S uc u al equa ion modelling o me cu y in a-skele al a iabili y on
a chaeological human emains
Noemi Ál a ez-Fe nández
a,b,
⁎, An onio Ma ínez Co izas
a,c
, Olalla López-Cos as
d,e,
a
CRETUS, EcoPas (GI-1553), Facul ade de Bioloxía, Uni e sidade de San iago de Compos ela, 16782, Spain
b
Boscalia Technologies S.L., Spain
c
Bolin Cen e o Clima e Resea ch, S ockholm Uni e si y, S ockholm SE-10691, Sweden
d
EcoPas (GI-1553), CRETUS, A ea o A chaeology, Depa men o His o y, Uni e sidade de San iago de Compos ela, 15782, Spain
e
A chaeological Resea ch Labo a o y, S ockholm Uni e si y, Wallenbe glabo a o ie , SE-10691, Sweden
Labo a o io de An opología Física, Facul ad de Medicina, Uni e sidad de G anada, 18012, Spain
HIGHLIGHTS GRAPHICAL ABSTRACT
•Skele ons ha e a dual ole (sink and
sou ce) on Hg dynamics in g a es.
•Skele al Hg a iabili y seems o be a -
ec ed by an e-mo em s a us o bone s uc-
u e.
•Bone c ys allini y con ols he ole o bone
componen s on Hg e en ion.
•Soil plays a mino oleon Hg con en in a -
chaeological skele ons.
ABSTRACTARTICLE INFO
Edi o : Mae Sexaue Gus in
Keywo ds:
Hg
PLS-SEM
Skele on
Soil/sedimen s
Os eoa chaeology
Diagenesis
A chaeological bu ial en i onmen s a e use ul a chi es o in es iga e he long- e m ends and he beha iou o me -
cu y. In o de o unde s and he ela ionship be ween me cu y, skele ons and soil, we applied Pa ial Leas Squa es -
S uc u al Equa ion Modelling (PLS-SEM) o a de ailed, mul isampling (n = 73 bone samples +37 soil samples) design
o wo a chaeological g a es da ing o he 6 h o 7 h cen u ies CE (A Lanzada si e, NW Spain). Me cu y con en was
assessed using a DMA-80, and da a abou bone s uc u e and he g a e soil/sedimen s we e ob ained using FTIR-
ATR spec oscopy. The heo e ical model is suppo ed by p oxies o bone s uc u e, g a e soil/sedimen s, and loca ion
o he bone wi hin he skele on. The gene al model explained 61 % o me cu y a iance. Addi ionally, Pa ial Leas
Squa e –P edic ion O ien ed Segmen a ion (PLS-POS) was also used o check o segmen a ion in he da ase . POS e-
ealed wo g oup o samples depending on he bone phase (hyd oxyapa i e o collagen) con olling he Hg con en ,
and he co esponding models explained 86 % and 76 % o Hg a iance, espec i ely. The esul s sugges ha me cu y
beha iou in he g a es is complex, and ha me cu y concen a ions we e influenced by i) he an e-mo em s a us o he
bone ma ix, ela ed o he weigh o each bone phase; ii) pos -mo em e olu ion o bone c ys allini y, whe e bone loses
me cu y wi h inc easing al e a ion; and iii) he p oximi y o he skele al pieces o me cu y a ge o gans, as
Science o he To al En i onmen 851 (2022) 158015
⁎Co esponding au ho a : CRETUS, EcoPas (GI-1553), Uni e sidade de San iago de Compos ela, 16782, Spain.
E-mail add ess: n.al a ez. e [email protected] (N. Ál a ez-Fe nández).
h p://dx.doi.o g/10.1016/j.sci o en .2022.158015
Recei ed 31 May 2022; Recei ed in e ised o m 2 Augus 2022; Accep ed 9 Augus 2022
A ailable online 13 Augus 2022
0048-9697/©2022 TheAu ho s. Published by Else ie B.V. Thisis an openaccess a icleunde heCC BY-NC-ND license (h p://c ea i ecommons.o g/licenses/by-nc-nd/4.
0/).
Con en s lis s a ailable a ScienceDi ec
Science o he To al En i onmen
jou nal homepage: www.else ie .com/loca e/sci o en
decomposi ion and collapse o he ho acic and abdominal so issues causes a seconda y me cu y en ichmen in
bones om he body unk du ing ea ly pos -mo em. Skele ons p o ide a sou ce o me cu y o he soil whe eas soil/
sedimen s con ibu e li le o skele al me cu y con en .
1. In oduc ion
The cycle and oxicology o me cu y a e widely esea ched due o i s
oxici y. Me cu y is a global public heal h conce n acco ding o he Wo ld
Heal h O ganiza ion (WHO, 2020) and, in gene al e ms, i has been
ega ded as he mos oxic non- adioac i e elemen (Pushie e al., 2014).
I is ha m ul e en a e y low doses, and has no known biological unc ion.
Fu he mo e, me cu y is a wo ldwide dis ibu ed pollu an . The e a e bo h
an h opogenic and non-an h opogenic sou ces o he en i onmen , and
once i is bioa ailable me cu y bioaccumula es and biomagnifies in ood
chains and is e y pe sis en in ecosys ems (E e s, 2018;Mo el e al.,
1998;Tang e al., 2020).
Me cu y has di e en chemical ac ions ha occu na u ally in he
en i onmen ; including elemen al me cu y (Hg
0
), ino ganic me cu ous
(Hg
+
) and me cu ic (Hg
2+
) sal s, as well as o ganic compounds such as
me hyl- and e hyl‑me cu y (Be lin e al., 2015). The p ima y oxic e ec s
o his me al a e caused by i s capaci y o bind o sul hyd yl g oups, and,
o a lesse ex en , o hyd oxyl, ca boxyl, and phospho ylg oups. This abili y
o bind such compounds a ec s non-p o eins con aining hiol g oups,
educed glu a hione and p o eins con aining cys eine, in addi ion o i s
in e ac ions wi h ionic channels, anspo e s, and enzymes (Sil a-Filho
e al., 2021;Tchounwou e al., 2003). Me cu y's di e en chemical o ms
do no necessa ily sha e abso p ion pa hs o beha iou once hey a e inside
he o ganism, which makes i oxicologically complex (Be lin e al., 2015;
Cla kson, 1997). Fu he mo e, oxici y is also dependen on he dose,
ime, and way o exposu e (Abass e al., 2018). This a iabili y leads o a
wide ange o a ge o gans and clinical symp oma ology. A low-dose
and ch onic exposu e (i.e. en i onmen al) kidneys and li e a e he main
a ge o gans (Ga cía e al., 2001).
As a esul o e o s o elimina e he di ec sou ces o me cu y pollu ion,
which in some cases ha e led o acu e poisoning cases, mos s udies on
me cu y oxicology ocused upon he e ec s o high-dose exposu es (see
Ál a ez-Fe nández e al., 2020); despi e cases a e p edominan ly o low-
dose (Holmes e al., 2009). Ch onic exposu e a low doses is p ima ily
ela ed o an h opogenic emissions and he esul ing en i onmen al pollu ion.
In hese ci cums ances (low dose and ch onic), me cu y has impo an conse-
quences o heal h, such as ch onic in oxica ion and inc eased isk o de elop-
men al diso de s in child en (Budnik and Cas eleyn, 2019;Ha e al., 2017).
Unde s anding he me cu y cycle is a subjec o in ensi e esea ch
ecen ly (Gus in e al., 2020 and pape s con ained he ein). This cycle is
al e ed by he long esidence ime o me cu y in na u al sys ems
(Ou idge e al., 2018;S ee s e al., 2011). Cu en ly, 60 % o he annual
emissions o he a mosphe e come om me cu y p e iously deposi ed on
soil and wa e (Li e al., 2022). Me cu y in soils is mainly bond o o ganic
ma e , a empo a y sink un il i is e-mobilised (Gab iel and Williamson,
2004;Gębka e al., 2020;Skyllbe g e al., 2003). Ceme e ies a e a sou ces
o soil me cu y, especially when he inhuma ion o he body was pa o
he une a y i ual (Amuno, 2013;Jonke and Oli ie , 2012;Mohammed
and Abudei , 2020;P es es da Sil a e al., 2020;Spongbe g and Becks,
2000;Uslu e al., 2009;WHO, 1998). Human bodies ac as empo a y
sinks du ing li e as hey inco po a e me cu y om he en i onmen by
inhala ion, inges ion, e c. (Liu e al., 2011). Once bodies a e bu ied, hey
ac as a sou ce o me cu y –among o he elemen s – o he soil, and lea e
afinge p in on i (Ga cía-López e al., 2022;G a , 1986;Sobocká, 2004).
When a la ge numbe o bodies a e bu ied in he same place (i.e. long-
used ceme e ies), he ans e o me cu y is significan and could ha e an
ad e se impac on su ounding ecosys ems.
Na u al a chi es show ha humans ha e been impac ing he me cu y
cycle since a leas 3250 BCE (e.g., in he Sou h Ibe ian Peninsula)
(Leblanc e al., 2000). In NW Spain a mosphe ic me cu y pollu ion has
been da ed back o 500 BCE (Ma ınez-Co izas e al., 1999), and can be
ela ed o me cu y mining in he Ibe ian Peninsula du ing he I on Age.
An h opogenic emissions ma kedly inc eased o peak du ing he Roman
Empi e due o he ex ensi e mining and me allu gy o ulfil economic
demands. Pollu ion dec eased wi h he all o he Wes e n Roman Empi e.
Ál a ez-Fe nández e al. (2020) and López-Cos as e al. (2020) assessed
he impac o hese fluc ua ions on me cu y le els in pas -popula ions
h ough hei os eoa chaeological emains, showing simila ends o
hose p o ided by local econs uc ions o a mosphe ic pollu ion om
pea land a chi es. The e o e, me cu y con amina ion in skele ons can
complemen he da a ob ained om commonly used na u al a chi es and
p o ide mo e specific in o ma ion abou he impac o ce ain pollu an s,
like me cu y, on bio a (Cooke e al., 2020;Gus in e al., 2020). Howe e ,
he p ese ed a chaeological emains a e mainly bones and ee h (i.e.
skele ons), which a e no he main a ge o gans o me cu y and, in
some cases, ha e been bu ied o a long ime ( o a e iew o Hg s udies
on a chaeological popula ions see Ál a ez-Fe nández e al. (2020)).
Diagenesis is a majo issue when ace elemen al composi ion is s udied
in os eoa chaeological eco ds (Hedges, 2002). Fo he g a es s udied in
his wo k, we add essed he ac o s con olling me cu y con en in soil/
sedimen s in a p e ious in es iga ion (Ál a ez-Fe nández e al., 2021).
This s udy showed ha he bu ied indi iduals we e he only sou ce o
me cu y, and esul s ag eed wi h p e ious wo k ha ound no e idence
o diagene ic inco po a ion o me cu y om soil o bone (Emslie e al.,
2015;Kepa e al., 2012;Rasmussen e al., 2008, 2013b;Walse e al.,
2019;Yamada e al., 1995). Howe e , he po en ial impac ha pas
human demog aphy and une a y i uals may ha e had on he soil cu en
s a e emains o be in es iga ed. To do so, i is impo an o add ess he
ela ionship be ween me cu y and bone issue, bo h an e- and pos -mo em,
o be able o unde s and he ole o skele ons on me cu y dynamics in he
su ounding soil.
In his in es iga ion, wo pos -Roman-Ea ly-Medie al bu ials om he
a chaeological si e o A Lanzada (NW Spain) we e selec ed and a de ailed
soil and bone sampling scheme wasused o assess he a iabili y o me cu y
concen a ions in he skele ons and he soil. Bone samples o he bu ied
indi iduals we e aken as close as possible o he loca ion o he soil
samples. The weigh o po en ial ac o s de e mining he in a-skele al
a iabili y in me cu y concen a ions was assessed using Pa ial Leas
Squa es - S uc u al Equa ion Modelling (PLS-SEM), wi h he objec i e o
elucida ing: i) he ole o bone s uc u e (an e- and pos -mo em)inme cu y
con en ; ii) he ole o bone componen s on me cu y e en ion; iii) whe he
he bones we e a sou ce and/o sink o me cu y in he bu ial en i onmen ;
and i ) he ole o skele ons on he me cu y cycle.
2. Ma e ial and me hods
2.1. Bone and soil samples
Se en y- h ee bulk bone and 37 soil/sedimen (sil +clay ac ion)
samples associa ed wi h h ee skele ons ound in wo pos -Roman (AD
5 h cen u y) bu ials (T1 and T5) om A Lanzada si e (Sanxenxo
Po en ed a, NW Spain, 42° 25′46″N8°52′25″W) we e analysed
(Fig. 1). A Lanzada is an a chaeological si e wi h se e al occupa ional
phases. The mos ele an ones o his s udy a e a long-used habi a ional
a ea (B onze o pos -Roman imes), o he Wes , and a nec opolis wi h
wo une a y a eas (Roman and pos -Roman; exca a ed du ing he 1960s
and 1970s) o he Eas (López-Cos as, 2015;Rod íguez Ma ínez, 2017).
The g a es, T1 and T5, we e included in his s udy, we e disco e ed on
he Wes side o he si e du ing he las exca a ion campaign
(2016–2017), when a de ailed an h opological and pedological s udy was
N. Ál a ez-Fe nández e al. Science o he To al En i onmen 851 (2022) 158015
2
done (including he collec ion o soil samples om he bu ials). They we e
loca ed closed o each o he and, acco ding o he a chaeological ma e ial
(~6 h o 7 h cen u y CE), da ed o he same a chaeological laye o a mon-
umen al s uc u e ha was hough o be a chu ch (Rod íguez Ma ínez,
2017). The bu ials we e o ien ed Wes -Eas and exca a ed om dune
sands. The g ain size o he bu ials soil/sedimen s is domina ed by sand
(85 % in he bu ials' soil and 95 % in he soil ou side he bu ials; Ga cía-
López e al., 2022). The bu ial soils showed an en ichmen in P and a
dec ease in alkalini y ( om pH ~9 o 8), compa ed o he soil loca ed
om ou side he bu ials. They also showed a seconda y en ichmen in
sil +clay ac ions and o ganic ma e (Ál a ez-Fe nández e al., 2021),
soil componen s able o e ain me cu y. T1 and T5 a e no di ec ly ela ed
o he une a y a eas o he nec opolis p e iously ound, so he la ge num-
be o bodies he e was no expec ed o g ea ly influence he geochemis y
o he wo bu ials. Radioca bon da ing indica ed ha he indi iduals died
some ime be ween he 5 h o he 6 h cen u ies cal. AD. Bodies we e p oba-
bly deposi ed in wood boxes (i.e., co fins) since a clea colou pa e n o
ec angula shape was e iden in bo h bu ials (Fig. 1) and nails we e
ound a he edges. All he skele ons we e in supine posi ion wi h s e ched
a ms and legs (hands placed a bo h sides o he body o o e he pel ic
a ea; see Fig. 1). T1 is a single bu ial con aining he skele on o an elde ly
(>60 yea s) emale (L01) o sho s a u e (152 cm) when compa ed o
Ibe ian a chaeological popula ions om Roman and Medie al imes
(López Cos as, 2012;López Cos as e al., 2017). T5 is a mul iple coe aneous
bu ial con aining wo male skele ons, an adolescen (15–19 yea s) placedin
Wes -Eas o ien a ion (L06) and a ma u e adul (40–50 yea s) placed in
Eas -Wes o ien a ion (L07). Bo h had abo e-a e age s a u es (175 cm
and 170 cm, espec i ely) when compa ed o a chaeological popula ions
om he same pe iod (López Cos as, 2012;López Cos as e al., 2017). Sex
was es ima ed ollowing es ablished c i e ia on he innomina e and c anial
bones (see a summa y in Buiks a and Ubelake (1994)). Age was es ima ed
using s anda d iden ifica ion c i e ia o innomina e bone (au icula
su ace and pubic symphysis mo phology), ou h ib and epiphyseal usion
( o a summa y see Buiks a and Ubelake (1994)), and Ibe ian me hods o
g ow h and ma u i y o pos c anial bones (López-Cos as e al., 2012;
Rissech e al., 2013). S a u es we e calcula ed using he hume us and
emu maximum leng h (De Mendonça, 2000). All bone emains we e
well-p ese ed (Table SM 1), p esen ing he ollowing ab asion deg ees
(a e age o he whole p ese ed bony pa s, acco ding o B ickley and
McKinley, 2004): L01 deg ee 3, L06 deg ee 2, L07 deg ee 2. L06 and L07
ha e in ense physical (p essu e) and chemical al e a ion in localised a eas
(L06: ace and ee , L07: long bones epiphyses), ha esul ed in he loss o
some o he bones/bone a eas despi e hei low deg ee o supe ficial
ab asion and being well p ese ed (López Cos as e al., 2017).
Bulk bone (co ical) was mic o-sampled using a den is 's d ill, a e he
supe ficial laye was emo ed o a oid con amina ion. Due o he impo -
ance o p ese ing a chaeological skele ons, a small amoun was sampled
(~30 o 60 mg). Fou ypes o bone we e analysed when possible: i) ype
0a: e eb ae (spine) and ilium; ii) ype 1a: ibs; iii) ype 2a: long bones;
and i ) ype 3a: c ania (Table SM 1). This classifica ion is modified om
p e ious s udies in es iga ing in a-skele al a iabili y o me cu y and
o he elemen s om he la ge nec opolis a ea om A Lanzada si e
(Ál a ez-Fe nández e al., 2020;López-Cos as e al., 2016). The sampling
s a egy ollowed a mul isampling app oach o co e mos o he skele on
in a- a iabili y (Fig. 2 and Table SM 1). Fo e hical easons he numbe
o samples in L01 was educed due o he low bone densi y, ela ed o
se e e os eopenia (possibly a case o senile os eopo osis). A o al o
73 bulk powde bone samples we e analysed: 11 o L01, 31 o L06 and
31 o L07. Since he p esen esea ch is he con inua ion o p e ious
wo k (Ál a ez-Fe nández e al., 2021), in which he dis ibu ion o me cu y
in he bu ial soil/sedimen was assessed o hese wo g a es, bulk bone
sampling was in ended o be in ha mony wi h he soil/sedimen sampling
design, whe e samples we e collec ed in wo ansec s –longi udinal and
ans e se –along each skele on. The soil/sedimen samples used he e
and in he p e ious esea ch a e he same, excluding hose om ou side
bu ials ha we e no included in his s udy. As soil/sedimen samples
we e collec ed in he field, an exac ma ch was no possible o all he
bone samples, so he closes soil/sedimen sample o each bone sample
50 m
5 cm
5 cm
L001
L006
L007
T1
T5
A
B
C
Fig. 1. (A) and (B) Ae ial View o A Lanzada wi h app oxima ed loca ion o he g a es; (A) modified om Google Ea h 2020., (B) modified om Rod íguez Ma ínez, 2017.
(C) Tombs.
N. Ál a ez-Fe nández e al. Science o he To al En i onmen 851 (2022) 158015
3
was used. Median dis ance be ween bone and soil samples was <10 cm in
all cases. Un o una ely, he o he skele ons belonging o he la ge nec op-
olis a ea we e exca a ed >50 yea s ago wi hou collec ing soil samples.
2.2. Me cu y analyses
Bulk bone me cu y concen a ions we e de e mined using a DMA-80
(Miles on) hos ed a he labo a o y o he Eco oxicoloxía e Ecofisioloxía
Vexe al esea ch g oup (Depa amen o de Bioloxía Funcional, USC) ollowing
he p o ocol desc ibed in Ál a ez-Fe nández e al. (2020). Two s anda d
e e ence ma e ials we e used: es ua ine sedimen GBW07601a (670 ±
60 ng g
−1
) and bone meal SRM 1486 (2.3 ± 1.4 ng g
−1
). The quan ifica-
ion limi was 0.26 ng g
−1
. Mean eco e y o e e ence ma e ials was
107 % GBW07601a (698 ± 50 ng g
−1
) and 104 % SRM 1486 (2.5 ±
1.2 ng g
−1
). Quali y con ols ( eplica ed analyses e e y 10 samples) we e
done o 29 samples, epo ing an a e age di e ence o 1.9 ng g
−1
.
2.3. FTIR-ATR analyses
Bone and soil samples we e analysed by o al a enua ed eflec ance
Fou ie - ans o m in a ed spec oscopy (FTIR-ATR) using a spec ome e
Agilen Ca y 630 FTIR coupled wi h an ATR module, loca ed a EcoPas
labo a o ies (Facul ade de Bioloxía, USC). Spec a we e acqui ed in he
mid-in a ed egion (MIR) 4000–400 cm
−1
, by a e aging 100 scans a a
esolu ion o 4 cm
−1
. The equipmen was cleaned, and a backg ound was
collec ed be o e e e y measu emen . Peak iden ifica ion was done using
he {andu inha} R package (Ál a ez Fe nández and Ma ínez Co izas,
2020) ollowing he second de i a i e sum spec um me hod a e Z sco e
s anda disa ion. Assignmen o compounds ela ed o ib a ions is based
on hose epo ed in he li e a u e (see e e ences in Table SM 2), aking
in o accoun he limi a ions imposed on IR in e p e a ion o complex
samples (Coa es, 2000;La kin, 2017;Simonescu, 2012;Soc a es, 2004).
Se e al indices ela ed o he bone s uc u e we e calcula ed: i) he
in a ed spli ing ac o (IRSF), which is ela ed o hyd oxyapa i e c ys al-
lini y (Weine and Ba -Yose , 1990), as he sum o he ν
4
(PO
4
) peaks in en-
si ies a 600 and 559 cm
−1
di ided by he in ensi y o he alley be ween
hem (587 cm
−1
); ii) he mine al ma u i y index (MMI), which ep esen s
he p og essi e ans o ma ion o poo ly c ys allised non-apa i e domains
in o well c ys allised apa i e (Fa lay e al., 2010), as he a io 1019/
1111 cm
−1
o ν
3
(PO4); iii) he CO
3
/PO
4
(CP) a io, which indica es he
ca bona e con en o hyd oxyapa i e, di iding he ν
3
(CO
3
) peak in ensi y
a 1409 cm
−1
by he ν
3
(PO4) peak in ensi y a 1019 cm
−1
(G unenwald
e al., 2014); and i ) he collagen con en es ima ed h ough he Amide I/
PO
4
a io (AmP), di iding he peak in ensi y o ν1(Amide I)banda
1638 cm
−1
by he ν
3
(PO
4
) band a 1019 cm
−1
(T ueman e al., 2004a).
2.4. S a is ics
2.4.1. Desc ip i e me hods
S a is ical analyses and g aphs we e done using he so wa e R (R Co e
Team, 2021). No mali y was assessed using he Shapi o-Wilk es , as i is
he mos obus e en when he sample size is ela i ely small (Yap and
Sim, 2011). Gi en he non-pa ame ic na u e o he da a, desc ip i e
analysis was based on median and in e qua ile anks (IQR). To de ec
significan di e ences among g oups, Mann-Whi ney-Wilcoxon was
applied when compa ing wo g oups, o K uskal-Wallis combined wi h
pai wise Mann-Whi ney-Wilcoxon when mo e han wo g oups we e
in ol ed. The p- alues we e conside ed significan when p <0.05.
2.4.2. Pa ial leas squa es s uc u al equa ion modelling
To assess me cu y in a-skele al dis ibu ion, Pa ial Leas Squa es
S uc u al Equa ion Modelling (PLS-SEM) was applied, using he Sma PLS
so wa e (Ringle e al., 2015). This me hod enables he es ima ion o
complex models ollowing a causal-p edic i e app oach wi hou imposing
dis ibu ional assump ions on he da a ma ix (Hai e al., 2019a;Sa s ed
e al., 2017) while aiming o maximise he explained a iance o he depen-
den la en cons uc s (Hai e al., 2019b). The inne and ou e models a e
ep esen ed in Fig. 3. The la en cons uc s can be classified in o h ee
g oups: i) hose ela ed o bone componen s, collagen and hyd oxyapa i e
(HAP), o assess hei ole on he skele on me cu y con en , and he deg ee
o c ys allini y o he mine al phase (HAPc), as i can a ec he elease o
me cu y om he bone issue o he su ounding soil; ii) soil componen s
Fig. 2. Sampling design. Sampled bones colou ed in ligh pu ple wi h he sampling a ea highligh ed wi h do s in da k pu ple. See u he in o ma ion in Table SM 1.
N. Ál a ez-Fe nández e al. Science o he To al En i onmen 851 (2022) 158015
4
in he sil +clay ac ion, among which me cu y (Hg), p ima y silica es (p-
silica es) and clay we e selec ed o assess he deg ee o in e ac ion be ween
bone and soil, as he bone was ound o be a po en ial sou ce o me cu y o
he soil; and iii) loca ion, as i is suscep ible o a ec me cu y con en
(Ál a ez-Fe nández e al., 2021).
Measu ed a iables included: i) o collagen he FTIR-ATR bands a
3280 (Amide A and –OH ib a ion), 1638 (Amide I) and 1545 cm
−1
(Amide II) (Goo magh igh e al., 2006;Ma ínez Co izas and López-
Cos as, 2020;T ueman e al., 2004b); ii) o HAP bands a 870 (ν
2
(CO
3
)),
712 (ν
4
(CO
3
)), 690 (ν
4
(PO
4
) and –OH), and 470 cm
−1
(ν
2
(PO
4
)) (Bax e
e al., 1966;Flee , 2009;Rey e al., 2011;Zammel e al., 2021); iii) o
HAPc he band a 1019 cm
−1
cha ac e is ic o ν
3
(PO4) (Dal Sasso e al.,
2016;Rey e al., 1991); i ) o soil Hg, concen a ions ob ained in a p e i-
ous wo k on he same g a es we e used (Ál a ez-Fe nández e al., 2021);
) o p-silica es he bands a 606 (Si, Al e ahed al de o ma ion), 585
(Si e ahed al b ea hing) and 531 cm
−1
(Si –O–H ib a ions)
(McKeown, 2005;Pé ez-Rod íguez e al., 2016); i) o clay, bands a
3694 and 3621 cm
−1
cha ac e is ic o kaolini e –OH s e ching
(Vaculíko á e al., 2011); and ii) o loca ion, he longi udinal axis (x-
axis) o each skele on se ing as ze o he pel ic a ea (app oxima ely a S3
sac al e eb a). All measu ed a iables we e se on eflec i e mode.
Gi en he composi ional na u e o he mani es a iables o la en con-
s uc s bone Hg and soilHg, concen a ions we e p e iously ans o med by
na u al loga i hm. The da a ma ix (Table SM 3) was s anda dised and he
pa h me hod was chosen o he weigh ing scheme (Hai e al., 2017). Two-
ailed boo s apping was applied o es he s a is ical significance o pa h
coe ficien s choosing he bias-co ec ed and accele a ed (BCa) me hod o
he confidence in e als (Hai e al., 2017). The significance le el was se
o 0.05. Pa ial Leas Squa e P edic ion-O ien ed Segmen a ion (PLS-POS)
was also used o iden i y segmen s (i.e., g oups o samples) in he da a se
(Becke e al., 2013).
3. Resul s
3.1. Me cu y a iabili y
Bone me cu y concen a ions a ied om 2 o 4 ng g
−1
o L01, om 1
o 39 ng g
−1
o L06,and om1 o18ngg
−1
o L07 (Table 1). Di e ences
in me cu y be ween indi iduals we e significan (p- alue <0.01). L01
bones we e he leas en iched in me cu y ollowed by L07, while L06 had
he highe me cu y concen a ions. To e alua e me cu y a iabili y
among bone ypes he ou g oups we e conside ed. No significan di e -
ences we e ound be ween ype 0a ( e eb ae+ilium) and ype 1a ( ibs)
and be ween ype 2a (long bones) and ype 3a (c ania). The e o e, hese
g oups we e combined in o wo final g oups ( ype1 = 0a + 1a: spine,
ilium, ibs; and ype2 = 2a + 3a: long bones, c ania). F om he e on we
will e e exclusi ely o hese wo bone ypes. Me cu y concen a ions
Fig. 3. PLS-SEM inne model and ou e model. HAP: hyd oxyapa i e; HAPc: hyd oxyapa i e c ys allini y; p-silica es: p ima y silica es.
Table 1
Summa y o Hg concen a ions in bone pe indi idual and bone ype (ng g
−1
). L01
is a emale and L06 and L07 a e males.
Minimum Median Maximum IQR
Indi idual L01 1.97 2.94 4.31 0.94
L06 1.19 9.21 38.58 11.69
L07 1.48 5.72 17.86 3.32
Bone ype Type 1 2.15 8.92 38.58 8.35
Type 2 1.19 3.34 17.86 2.45
N. Ál a ez-Fe nández e al. Science o he To al En i onmen 851 (2022) 158015
5
we e ound o be s a is ically highe (p- alue <0.01) in bone ype 1
( a ying om 2 o 39 ng g
−1
) han in bone ype 2 ( a ying om 1 o
18 ng g
−1
)(Table 1).
3.2. FTIR-ATR spec a
Fig. SM 1 gi es an o e iew o he ull bone spec a in he ange
4000–400 cm
−1
. The a e age spec um shows he mos cha ac e is ic
peaks, all he spec a enable o de ec he egions wi h he highes a iabil-
i y among samples, while he second de i a i e spec um loca es peaks,
including hose wi h lowe abso bance. The clay egion, ~3700 o
~3610 cm
−1
, showed high a iabili y al hough a e age band in ensi y is
low. The egion om ~3600 o ~2800 cm
−1
, wi h bands ela ed o
collagen ib a ions (amide A and B, alipha ics and –OH g oups), showed
mode a e a iabili y and low abso bance. The Amide I collagen band, a
~1660 cm
−1
, showed mode a e abso bance bu high a iabili y. Abso p-
ions in he egion ~1450 o ~1300 cm
−1
, co esponding o he ν
3
(CO
3
)
ib a ion modes, showed mode a e in ensi y and a iabili y. F om ~1200
o ~950 cm
−1
, he egion wi h he cha ac e is ic ν
3
(PO
4
)andν
1
(PO
4
)
ib a ional modes o phospha es, showed high o mode a e in ensi ies
and low a iabili y. Bands be ween ~900 and ~700 cm
−1
,o ν
2
(CO
3
)
and ν
4
(CO
3
) ib a ional modes o ca bona es, showed mode a e in ensi ies
and a iabili y. The ypical double o he ν
4
(PO
4
) ib a ional mode o
phospha e, be ween ~690 and ~500 cm
−1
, showed high in ensi ies, bu
low a iabili y.
3.3. Bone s uc u e indices
Bone me cu y was sligh ly co ela ed wi h he in a ed spli ing ac o
(IRSF), mine al ma u i y index (MMI), ca bon/phospha e a io (CP), and
amide I/phospha e a io (AmP). The co ela ion wi h IRSF and MMI was
nega i e wi h alues a ound ~−0.36 (Table SM 4). Co ela ion wi h CP
and AmP was posi i e, wi h alues o 0.34 and 0.50 espec i ely. Va iabil-
i y was minimal when compa ing indi iduals IRSF (L01: 4.3 ± 0.3, L06:
3.9 ± 0.9, L07: 4.1 ± 0.5), MMI (L01: 9.0 ± 0.9, L06: 8.2 ± 1.0, L07:
9.1 ± 2.0) and CP (L01: 0.07 ± 0.02, L06: 0.09 ± 0.06, L07: 0.09 ±
0.04). While o AmP significan di e ences we e ound be ween T1
(L01: −0.03 ± 0.00) and T5 (L06: 0.09 ± 0.06, L07: 0.09 ± 0.04)
(Table 2). Simila esul s we e ob ained when compa ing bone ype, wi h
low a iabili y in IRSF ( ype 1: 4.0 ± 0.7, ype 2: 4.2 ± 0.6), MMI ( ype
1: 8.5 ± 1.2, ype 2: 9.3 ± 1.7), CP ( ype 1: 0.09 ± 0.04, ype 2: 0.08 ±
0.03) and significan di e ences in AmP ( a ying om −0.03 o 0.05 in
bone ype 1 and om −0.03 o −0.00 in bone ype 2). The bone ype 2
sample L06.30 was iden ified as an ou lie o AmP wi h a alue o 0.07.
3.4. PLS-SEM model
A PLS-SEM model (Fig. 3) wi h 8 la en cons uc s (i.e., inne model)
and 16 indica o a iables (i.e., ou e model) was es ed in he whole da a
se (G0). This gene al model (G0) accoun ed o 61 % o Hg a iance. Addi-
ionally, PLS-POS analysis e ealed he exis ence o wo g oups (Table SM
3). The same model (Fig. 3) used o G0 was applied o he wo g oups
(G1 and G2). The model o G1 accoun ed o 86 %, and he model o G2
accoun ed o 76 % (Table SM 7 and Fig. SM 2) o he me cu y a iance.
In e ac ions be ween LVs (i.e., pa hs)in G0accoun o 62 % o collagen
a iance, 59 % in G1 and 66 % in G2. The models also explain ~90 % o
hyd oxyapa i e (HAP) a iance (93 % G0, 95 % G1, 92 % G2). G0 also
accoun s o 38 % o soil Hg a iabili y, G1 accoun s o 17 % and G2 o
60 %. Table SM 8 shows he pa h coe ficien s o G0, G1 and G2 wi h he
significance le els. The pa h coe ficien s om HAP o bone Hg (HAP➔bone
Hg) a e significan in he h ee models and high and posi i e in G0, mode -
a e/high and nega i e in G1 and s ong and posi i e in G2. The pa h coe fi-
cien s hyd oxyapa i e c ys allini y (HAPc)➔bone collagen and HAP we e
significan and nega i e in all he models; while he pa h coe ficien s
HAPc➔bone Hg we e high and nega i e o G1 and posi i e o G2, bu
no significan o G0. Bone collagen➔bone Hg was he only significan
o G1, in which he pa h coe ficien is mode a e and posi i e. Loca ion➔
bone Hg pa h coe ficien s we e low and nega i e o he h ee models,
being significan only o G0. Soil clay had a low and nega i e e ec on
bone Hg in he h ee models. Soil p ima y silica es (p-silica es) had a nega-
i e coe ficien sin all he models, being mode a e and significan only in G0
and G2. Soil clay showed e y low and non-significan pa h coe ficien s on
soil Hg in all he models, while soil p ima y silica es had a nega i e pa h
coe ficien in all o hem, bu significan only o G0 and G2. Finally, bone
Hg has a low posi i e pa h coe ficien on soil Hg in all he models, which
we e significan o G0 and G2.
Fig. 4 shows he di ec and indi ec e ec s o all he model pa hs
(Table SM 9). Soil p-silica es had a nega i e e ec (di ec and indi ec )
bo h on bone Hg and soil Hg in all models including: i) Loca ion which
has a nega i e indi ec e ec on soil Hg and a nega i e di ec e ec on
bone Hg; ii) Bone collagen had a di ec posi i e e ec on bone Hg, bu i
was only significan o G1; iii) Hyd oxyapa i e c ys allini y (HAPc) had a
nega i e indi ec e ec on soil Hg and a nega i e indi ec e ec on HAP
and bone collagen; i ) HAPc had a posi i e e ec on bone Hg, being di ec
o G0 and G2 and indi ec o G1. HAPc also had a nega i e e ec on bone
Hg, indi ec in G0 and G1 and di ec in G2; ) HAP had an indi ec e ec on
soil Hg, being posi i e in G0 and G2 and nega i e, bu non-significan o
G1; i) HAP had a di ec e ec on bone Hg, posi i e in G0 and G1 and
nega i e in G1; ii) Bone Hg had a posi i e di ec e ec on soil Hg. The
G0 esiduals (Fig. SM 4) o L06 and L07 showed an unde es ima ion o
he samples in he ho acic a ea while o e es ima es he samples om
long bones samples; no clea pa e n was obse ed o L01 esiduals. In
G1 and G2 esiduals we e low and show no clea spa ial pa e n.
Samples assigned o G1 and G2 did no espond o significan di e -
ences in Hg con en , bone s uc u e indices, o o di e ences be ween
g a es and among indi iduals, bu a possible end in he classifica ion
based on a isu o he skele ons was iden ified. Samples in g oup-1 we e
less ossified ( hinne co ical) and had highe chemical al e a ion (su ace
ab asion) han hose in g oup-2 (Table SM 1). Fu he in o ma ion abou
he model can be ound in he Supplemen a y Ma e ial, including block
unidimensionali y, ou e model summa y, inne model summa y, inne
model pa hs summa y and o al/di ec /indi ec e ec s.
4. Discussion
4.1. Me cu y con en
Me cu y con en a ied among indi iduals and dec eased wi h age
(L01 <L07 <L06). Howe e , he small sample size (n = 3 indi iduals)
does no allow us o disca d a spu ious ela ionship, as many o he ac o s
(e.g., exposu e du ing li e) could also be esponsible o his pa e n. In a
p e ious s udy, he impac o me cu y a mosphe ic concen a ions in A
Lanzada Roman (n = 43) and pos -Roman (n = 33) popula ion was
assessed by analysing h ee ypes o bones ( ibs, long bones, and c ania).
No ela ionship was ound be ween me cu y con en and age in any o
he wo coho s (Ál a ez-Fe nández e al., 2020), in line wi h esul s om
ecen esea ch om biopsies (Babuśka-Roczniak e al., 2021;Zioła-
F ankowska e al., 2017) and au opsies (Domingo e al., 2017;Yoo e al.,
2002). Va ia ion in me cu y concen a ions among he h ee analysed
indi iduals (L01, L06, and L07) may be explained by indi idual di e ences
Table 2
Summa y o AmP concen a ion in bone pe indi idual and bone ype. L01 is a
emale and L06 and L07 a e males.
Minimum Median Maximum IQR
Indi idual L01 −0.028 −0.025 −0.021 0.003
L06 −0.033 −0.016 0.071 0.025
L07 −0.030 −0.021 0.051 0.010
Bone ype Type 1 −0.033 −0.015 0.052 0.022
Type 2 −0.031 −0.024 −0.002 0.004
N. Ál a ez-Fe nández e al. Science o he To al En i onmen 851 (2022) 158015
6
Fig. 4. PLS-SEM models: di ec and indi ec e ec s; significan pa hs a e highligh ed wi h (*).
N. Ál a ez-Fe nández e al. Science o he To al En i onmen 851 (2022) 158015
7
in an e-mo em exposu e and accumula ion. P e ious esea ch a A Lanzada
sugges ed ha me cu y exposu e was mainly ela ed o a mosphe ic pollu-
ion du ing bo h Roman and pos -Roman imes (Ál a ez-Fe nández e al.,
2020;López-Cos as e al., 2020). Exposu e o la ge doses o me cu y was
e y unlikely as he median concen a ion was ~6 ng g
−1
o in L01, L06,
andL07and21±23ngg
−1
o he pos -Roman coho om he nec opo-
lis. L06 and L07 we e likely o sha e a simila en i onmen al exposu e, as
hey we e bu ied in a double non-consecu i e bu ial ha indica es ha
bo h died a he same ime o wi hin a ew days o each o he . L01 could
ha e li ed a ew decades ea lie o la e , and his indi idual could ha e
had di e en en i onmen al exposu e. Di e ences in en i onmen al expo-
su e could explain in pa he sligh di e ences obse ed he e, as well as
hose ound in he indi iduals bu ied in he une a y a eas o he same
a chaeological si e (Ál a ez-Fe nández e al., 2020). In con as , me cu y
accumula ion appea s o be ela ed o indi idual ac o s such as daily li e
habi s, occupa ional ac i i ies, and body mass index (Bjø klund e al.,
2017;Zioła-F ankowska e al., 2017). So, di e ences in me cu y con en
among he s udied indi iduals may be explained by non-measu ed cha ac-
e is ics, such as body mass. I s a u e is used as a p oxy o body mass (no -
wi hs anding any limi a ions), he same end as ha obse ed in me cu y
con en is ound: L01 (152 cm) <L07 (170 cm) <L06 (175 cm). L06 and
L07, bo h males, had highe s a u e, alle han he Ibe ian a e age o
he pe iod, hus may ha e had highe body mass han L01, a senile
woman o sho e s a u e. L01 also su e ed om ad anced os eopenia,
p obably as a side e ec o he age (>60 yea s-old) ha could ha e a ec ed
me cu y accumula ion/ elease (Lanocha e al., 2013;Zioła-F ankowska
e al., 2017). Old age could also ha e p e en ed he om pa icipa ing in
ac i e occupa ional ac i i ies, and she may ha e had a mo e seden a y
li es yle compa ed o he wo men. The e is an ongoing deba e ega ding
he associa ion be ween me cu y and os eopenia wi h con as ing esul s.
Some s udies sugges me cu y a ec s bone me abolism (Suzuki e al., 2004;
Tang e al., 2022), while o he s did no find any significan associa ion
be ween me cu y exposu e and his pa hology (see a e iew in Jalili e al.,
2020). The e is only one s udy whe e me cu y was ound o dis up calcium
me abolism, in goldfish scales, and me cu y dose was high (10
−7
Mme hyl-
me cu y exposu e du ing 2, 4, and 8 days in Suzuki e al. (2004)). I is
unlikely ha me cu y exposu e was he cause o os eopenia in L01.
Bone ype showed significan di e ences in me cu y con en . Bone ype
1 (spine+ilium+ ibs) had highe me cu y concen a ion han bone ype 2
(long bones+c ania). Rasmussen e al. (2017) ound ha ibs had a highe
me cu y con en compa ed o long bones when indi iduals we e ea ed
wi h me cu y emedies, and hey p oposed wo causes: i) he bone u n-
o e , which is quicke in sho bones, and ii) he collapse o Hg-en iched
so issues con ained in he ho ax. Bo h causes a e possible and may co-
exis as an e- and pos -mo em mechanisms, and he e o e accoun o me -
cu y a iabili y be ween he wo analysed bone ypes. Type 1 bones also
ha e a p io i highe u no e a es han ype 2 bones, pa icula ly long
bones such as he emu (Hedges e al., 2007). The e o e, he highe Hg
con en in bone ype 1 may be ela ed o inc eased exposu e o me cu y
in he mos ecen yea s be o e dea h, and possibly he main cause o a i-
abili y. A la ge p opo ion o an adul emu is o med du ing g ow h spu s
(Hedges e al., 2007), and i was p oduced when he indi idual was possi-
bly less engaged in ac i i ies ela ed o me al exposu e (i.e., me allu gy).
Emslie e al. (2019, 2015) obse ed ha hume i had highe me cu y
con en han emo a and ibiae. They hypo hesised ha highe me cu y
con en may be due o bone emodelling a es, ela ing hem o biomechan-
ical egula ion as consequence o he use o hea y ools and o he daily li e
ac i i ies. Howe e , hey did no explain why skele al ma ke s we e mo e
exp essed in hume i han on lowe limbs. Since hey only analysed long
bones (hume i, emo a, and ibiae), he compa ison wi h small bones
om ho acic and abdominal a ea was no possible. An al e na i e hypo h-
esis can be conside ed: he seconda y me cu y en ichmen o bones
om he body unk occu ed du ing ea ly pos -mo em, due o he
decomposi ion and collapse o he ho acic and abdominal so issues
(no e ha du ing he b eakdown o so issues he body mass c ea es
anoxic condi ions (Janaway e al., 2009)). This finding is consis en wi h
Ál a ez-Fe nández e al. (2021), who ound ha me cu y con en in he
soil/sedimen s om hese wo bu ials s udied he e was ela ed o p oximi y
o he ho acic/abdominal a ea. The main a ge o gans o me cu y,
when exposu e is low and ch onic, a e placed in he abdomen and ho ax
(i.e.: kidneys and li e ) and ha lungs and diges i e appa a us can also
accumula e some me cu y (Lech and Sadlik, 2004). Rasmussen e al.
(2013a, 2013b) also measu ed me cu y con en in he soil close o he
kidneys, li e and lung (n
kidneys
=3,n
li e
=4,n
lungs
= 4 indi iduals)
e ealing he same end.
Fo he whole popula ion and a chaeological/his o ical amewo k o A
Lanzada, no e idence was ound o di ec use o me cu y ha caused an
acu e in oxica ion/high le el exposu e. On he con a y, i seems o ha e
been low and ch onic, h ough a mosphe ic pollu ion (Ál a ez-Fe nández
e al., 2020); a pa hway ha is mo e likely o explain a highe p opo ion
o he me cu y a iabili y han he u no e a e in he indi iduals s udied
he e. Rasmussen e al. (2017), also sugges ed ha di e ences in me cu y
con en o indi iduals we e due o ea men wi h me cu y-con aining
emedies, wi h me cu y concen a ions be ween 80 and 78,727 ng g
−1
in
co ical bone. Se e al s udies epo ed alues in kidneys and li e om
popula ions exposed o me cu y. Johansen e al. (2007) ound alues o
1400 ng g
−1
in kidneys, and 540 ng g
−1
in li e (in G eenland popula ions
wi h high fish consump ion), and Ga cía e al. (2001) epo ed alues o
250 ng g
−1
in kidneys and 140 ng g
−1
in he li e ( om a Spanish popula-
ion li ing nea o an indus ial a ea). Thus, he highe me cu y con en on
he ho acic and abdominal egion may be a p oxy finge p in o he so
issues placed he e, which include he main a ge o gans o me cu y
when exposu e is low and ch onic (Holmes e al., 2009). The influence o
he ho acic andabdominal a ea could be ex ended o he long bones placed
nea hem, especially he hume us. Depending on he bu ied body posi ion,
ulna and adius may be pa allel o he legs a he han o e he abdomen
(no e ha body posi ion in A Lanzada included s e ched a ms; López-
Cos as, 2015;seealsoFig. 1). In e es ingly, he e we find ha L07.20 ( adius
o e he ho ax a he ~L1 e eb a) and L06.22 (hume us) had highe
me cu y concen a ions han he a e age o bone ype 2. The e o e, bone
could be ac ing as a sou ce o me cu y o he soil, bu also as a sink o
he me cu y eleased om he body's so issues du ing decomposi ion
(c . Ál a ez-Fe nández e al. (2021)).
4.2. Bone s uc u e indices
Bone IR indices ela ed o hyd oxyapa i e mine al s uc u e (IRSF and
MMI) co ela ed nega i ely wi h me cu y con en , sugges ing ha me cu y
ends o be highe in newly- o med bone issue, and lowe in ma u e bone
issue. The ela ionship wi h he mine al ma u i y index (MMI) and he
in a ed spli ing ac o (IRSF) indica es ha Hg con en inc eased in bone
a eas ha we e ecen ly c ea ed o emodelled (Fa lay e al., 2010).
Hyd oxyapa i e con en and c ys al size inc ease wi h bone ma u i y,
nega i ely a ec ing Hg e en ion in he bone s uc u e since la ge c ys als
ha e lowe specific eac i e su ace o binding. As expec ed, ca bona e/
phospha e (CP) and amide I/phospha e (AmP) a ios co ela ed posi i ely
wi h bone me cu y. This is consis en wi h hei nega i e co ela ion wi h
IRSF and MMI. New- o med and emodelled bone issues ha e a highe p o-
po ion o collagen fib es and highe a es o C0
3
/PO
4
(Akkus e al., 2003)
since hey a e less mine alized. Di e ences obse ed be ween new and
emodelled bone (high me cu y) and ma u e bone (low me cu y) could
espond o wo di e en p ocesses: i) an an e-mo em p ocess: me cu y is
eleased o dilu ed due o bone issues ma u a ion and mine aliza ion,
since bo h la ge c ys als and lowe collagen con en limi me cu y e en-
ion; ii) a pos -mo em p ocess: me cu y eleased du ing body decomposi-
ion is easily accumula ed in new and emodelled bone due o i s la ge
collagen con en and smalle hyd oxyapa i e c ys al size.
IRSF and MMI showed no di e ences ega ding indi iduals, consis en
wi h p e ious esea ch showing independence o he deg ee o bone
mine aliza ion wi h espec o he age and sex o adul s, and mo e ma u e
indi iduals (Boi in and Meunie , 2002). Di e ences in he deg ee o mine -
aliza ion seem o be ela ed o he age o he issue (Bala e al., 2013). Fo
N. Ál a ez-Fe nández e al. Science o he To al En i onmen 851 (2022) 158015
8
bone issue, he deg ee o mine aliza ion depends on he u no e a e
(Boi in, 2007). Howe e , di e ences we e no ound o IRSF and MMI
be ween bone ypes 1 and 2, ha ha e di e en u no e a e (Hedges
e al., 2007). The e a e no di e ences in AmP be ween L06 and L07, in
ag eemen wi h s udies om biopsies (Danielsen e al., 1994) and au opsies
(Wang e al., 2002), in which di e ences in collagen s uc u e –in non-
pa hological indi iduals –we e only ound o indi iduals olde han
~60 yea s. These s udies indica ed ha changes in collagen a e age-
independen and ha changes in he u no e a e may exe some con ol.
Pa hologies like os eopenia can modi y he bone u no e a e al e ing
collagen con en (Bala e al., 2013;Paschalis e al., 1997), which may
explain why AmP is highe in L06 and L07 compa ed o L01, as L01 was
an elde ly woman wi h os eopenia and L06 and L07 we e a ju enile and
a ma u e adul wi h no pa hological ma ke s. AmP is also highe in bone
ype 1 han in ype 2, pe haps also ela ed o hei di e en u no e a es
(Akkus e al., 2003), and he ac ha long bones a e la gely mine alized
wi h a hick co ex, especially hose which a e ela ed o locomo ion.
Conside a ion also may be gi en o he mic o-sampling s a egy, and o
he ac ha no me aphyseal a eas we e sampled o L06, so small
di e ences be ween samples can be ela ed o bone a iabili y and only
consis en co ela ions a e conside ed he e.
4.3. Me cu y in a-skele al a iabili y
The h ee models (G0, G1 and G2) explained an ele a ed p opo ion o
me cu y a iabili y as well as ha o bone hyd oxyapa i e (>90 %), while
o bone collagen he pe cen age o a iance explained was high o mode -
a e(~60 %). G1 and G2 di e ed on hei p edic i e powe o soil me cu y;
G1 explained 17 %, whileG2 explained60 %. In G1, bone me cu y had a no
significan e ec on soil me cu y wi h a o al pa h e ec o 0.18. Fo G2,
bone me cu y had a significan posi i e e ec on soil me cu y wi h a o al
pa h e ec o 0.37. The PLS-POS g oups seemed o be esponding o di e -
ences in bone ossifica ion (co ical hickness) and su ace chemical al e -
a ion (ab asion), being bo h mo e significan ac o s in G1 han in G2.
Model G2 is mos like G0, indica ing ha he samples in his segmen
domina ed he gene al model. The main di e ence be ween he POS
models elies in he bone componen 's in e ac ion wi h bone and soil
me cu y. The bone mine al phase had a significan impac on bone Hg in
he h ee models, while bone collagen was only ele an in G1. The bone
mine al phase componen s a ec ed soil me cu y in opposi e ways in G1
and G2. Thus, POS models seem o ha e cap u ed in a-skele al a iabili y
ela ed o he weigh o he bone componen s on me cu y accumula ion.
Hyd oxyapa i e (HAP) had a posi i e e ec on bone me cu y in G0 and
G2, while i was nega i e inG1. Collagen was no ele an in G0and G2 bu
domina ed bone me cu y a iabili y in G1. The e o e, he PLS-POS g oups
esponded o di e ences in he bone ac ion con olling me cu y con en .
In G1, he con ol was exe ed by he o ganic ac ion o he bone
(i.e., collagen), while he ino ganic ac ion (i.e., hyd oxyapa i e)
con olled he me cu y con en in G2. Bone is composed on a e age o
~70 % HAP, ~20 % collagen ( ype I), and ~8 % o wa e pe weigh
(Auga and Scho lemme , 2006;Cu ey, 2008). As bones o ype 1 a e
expec ed o ha e highe u no e a es, hen he p opo ion o HAP/
collagen will be p esumably lowe han in g oup 2 (Akkus e al., 2003).
Thus, small inc eases in collagen in ype 1 bones could ha e a la ge impac
on me cu y accumula ion, emphasizing he con ibu ion o he less
abundan componen . This di e ence in he bone ac ion con olling
bone me cu y con en eflec s a di e en beha iou ela ed o he an e-
mo em s a us o he bone issue s uc u e. Me cu y ends o bind o so
bases like S, P, and N; bonds wi h S-con aining g oups being especially
s able (Schus e , 1991). HAP can s abilise Hg as (Hg)
3
(PO
4
)
2
(Ce ini-
Sil a e al., 2021). Bone collagen s uc u e is cha ac e ised by a epea ing
amino acid mo i (Gly-X-Y), whe e X and Y can be any amino acid
(Hulmes, 2008). All amino acids con ain N-g oups, bu Cys eine (-SH,
Cys) and Me hionine (-S-CH
3
, Me ) also con ain S-g oups. Gauza-
Włoda czyk e al. (2017) es ima ed ha he amoun o Cys and Me pe
100 g o collagen in bo ine bone we e espec i ely 0.32 g and 1.07 g.
Hence, he e is he po en ial o o m Hg S compounds, and hey may
ha e an an e-o pos -mo em o igin. To ou knowledge, he mechanisms o
me cu y inco po a ion in o he bone du ing li e a e unknown, bu his pa h-
way canno be disca ded. On he o he hand, du ing he main phase o so
issues b eakdown in he pos -mo em span, he educing condi ions needed
o o m Hg S compounds a e me (Janaway e al., 2009;Schus e , 1991).
The p esence o compounds ha bind me cu y s ongly explains why he
pa hway om bone me cu y o soil is no significan when collagen con ols
bone me cu y con en , i.e., he elease o me cu y o he soil may be e y
low un il he collagen s uc u e is s ongly deg aded.
The ela ionship be ween hyd oxyapa i e c ys allini y (HAPc)
(i.e., deg ee o c ys allini y) and bone Hg was bo h nega i e and posi i e
in he h ee models sugges ing a dual ole o bone c ys allini y. This can
be ela ed o he po en ial o he skele on o be bo h p ima y/seconda y
sou ce and sink (Ál a ez-Fe nández e al., 2021). The models sugges ha
HAPc a ec s bo h bone Hg elease (p ima y/seconda y sou ce) and e en-
ion/accumula ion (sink; we canno disce n e en ion o bone an e-mo em
con en om me cu y inco po a ed as he esul o he in e ac ion wi h
so issues du ing hei decomposi ion). In a chaeological bone, as
hyd oxyapa i e c ys allini y inc eases, bone s uc u e is al e ed (Nielsen-
Ma sh and Hedges, 2000) and, hus, as ound he e, a nega i e ela ionship
wi h collagen and bone hyd oxyapa i e s uc u e is expec ed. In G1 HAPc
has a di ec nega i e e ec on bone me cu y, as bone me cu y is con olled
by collagen. Bu he e is also a sligh ly significan posi i e indi ec e ec ,
ela ed o HAPc as a p oxy o bone al e a ion. Despi e bone me cu y
being con olled by collagen, i is e y likely ha HAP also con ains
me cu y and as HAP is al e ed bone me cu y will be los . In G2, HAPc has
a di ec posi i e e ec on bone me cu y, eflec ing ha HAP con ols me -
cu y con en in his g oup. When conside ing he an e-mo em condi ion o
hese bone samples, he highe he HAPc he highe he p opo ion o HAP
(Akkus e al., 2003). Bu , when HAPc inc eases as a esul o pos -mo em
bone al e a ion, he any accumula ed me cu y in HAP can be eleased.
Loca ion (i.e., he sample posi ion ega ding he pel ic a ea) has a
nega i e e ec bo h on bone and soil me cu y con en . Samples close o
he ho acic/abdominal a ea had highe me cu y concen a ions compa ed
o hose a he away. This is consis en wi h wha has been al eady dis-
cussed in Sec ion 4.1., and ela ed o he loca ion o he main a ge o gans
o me cu y in he ho acic/abdominal a ea (Be lin e al., 2015;Ga cía
e al., 2001). I is known ha bones may ac as sink o some elemen s du ing
body decomposi ion and hen elease hem o he bu ial en i onmen
(Hedges, 2002). The ela ionship be ween loca ion and bone me cu y
con en s ongly sugges s ha bones ac as a empo a y sink o me cu y.
4.4. Role o soil componen s
Mos soil p ima ysilica es lack he capaci y o e ainme cu y (Schus e ,
1991), consis en wi h hei dilu ion e ec on bone and soil me cu y
con en . The highe he amoun o p ima y silica es he ewe soil compo-
nen s wi h he po en ial o e ain me cu y. Clay had a sligh ly nega i e
e ec on bone me cu y con en . The sil +clay ac ion is a eac i e compo-
nen o he soil and ends o be en iched in me als (Qin e al., 2014). The e-
o e, a highe sil +clay con en leads o an inc ease in me cu y
concen a ions (Ál a ez-Fe nández e al., 2021;Cooke e al., 2020;
Schus e , 1991) ha limi s he amoun o me cu y ha emains a ailable
o be s abilised in bone.Howe e , he weigh o he clay in he h eemodels
is low (−0.13 G0, 0.01 G1, −0.17 G2). Bone me cu y shows a significan
di ec posi i e e ec on soil me cu y in G0 and G2. This suppo s he idea
ha skele ons a e po en ial sou ces o me cu y o he soil/sedimen s in
he bu ial en i onmen (Ál a ez-Fe nández e al., 2021). I also suppo s
he con en ion ha bone me cu y concen a ions a e no diagene ic when
he geology o he a ea does no con ain me cu y o es, based on he da a
om a chaeological skele ons (Emslie e al., 2015;Kepa e al., 2012;
Rasmussen e al., 2013a;Walse e al., 2019;Yamada e al., 1995). This is
also likely he case o he bu ials s udied he e because soil/sedimen
me cu y concen a ions a e on a e age 0.9 ± 0.7 ng g
−1
in he dune
sands laye whe e he skele ons we e bu ied (Ál a ez-Fe nández e al.,
N. Ál a ez-Fe nández e al. Science o he To al En i onmen 851 (2022) 158015
9
