to analyse changes in Hg concentrations in turtle carapac-es over time. To date, we have only been able to produce qualita-tive data in the absence of a suitable reference material thatcontains Hg in a similar matrix to keratin.

Faced with this shortcoming, we recommend that future studiesaim to develop a suitable reference material. This could be

L. Schneider et al. / Chemosphere 119 (2015) 163–170 169

produced from the keratin layer of a whole turtle carapace, blend-ing it to homogenise the sample and pressing this material into apellet form in a mechanical press. The Hg concentration of thematerial could be characterised using acid digestion as far as otherreference materials such as DORM-2.

Turtles are long lived animals, and it is not uncommon toencounter individuals that are 50 years old and potentially morethan 100 years old (Legler and Vogt, 2013). Using such matureturtles, it may be possible to determine changes in Hg bioaccumu-lation after commencement of mining and industrial activities overlong periods. This approach needs to be tested in an area where ahistory of Hg contamination can be matched with changes in Hgconcentrations in the growth rings.

While Hg concentrations in sediment cores can provide a his-torical record of contamination in sediments (Hollins et al., 2011;Schneider et al. in press), we call for researchers to explore theuse of turtle carapaces to understand the effects of anthropogenicactivities on changes of Hg bioaccumulation rates in organisms.

4.6. Limitations to and considerations for the use epidermis of turtlesand caimans to study Hg bioaccumulation

4.6.1. Morphological differencesCarapace morphology should be taken into consideration when

choosing a turtle species to study Hg concentrations as they vary inkeratin concentration and composition. The mode of carapacedevelopment determines the protein make up of turtle carapace(Alibardi, 2004) and consequently also the potential for Hg bindingto the epidermis. It is therefore necessary to understand the modeof carapace development of any studied species to understand theprocess of Hg bioaccumulation and to draw appropriate compari-sons between species.

4.6.2. Information on growth rings developmentIf Hg bioaccumulation is to be understood from concentration

variations across turtle carapace growth rings, the process of for-mation of growth rings needs to be known for a given species.However, it remains unclear as to what triggers the formation ofnew scute layers and whether growth rings actually representspecific periods or events. Existing studies suggest new scute lay-ers formed primarily after a major cessation of growth, such asduring hibernation, prolonged drought, or heavy rainfall (Berry,2002; Wilson et al., 2003; Legler and Vogt, 2013; SupplementaryFig. 5). This interpretation of Hg concentrations measured ingrowth rings requires a more thorough knowledge of formation.

4.6.3. Carapace sheddingIn some turtles, the superficial part of the hard corneous layer is

shed as thin but compact flakes in spring-summer so that the scuteis enlarged. This process also occurs through the formation of ashedding plane. The keratin of scutes of turtle species which shedtheir corneous layer cannot be used for understanding long-termHg bioaccumulation behaviour as it will usually reflect an exposuretime of one year. This is the case with many Emydid turtles such asChrysemys picta and all species of Graptemys, Pseudemys, andTrachemys, Dermatemys mawii (Vogt 1981; Legler and Vogt, 2013).

5. Conclusions

We have demonstrated that it is necessary only to collectkeratin layers of turtles and crocodilians dermis to biomonitorenvironmental mercury and characterise Hg bioaccumulation inthese reptile groups. Our results contribute to a growing body ofevidence that indicate reptile epidermis is suitable for Hg monitor-ing and establishing non-invasive sampling methods for predicting

Hg concentrations in reptile muscle tissues. Hg concentrations inkeratin were found to be a reliable tissue for measuring Hg bioac-cumulation in turtles, and a better predictor of exposure to Hg thanHg concentrations in muscle tissues for both turtles and caimans.Disagreement between our results and those of previous studiesmay be attributed to inconsistency in dermis sampling methods.Only 9 of 31 papers reporting Hg concentrations in turtle carapaceand only 1 in 7 papers reporting Hg concentrations in crocodiliandermis have specified whether keratin or mixed samples wereused for Hg analysis. Preliminary LA-ICPMS results indicate itmay be feasible to determine Hg bioaccumulation by turtles overperiods exceeding 50 years. The acquisition of quantitative datausing this approach will require development of a reference matrixmaterial for calibration of measured Hg concentrations in keratin.Consideration of turtle carapace morphology is required whenselecting target species for such study, with hard-shelled turtlespecies that display regular growth ring formation recommended.It is also important to select species that do not shed their carapa-cial scutes in order to predict Hg bioaccumulation over long peri-ods. The ability to determine Hg bioaccumulation in turtles overtime would make a valuable advance to the study of the historyof contamination in industrial sites, which have until now reliedon the analysis of Hg concentrations in sediments.

Acknowledgements

We would like to acknowledge Lorenzo Alibardi for his insight-ful comments on the molecular composition of turtle carapace andcaiman skin and its potential relation to Hg binding. We thank theCNPQ (Conselho Nacional de Desenvolvimento Científico e Tec-nológico) who funded the BAJAQUEL Project (processo 408760/2006-0 Granted to Ronis Da Silveira). Laboratory analyses werefunded by the Ecochemistry Laboratory at the University of Can-berra and by the Research School of Earth Sciences at the Austra-lian National University, and were performed with the assistanceof Les Kinsley. Chelonian and caiman samples were collected inconjunction with INPA and UFAM (Universidade Federal do Ama-zonas, Brazil) under the auspices of IBAMA collection permit13347-1, IBAMA/CITES export permit 11BR006659/DF, CITESimport permit 2011-AU-635980 and AQUAS quarantine permitNA11102340. All procedures were conducted in accordance withthe Brazilian government permit requirements for the BAJAQUELproject. Euthanasia and sample extraction from live animals wasassisted by the veterinarian Augusto Kluczkovski Jr.

Appendix A. Supplementary material

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.chemosphere.2014.05.065.

References

Alibardi, L., 2003. Immunocytochemistry and keratinization in the epidermis ofcrocodilians. Zool. Stud. Taipei 42, 346–356.

Alibardi, L., 2004. Dermo-epidermal interactions in reptilian scales: Speculations onthe evolution of scales, feathers, and hairs. J. Exp. Zool. 302, 365–383.

Appelquist, H., Asbirk, S., Drabæk, I., 1984. Mercury monitoring: mercury stability inbird feathers. Mar. Pollut. Bull. 15, 22–24.

Baldwin, S., Deaker, M., Maher, W., 1994. Low-volume microwave digestion ofmarine biological tissues for the measurement of trace elements. Analyst 119,1701–1704.

Berry, K.H., 2002. Using growth ring counts to age wild juvenile desert tortoises(Gopherus agassizii) in the wild. Chelonian Conserv. Biol. 4, 416–424.

Bezerra, M.F., Lacerda, L.D., Costa, B.G., Lima, E.H., 2012. Mercury in the sea turtleChelonia mydas (Linnaeus, 1958) from Ceará coast, NE Brazil. An. da Acad. Bras.,de Ciências. 84, 123–128.

170 L. Schneider et al. / Chemosphere 119 (2015) 163–170

Bezerra, M.F., Lacerda, L.D., Lima, E.H.S.M., Melo, M.T.D., 2013. Monitoring mercuryin green sea turtles using keratinized carapace fragments (scutes). Mar. Pollut.Bull. 77, 424–427.

Burger, J., Gochfeld, M., Rooney, A.A., Orlando, E.F., Woodward, A.R., Guillette Jr., L.J.,2000. Metals and metalloids in tissues of American Alligators in Three FloridaLakes. Arch. Environ. Contam. Toxicol. 38, 501–508.

Congdon, J.D., Dunham, A.E., Sels, R.C.V.L., 1994. Demographics of commonsnapping turtles (Chelydra serpentina): implications for conservation andmanagement of long-lived organisms. Am. Zool. 34 (3), 397–408.

Da Silveira, R., Magnusson, W.E., Campos, Z., 1997. Monitoring the Distribution,Abundance and Breeding Areas of Caiman crocodilus crocodilus andMelanosuchus niger in the Anavilhanas Archipelago, Central Amazonia. Brazil.J. Herpetol. 31, 514. http://dx.doi.org/10.2307/1565603.

Da Silveira, R., Magnusson, W.E., 1999. Diets of Spectacled and Black Caiman in theAnavilhanas Archipelago, Central Amazonia, Brazil. J. Herpetol. 33, 181–192.

Da Silveira, R., Magnusson, W.E., Thorbjarnarson, J., 2008. Factors affecting thenumber of caimans seen during spotlight surveys in the mamirauá reserve,Brazilian Amazonia. Copeia 2008, 425–430.

Da Silveira, R., Campos, Z., Thorbjarnarson, J., Magnusson, W.E., 2013. Growth ratesof black caiman (Melanosuchus niger) and spectacled caiman (Caiman crocodilus)from two different Amazonian flooded habitats. Amphibia-Reptilia 34, 437–449.

Day, R.D., Christopher, S.J., Becker, P.R., Whitaker, D.W., 2005. Monitoring mercuryin the loggerhead sea turtle, Caretta caretta. Environ. Sci. Technol. 39, 437–446.

De Deus, C., Da Silveira, R., Py-Daniel, L.H.R., 2003. Piagacu-Purus: Bases Científicaspara Criação de uma Reserva de Desenvolvimento Sustentável. Instituto deDesenvolvimento Sustentável Mamirauá, Manaus, AM. Brasil.

R Development Core Team. R Development Core Team, 2013. R: A language andenvironment for statistical computing. R Foundation for Statistical Computing,Vienna, Austria. ISBN 3-900051-07-0, URL http://www.R-project.org. RFoundation for Statistical Computing, Vienna, Austria, 2013.

Eggins, S., 2012. Mercury bioaccumulation in chelonians and caimans of theAmazon Basin. PhD Thesis, Australian National University, Canberra, ACT,Australia.

Eggins, S.M., Kinsley, L.P.J., Shelley, J.M.G., 1998. Deposition and elementfractionation processes during atmospheric pressure laser sampling foranalysis by ICP–MS. Appl. Surf. Sci. 1998 (127), 278–286.

Fadini, P.S., Jardim, W.F., 2001. Is the Negro River Basin (Amazon) impacted bynaturally occurring mercury? Sci. Total Environ. 275 (1–3), 71–82.

Golet, W.J., Haines, T.A., 2001. Snapping turtles (Chelydra serpentina) as monitors formercury contamination of aquatic environments. Environ. Monit. Assessement71, 211–220.

Green, A.D., Buhlmann, K.A., Hagen, C., Romanek, C., Gibbons, J.W., 2010. Mercurycontamination in turtles and implications for human health. J. Environ. Health72, 14–22.

Grillitsch, B., Schiesari, L., 2010. The Ecotoxicology of Metals in Reptiles. In: Sparling,D., Linder, G., Bishop, C., Krest, S., (Eds.), Ecotoxicology of Amphibians andReptiles, second ed., CRC Press, pp. 337–448.

Heaton-Jones, T.G., Homer, B.L., Heaton-Jones, D.L., Sundlof, S.F., 1997. MercuryDistribution in American Alligators (Alligator mississippiensis) in Florida. J. Zool.Wildlife Med. 28, 62–70.

Hollins, S.E., Harrison, J.J., Jones, B.G., Zawadzki, A., Heijnis, H., Hankin, S., 2011.Reconstructing recent sedimentation in two urbanised coastal lagoons (NSW,Australia) using radioisotopes and geochemistry. J. Paleolimnol. 46, 579–596.

IUCN, 2013. The IUCN Red List of Threatened Species [Internet]. http://www.iucnredlist.org.

Legler, J., Vogt, R.C., 2013. The Turtles of Mexico. University of California Press,Berkeley, Ca.

Maher, W., Forster, S., Krikowa, F., Snitch, P., Chapple, G., Craig, P., 2001.Measurement of trace elements and phosphorus in marine animal and planttissues by low-volume microwave digestion and ICP–MS. At. Spectrosc. 22,361–370.

Martinelli, L.A., Ferreira, J.R., Forsberg, B.R., Victoria, R.L., 1988. Mercurycontamination in the Amazon: a gold rush consequence. Ambio 17, 252–254.

Pfeiffer, W.C., Lacerda, L.D., Salomons, W., Malm, O., 1993. Environmental fate ofmercury from gold mining in the Brazilian Amazon. Environ. Rev. 1, 26–37.

Rainwater, T.R., Wu, T.H., Finger, A.G., Cañas, J.E., Yu, L., Reynolds, K.D., et al., 2007.Metals and organochlorine pesticides in caudal scutes of crocodiles from Belizeand Costa Rica. Sci. Total Environ. 373, 146–156.

Sakai, H., Saeki, K., Ichihashi, H., Suganuma, H., Tanabe, S., Tatsukawa, R., 2000.Species-specific distribution of heavy metals in tissues and organs ofloggerhead turtle (Caretta caretta) and Green Turtle (Chelonia mydas) fromJapanese Coastal Waters. Mar. Pollut. Bull. 40, 701–709.

Schneider, L., Belger, L., Burger, J., Vogt, R.C., 2009. Mercury bioacumulation in fourtissues of Podocnemis erythrocephala (Podocnemididae: Testudines) as afunction of water parameters. Sci. Total Environ. 407, 1048–1054.

Schneider, L., Belger, L., Burger, J., Vogt, R.C., Ferrara, C.R., 2010. Mercury levels inmuscle of six species of turtles eaten by people along the Rio Negro of theAmazon basin. Arch. Environ. Contam. Toxicol. 58, 444–450.

Schneider, L., Belger, L., Burger, J., Vogt, R.C., Jeitner, C., Peleja, J.R.P., 2011a.Assessment of non-invasive techniques for monitoring mercury concentrationsin species of Amazon turtles. Toxicol. Environ. Chem. 93, 238–250.

Schneider, L., Ferrara, C.R., Vogt, R.C., Burger, J., 2011b. History of turtle exploitationand management techniques to conserve turtles in the rio negro basin of theBrazilian Amazon. Chelonian Conserv. Biol. 10, 149–157.

Schneider, L., Peleja, R.P., Kluczkovski Jr., A., Freire, G.M., Marioni, B., Vogt, R.C., et al.,2012. Mercury concentration in the spectacled caiman and black caiman(Alligatoridae) of the Amazon: implications for human health. Arch. Environ.Contam. Toxicol. 63, 270–279.

Schneider, L., Maher, W., Green, A., Vogt, R.C., 2013. Mercury Contamination inReptiles: An Emerging Problem with Consequences for Wild Life and HumanHealth. In: Kim, K.H., Brown, R.J.C. (Eds.). Mercury: Sources, Applications andHealth Impacts. Nova Science Publishers, Hauppauge, NY, pp. 173–232.

Toni, M., Valle, L., Alibardi, L., 2007. Hard (beta-) keratins in the epidermis ofreptiles: composition, sequence, and molecular organization. J. Proteome Res. 6,3377–3392.

Turnquist, M.A., Driscoll, C.T., Schulz, K.L., Schlaepfer, M.A., 2011. Mercuryconcentrations in snapping turtles (Chelydra serpentina) correlate withenvironmental and landscape characteristics. Ecotoxicology 20, 1599–1608.

Vogt, R.C., 1981. Natural History of Amphibians and Reptiles in Wisconsin.Milwaukee, Wis.: Milwaukee Public Museum.

Vogt, R.C., 2008. Amazon Turtles. Grafica Biblos, Lima, Peru.Wilson, D.S., Tracy, C.R., Tracy, C.R., 2003. Estimating age of turtles from growth

rings: a critical evaluation of the technique. Herpetologica. 59-178-174.Yanochko, G.M., Jagoe, C.H., Brisbin Jr, I.L., 1997. Tissue mercury concentrations in

alligators (Alligator mississippiensis) from the Florida Everglades and theSavannah River site, South Carolina. Arch. Environ. Contam. Toxicol. 32, 323–328.

Ziegler, T., Olbort, S., 2007. Genital structures and sex identification in crocodiles.Crocodile Spec. Group Newsl. 26, 16–17.