j
, ?$ . ~nl~t!kll
0730-7265101 59.00 ?
~nvironmenruiTonicoiagy and Chemtstry. Vol. 20, No.
10- PP. 1276-2286
Printed in
ill;
EF
AN EVALUATION OF METHODS FOR CALCULATING MEAN SEDIMENT QUALITY
GUIDELINE QUOTIENTS AS INDICATORS OF CONTAMINATION AND ACUTE
TOXICITY TO AMPHIPODS BY CHEMICAL MIXTURES /
ntiinar
,\rscnic
~;)dnial
cilroinlt
coppel
RUSSELL FAIREY,"?
EDWARDR. LONG,,' CASSANDRAA. ROBERTS,?BRIANS. ANDERSON,$ BRYN M. P H I L L I ~ ~ , ~ic;,d
blcrcnr!'
JOHNW. HUNT,$HOWARDR. PUCKETT/I
a n d CRAIGJ. WILSON# silver
tMoss Landing Wlariile Labosatones. San rase State University Foundation, 754.4 Sandholdt Road, Moss Landing, California 95039, USA
jNOAA, NCCOSICenter for Coastal Monitoring and Assessment, 7600 Sand Point Way NE, Seattle. Washington 98115, USA
$University of California-Davis. Marine Pollution Studies Laboratory, 34500 Highway 1, Monterey, California 93940, USA
I/Caiiiornia Department of Fish and Game, Marine Pollution Studies Laboratory, 34500 Highway I, Monterey, California 93940. USA
Kalifornia State Water Resources Conrrul Board, Bays and Estuaries Unit, PO Box 944213, Sacramento, Califorma 924W-2130, USA
Zinc
.kitill
i
e~~driil
Toel PC
(Received 5 July 2000; Accepred 18 F e b r ~ ~ o2U01)
iy Abstracl-Mean sediment quality guideline quotients (mean SQGQs) were developed to r$presenl the presence ofchemicalmixtures
in sediments and are derived by normalizing a suite of chemicals to their respective n~mcricaisediment quality guidelines (SQGsj.
Mean SQGQs incorporate tlle number of SQGs exceeded and the degree to which they are exceeded and are used for comparison
with observed biological effects in the laboratory or field. The currcnt research makes it clear, however, that the number and type
of SQGs used in the derivation Of these mean quotients can influence the ability of mean SQGQ vaiues lo correctly predict acute
toxicity to mmine amphipods in laboratory toxicity tesu. To determine the oprimal predicrive ability of mean SQGQs, a total of IS
differen1 chemic3i combinations were developed and compared. The ability of each set of meall SQGQs to correctly predict the
presence and absence of acute toxicity to amphipods was determined using three independenl databases (n = 605. 7753, 726).
Cnlculiited mean SQGQ values for all chemical combinations ranged from 0.002 to 100. The mean SQGQ that was most predictive
of ncute toxicity to amphipods is calculated as SQGQl = ((1([cadminm]l4.2l)~[copperj/270)([lead]lll2.l8~([silve~lll.77)~[zir~c]l
ilO)([total clilo1-danel/6)([dieid1inll8)([toml PAI-I,,lIl,S00)([toral PCE]l400))/9). Both the incidence and magnitude of ncute toxicity
to amphipods increased with increasing SQGQl values. To provide better comparability between regions and national surveys,
SQGQl is recommended lo serve as the standard method for combination of chemicals and respective SQGs when calculating
mean SQGQs.
ICeywords-Sediment
quality guideline quotients
Amphipod
Dl
.rota1 dl
llieliiilil
eo
tiigh )nr.
LOW
,
,
ERhlQ
used an
alUrndU
chlorin;
tles for
studies
iodivid!
present or absent [12]. Chemical mixtures can also be reprcbinatior
sented using SQGs and are generally calculated bynormalizing
initially
each chemical found in the sediment to its respective SQG
irnents :
value and then averaging the resulting norinrtlized values for
parison.
a given suite of chemicals. These normalized chemical sum. , tested u
maries (hereafter referred to as mean sediment quality guide.
flirtr re
ntteinpt
line quotients, or mean SQGQs) represent complex chemic~l
ability !
mixtures within each unique sediment sample as a quantitative
the I? c
numeric value that incorporates both the magnitude and num.
01' cliet~
bet of SQGs exceeded. Mean SQGQs can be used to predict
lines (1
the probability of biological effects in the laboratory or field
!';uiaus
as previously demonstrated [I?-141. In this article, we report
respecti
and compare a variety of methods for calculating mean SQGQs
In a
with the objective of proposing a standardized method for
allditioi
SQGQ calcuiation that improves the ability of SQGQs to PIC
ot indi.
dict whole sediment acute effects on amphipods.
indivio
lMETHODS sensus
were st
This research evaluated the type and number of andytes in
methoc
the SQGQ calculation to find chemical combinations that best
predicted biological effects, as indicated by marine arnpl~po~ Plllpod
latti'e c
mortality in sediment toxicity tests. Mean SQGQS we:e
Sampie
culated using effects range-mpdian (EP\hil) [15,16] and
of 15 ir
effects level (PEL) 1171 SQG values for nomelizing sedime"'
ical cr
chemical concentrations. Tbe ERM and PEL sediment 4~1:d~'Y
~QGQ
guidelines have been publibhed for 10 individual t a c e memis
three individual pesticides. !
: individual polycyclic ~ o n l ~ " " chen71~
h~drocarbons IPAHsl, three sroupin~sof individual PAHs'
Toxicity
Chenlical mixtures 8
INTRODUCTION
Althottgh the use ofempirically derived sediment quality
guidelines (SQGs) in sediment monitoring and assessment has
been die subject of debate, recent reports suggest SQGs continue
to be widely used to predict when chemical concentrations are
likely to be associaled with a measurable biological response
[I-51. Use of SQGs has been encouraged by recent research
that indicates reasonable predictive ability of SQGs [6], in combinatio~iwith limited application or regional specificiry of chemical guidelines derived through other methods [7-1 I]. The increasing use of SQGs results from a practical need for protective
management tools when identifying areas where anthropogetlic
chen~icalsmay present a risk to benthic biota. Utifortunarely,
the inappropiiate application of SQGs has resulted in criticisn~
of their use in sediment management. Tliere is a need to continually r e e x a ~ n ethe appropriate use of SQGs as management
lools and to refine uses of SQGs to better predict toxicity and/
or biological community impairment.
E~npiricallyderived SQGs have been generated from laree
sets of sytlupticaily collected chemical and biological data 161.
Sediment quality guidelines were developed primarily from
field-collecred sediments using statisrical approaches lhat 2ssociate chemical concentration and biological response. They
were established to demonstrate the individual chemical concentrations i t which biological effects were expected to be
'
To whom conesponacnce lllay be ~ddressed
!hue:> @mirnl.caislate.eduj.
,
Env~run.Toxicol. Cltenr 20, 2001
c,lculc!ion of m.:an sediment quality guideline quotients
2277
yble 1. Mean sediment quality guideline quotients (SQGQ) chemical combinations using effecrs range median (ERM) sediment quality guideline
(SQG) values
C-
Mean SQGQs
using
ERM guldehnes-
/
-
ERMO 1
~~limOnY
~rssnic
cadmium
~l~roni~um
COQP~~
~ead
~ercury
Silver
Zinc
Tolsl DDT P C )
~ ~ chlordane
t d
Dieldrin
Endtin
Tc,tal PCBs
LOW
mol wt PAHs
High mol wr PAHs
ERMQ2
Arsenic
Cadmium
Cliromium
Copper
Lead
Mercury
Silver
Zinc
?-Methylnaphthalene
Dibenz[a,lt]anthracenc
Acenuphthene
Acenaplithylene
Anthracene
Benz[a]anthracene
Benzo[alpyrene
Chrysene
Fluoiene Fiuoranthenr Naphthalene Pllenantlirene Pyrene p 'p '-DDE Toral DDT Total PCBs ERMQ3
Cadmium
Chromium
Copper
Mercury
Zinc
Total DDT
Dieldrin
Total PCBs
Low mol wt PAHs
High mol wt PAHs
ERMQ?
Antimony
Arsenic
Cadmium
Chromium
Copper
Lend
Silver
Zinc
Total DDT
Total chlordane
Dieldrin
Total PCBs
Low mol wt PAHs
Total PAHs
ERMQ5
Arsenic
Cadmium
Chromium
Copper
Lead
Mercury
Silver
Zinc
Toral chlordane
Dieldrin
Total PCBs
Low mol wt PAHs
High mol wt PAHs
ERMQ6
Arsenic
Cadmium
Chromium
Copper
Lead
Mercury
Siluex
Zinc
Total chlordane
Dieldrin
Total PCBs
Low mol w t PAHs
Total PAHs
-
'C:<MQ cLen:ccl comhtn1l:ons us'd c~iccisr lnne rn:di~i~SQGs :Ij.Ihl :.icntlcll uzllcs in1 mu-11q~0r1~1.r
CJI:LIIJIIOP C\CC:II ERhlQl. \\III;.I
LcrJ .. org:nlc cn:o-n norn;ilzed SQG (351 t31~ciur 10131 DDT OC = 3 r p n l c ; l r J i n : PCHa = ?olyc5.lsr.ni:e.i Slphrn~lsP!%H, ,<d!c\cl~c
mmstic hydrocarbons; mol wt = molecular weight
two summed organochlorine
pesticides, and the sum of 18 polychlonliated bipheiiyl congeners (PCBs) /IS-171. Guidelinevalues for these 32 chemicals were selected because of published
s~udiesdemoiistrating the reliability and predictive ability of lhe
illdividual guideteiiiie vaiues [6,16]. Two different chemical combinations that were used in previous studies were compared
initially and evaluated for their ability. to correctly classify sediments as toxic or nontoxic. Based on the results, of this comparison. 10 additional chemical combinations were sequentially
lesled using various combinations of the same chemicals and
their respective SQGs. Each successive colnbination was an
"lernpt by tile investigators to improve mean SQGQ predictive
:lbility based on the results of the previous combination. Six of
12 calcuiatio~lmethods were based on various combinations
~~chenlicais
that were divided by their respective ERM guidelines (Table 1). Six PELQ calculation methods were based on
"'L'.ious con~binationsof chemic.dls chat were divided by their
rciPective PEL guidelines (Table 2).
In an attempt to evaluate mean SQGQs that incorporated
idditionaltypes of SQGs, an assessment of the predictive abilig,
Yindividual SQGs was perfornled. The assessment included
'lldividualSQGs developed using correlative approaches. con'ensus approaches, and theoretical approaches. Individual SQGs
selected for inclusion in meal? SQGQ quotient calculation
!"etiiods when the values best predicted acute toxiciry to amfor that ccheinical or chemical class and were represenP1ilp~ds
jaiiveoichemical concenua1ions most commonly found in field
L"mPles
from the database (italicized entries in Table 3). A total
l5 individual SQGs were selected for use in additional chem:ql , 'ombinatiotis. These SQGs were incorporated into six VG9calculation methods that were based on combiiiatio~isof
"'emical~di\'ided by a mixture of respective E m , PEL. or
""cr se!tcrcd indjvidl,:,i SOGQ:T;~~I-1) Each mean quotielit calculatio~imethod included chemicals
and their respective SQGs for each of four major chemical
groups (metals, pesticides, PAHs, PCBs). A total of 18 chemical combinations were evaluated.
All mean quotient calculations attempted to use the same
summalion methods for cheinical classes (total chlordane, total
DDT. total PCBs, low ~nolecularweight PAHs, high molecular
weight PAHs. total PAHs). This step was necessary for valid
comparison of mean quotient calculations but presented iimitaxions with large lnultistudy data sets because of differences
in analyte lists among studies. The methods for summation of
chemical classes used in this study are give11 in the Appendix
[lj,lS-201. Itidividual chenucal quotients are calculated by
divicing the measured concentration or class summation concentration in the sediment sample by the respective publisiied
sediment quality guideline value for each cheniical for which
individual SQGs were derived. A value greater than one indicales the chemical concentratioti in that sample exceeded its
respective SQG. The mean SQGQ is obtained by calculati~~g
the inesn of the resulting individunl quotients for a given combination of chemicals. A generalized example of the calculation is
Inem sediment quality gmde!ine quollent
where 17 = total nu~iiberof analytes. Sampits that were found to have chemical concentrations Table 2. Mean sediment quality guideline quotient (SQGQ:) chemical combinations using probable effects level (PEL) SQG \
Meon SQGQs using PEL guidelines"
PE'LQI
Arsenic
Cadmium
Chromium
Copper
Lead
Mercury
Silver
Zinc
Total DDT (OC)
Total chlordane
Dieidtin
Lindane
Total PCBs
Low rnol wr PAHs
High mol wt PAHs
PELQ2
Cadmium
Chromium
Copper
Lead
Mercury
'
Nickel
Silver
Zinc
2-Methylnaphthalene
Dibenz[a,h]mtI~racene
Acenaphthene
Acenaphthylene
Anthracene
Benz[a]anthracene
Benzo[nlpyrene
Chrysene
Fluorene
Fluoranthene
Naphthalene
Phenantluene
Pyrene
p 'p '-DDE
Total DDT
Total PCBs
PELQ3
Cadmium
Cluomium
Copper
Mercury
Zinc
Total DDT
Die!dnn
Total PCBs
Low mol wt PAHs
High rnol wt PAHs
PELQ4
Arsenic
Cadmium
Chromium
Copper
Lead
Silver
Zinc
Total DDT
Total chlordane
Dieldrin
Total PCBs
Low rnol wt PAHs
Total PAHs
PELQ5
Arsenic
Cadmium
Chromium
Copper
Lend
Mercury
Silver
Zinc
Totai chlordane
Dieldrin
Lindaue
Total PCB8
Low mol wt PAHs
High mol wt PAHs
P E L Q ~ ~
Arsenic
\
Cadmium
Chromium
Copper
Lead
Mercury
Silver
Zinc
Total chlordane
Dieldrin
Lindane
Total PCBa
Low mol wt p ~ ~ , ~
Totnl PANs
--
'PELQ chemical conlbinations used probable effects level (PEL) SQGs [I?] as critical values for mean quotient calculariaos except f o r p ~ ~ ~ ,
which used on organic carbon normalized SQG [331 value for total DDT. OC = organic carbon: PCBs
polycyclic aromatic hydrocarbons; mol wt 5 molecular weight.
of the summation. to a value of one half of the given method
detection limits. When one or more required chericals were
not analyzed for a particular sample, that sample wns removed
from the base data set and not included in evaluations of mean
quotient calculation methods requiring that chemical. This step
was necessary for valid comparison of the summation methods
and the data sets to which they could be applied.
This study used chemical and biological data from tluee
readily available sources as the basis for comparisons of predictive ability. The sources were the State of California's Bay
Protection and Toxic Cleanup Program database (BPTCP), the
National Oceanic and Atmospheric Admilustration's (NOAA's)
national sediment toxicjty (SEDTOX) datnbase, and NOAA's
Biscaylle Bay survey in Florida. The BPTCP databaseincluded
605 coastal sediment sarnples for which synoptic chemical
analyses and toxicity tests (Eohalotori~rsesr~rariiisand Rhepo.~yliiuabronius) were performed [?I-271. These data were
used for initial comparisons between calculation methods developed by Fairey et al. 1211 and Long et al. [6] and 10 additional chemical mixture comhinations using ERM and PEL
guidelines for mean quotient calculation. These dala also were
used to evaluate the predictive abiliry of a variety of individual
SQGs with the objective of evaluating the predictive ability
of six nddirional chemical mixture combinations that combined
SQGs from different studies. The SEDTOX database included
3.753 sediment samples from studies throughout the coastal
United States for which synoptic chemical analyses and whole
sedin~entacute toxicit?. rests (Anzpelisca abdiia, R. nbronius)
were paired [2Sl. The SEDTOX data was used to confirm
predictive accuracy of mean quolient calculation methods using BPTCP data and to further evaluate mean quotient calculation methods thal included SQGs from mixed sources. The
Bisc~yneBay database [291 included 226 ;an>ples analyzed
=
polychlorinated biphenyls;P A H ~ ;
for sedimew chenlistry and loxicity (A.abdiCnJ and was used
as a third independent database for testins predictive accuriitj
Chemical data were. combined in dBase IV darabase filesmj
manipulated using dBase command proganls. Mean SQGQ niues derived from the various unique chemical combinntiooslu
each sample then were compared with synoptically measud
roxicity test responses to assess how well each cbemicd cmn.
bination predicted acute toxicity. Samples were classified as tonic
for the BFTCP study when sample survivd was signihcaiil!
different from controls and less than a criticd value dele~minc!l
using the minimum significant difference 90U1percentile tla\rd\
generated from BPTCP data (Eohausrori~rsesr~mrinssurvivd I*
sample was less thw 75% of survival in negative controis. 1
abronius survival in sanlpie was less than 77% of survii.d ill
negative controls) 17-1-27,7,30]. Samples were classified as 'oxK'
for the remaining smdies when samples were significanti!' dii
ferent from controls and less than SO% of negative conuols l3I!
MacDonald et al. 1321 compared mean quotient v31ues
five critical levels fto assess associations with measured bi":
logical effects (<0.1, <O.j, >0.5, >1.0, and >1.5). Longan"
MacDonald 117-1 used similar levelsbut also incltided 3"'*
ditional level (>1.3) to better evaluate greater mean SQGU
values. Seven critical levels (Tables 5 & 6) were used in
current study, with one additional cnticai level (LO.?) inc'udd
to better evaluate mean SQGQ values in the lowel' I~""'
where the majority of samples fell. In these analyses, we compared both the incidence '@' cenrages) of acute toxicity and average amphipod .veria
suly'~'
Ikf
above and below mean SQGQ critical values. The cn.
identification of oplimal predictive ability wei,e [hat
dence of toxicity increased steadily and average ~u'v'v" IdC'.
creased steadiiy as mean SQGQs increased. incidence $ < ~ I I ~ ~ '
icity wss IOU: (.=5%) and average survival hish (--.
'
'
-.Yo%)
ralcj]a~ionof mean sedimcflt quali? guideline quotients E~zviro~z.
Toxicol. Chen;. 20, 2001
3279
3. Sample counts and the incidence of acute loxicity to amphipods when effects mnge medians (ERMs), probable effects levels (PELS),
individual sedinxenc quality guidelines (SQGs) were exceeded using the Bay Prot6ction and Toxic Cleanup P r o s a m (BPTCP) data (,,
= 605)
o:hel.
-
C>&z&
No.
No.
NO.
sam-
"The number of samples exceeding individual SQGs and the percentage of tliose samples determined to be toxic were used i , ~
selecting which SQGs were reasonably plsdictive of coxicity to amphipods and were representative of concentrations comtnonly found in BPTCP samples. Selected SQGs (ili~iics)were used in mean calculation methods that incorporated individual SQGs from a variety of sources. PCB = poly-
cLlorinaledbiphenyl: PAN = polycyclic aromatic hydrocarbon: OC = organic carbon: N A = not applicable; mol wt = molecular weigl~r. ''!!GI,
'117j.
-
4 . Mean sediment quality guideline quolienL (SQGQ) chemical combinations using eCkcts range median (ERM) probable effects level
-
(PEL) and other individual SQGs values
Mean SQGQa uslng combinations of zndxv~dualSQGs"
\
"GQI
ca6rm"m
Co?per
Laid
Silver *. Llnc
/bra1
Clllordvile
Jieldrin ha1 P C B ~ 'rclta~P A H ~
SQGQ?.
Cadmium
Chromium
Copper
Lead
Silver
Zinc
Totni chlordane
Dleidrin
Endrin
Total PCBs
Tom! PAHs
SQGQ3
Cadmium
Chromium
Copper
Lead
Silver
Zin~
Lindane
Dieldnn
Endrin
Total PCBs
Total PAHs
SQGQ5
sQGQ4
Cadmium
Chromium
Copper
Lead
Silver
Zinc
Total chloidane
Dieldrin
Toul PCBs
Total PAHs
Arsenic
Cadmium
Copper
Lead
Siiver
Zinc
Total chlordane
Dieldrin
Lindane
Total PCBs
Total PAHs
. :
1
SOG.
'2 cllernical combinations used ERM, PEL, and other individrlnl SOGs $15 rricir-l
bip!lto!Jl:?.:\,.is = "o]\,cuc,i" .,.em." , . . '....'. SQGQG
Antimony
Arsenic
Cadmium
Chromium
Cupper
Lead
Mercury
Silver
Zinc
Toral chiardane
Dieldriii
Endrin
Lindane
Total PCBs
Total PAWS
--
6950
LLSU
-
~VL~O
IO.~LCOI,
I I . Clrern. ZU, ZUVI
R. Fairey tt q,
,
Table 5. Comparison of mean sediment quality guideline quotient chemical combinatio~isusing the Bay Protection and Toxic Cleanup proerl
-,m i
(BPTCP) data in = 6057; ERM = effects range median: PEL = probable effects level
SQGQ rangec
22.3
21.5
21.0
>0.5
cO.5
<0.2
<0.1
SQGQ rangef
22.3
21.5
>1.0
>0.5
<0.5
<O.?
<O.l
SUGQ ranges
ERMQI
(a = 528)b
ERMQ2
(n = 395)
n = 6. 100%
n = 17, 88%
n = 50, 78%
r~ = 122. 63%
rz = 406. 37%
,z = 227, 36%
n = 65, 31%'
n = 9, 33%
11 = 13, 46%
rz = 31.42%
n = 94, 50%
n = 301, 37%
n = 175, 36%
n = 65, 39%
ERIVIQ3
(11 = 569)
,z = 24, 50%
rt = 46, 61%
rr = 92, 62%
n = 205, 51%
n = 364, 38%
,L = 149, 38%
n = 38, 29%
PELQl
PELQZ
PELQ3
(n = 528)
(n = 394)
(n = 569)
n = 14. 100%
12 = 50, 76%
n = 91, 70%
n = 197, 53%
rr = 331. 36%
n = 88, 29%
rz = 29, 21%
= 20, 35%
n = 45, 44%
n = 36, 50%
n = 127. 48%
n = 267, 36%
a = 118, 37%
!I = 33, 34%
rt
SQGQl
(n = 561)
rt =
n =
n =
rz =
rr =
rt =
n =
39, 59%
85. 64%
142, 58%
274.51%
295, 35%
66, 35%
7, 14%
SQGQ2
SQGQ3
(n = 561)
(n = 563)
ERMQ4
!n = 533)
ERMQ5
( n = 533)
= 19. 58%
34, 65'7"
52, 63%
161, 53%
372, 38%
182. 37%
38, 19%
= 9, 100%
n = 23, 83%
it = 63, 78%
n = 147. 57%
r~ = 386, 37%
n = 202, 36%
rz = 53, 23%
n=
n =
rt =
n =
r~ =
n =
rt
PELQ4
= 528)
n
rr
r~
n
n
n
29, 66%
61. 66%
='113. 63%
= 241, 49%
= 287, 37%
= 63, 33%
= 2. 0%
=
=
SQGQ4
(n = 561)
= 8 , loo>
n = 20,85%
n = 55, 78% n = 136.60% 11 = 395. 36% n = 212.35%
17 = 53, 23%
11
PELQ5
(n = 533)
(JZ
tt
ERMQ6 (n = 531) n = 16,94%
rt = 55, 76%
n = 95. 69%
= 202,5390
n = 331, 35%
n = 81, 30%
it = 5 , 0 %
,I
sQGQ5
(n = 528)
PELQ6
= 531)
(ll
.
n = 13, 100%
n = 48, 75%
n = 87, 74%
= 199,5J%
n = 332, 35%
n = 82. 29%
n = 5.0%
11
sQOQ6
528)
(11 =
-
> I n the BPTCP database, 250 sediment samples were acutely toxic to amphipods (41%) while 355 were nor toxic. bThe count reflects the number of BPTCP samples for which the particular chemical combination co~ldbe calculated. 'For each SQGQ range, the number of samples with mean quotient values exceeding the respective critical value are give11 as well as
l~ercentageof those samples exceeding the criricnl value that were acutely toxic to amphipods.
mean SQGQs were low (<0.1); incidence of toxicity was 1iigl1
(>YO%) and average survival was low (<50%) when mean
SQGQs were high (>2.3); and when predictive ability was
similar across SQGQ ranges for different calculation methods.
snmple counts were compared to find methods that cotiectly
predicted toxicity f o r the largest number of samples in each
data set.
RESULTS
As sliown by tile sample count for the BPTCP database
(Table 5). the number of samples that lend themselves to a
partiiclar calculation method differs dramatically among
methods (11 = 394-569 out of 605). This variability is due to
periodic expansiolis of the analyte list for the BPTCP over the
program's seven years. In general, the lower the number of
analytes included in the calculation method, the more likely
the derivation could be completed for the majority ofsamples.
Results demonstrate an increased number of samples being
available for SQGQ calculation and biological comparison
when SQG che~nicalswere limited.
Using the BPTCP database, toxicity to marine amphipods
[E. csruarius and R. nbronius) in 605 sediment samples was
compared with calculated mean quotient values. Based on
BPTCP conlparisons with controls and the minim~imsiznificant difference. 150 samples were classified as acutely roxic
141%)and 355 were classified as nontoxic 159%). Calculared
mean quorient values irom sedimenr samples for all derivat~on
methods ranged fro111 0.031 (lowest multiple chemical indic:~tor) to 10.9 (greatest multiple chemical indicnror).
rile In the BPTCP data. there was a wide range in the patterns
in toxicity among the different calculation niethods (Table 51.
Generally, SQGQ values above the greater critical values wen
associated with greater proportions of toxic samples wliile
those below the lower critical valt~eswere associated will]
reduced proportions of toxic samples. In two cases (ERMQ?,
PELQZ), the incidence of toxicity did not increase appreciably '
with increasing chemical concentration. In four cases
(ERMQ3. ERMQ4, PELQ3, PELQ4), the increase in toxicity
was lninimal with increasing chemical concentration. One hy. '
brid combinatioii 01several types of sediment quality guidelines for mean quotient calculation (SQGQG) resulted in tile
greatest association with the incidence of toxicity to ampillpods (100% for values >2.3 and 100 % for values >1.5). In
general. predictive ability improved when mean quotienl cdculations utilized SQGs developed using a variety of COP.? .
parative statistical approaches (e.g.; ERM, PEL; consensus,
equilibrium partitioning). Associations with the incidence oi
toxicit). were lowest for all methods when mean quotient Values dropped below 0.1.
Using SEDTOX data, 2.753 samples were examined to'
associations between toxiciry to marine amphipods !E. 6'
trmrillu. A. abdirn, and R. abronius) and all 1 8 mean quotie"'
caiculation methods. Of these samplesl 154 were classified*
toxic (18%) and 2.169 were classified as nontoxic (82%). T'q
lower incidence of toxiciiy refleccs the pl-imay program 9"'
of identifying general spatial status aud temporal trends &lon!
the coastal margins oi the United States rather than inves"'
sation of conlamitlared sites as with EPTCP Calculated med1
(
!
.I
+
0
o
r,
2
5
.
5
.-?
5
a
01
~ ! Y o1
i 0(! asaz* IZE'I al!lllli (%ZZ) s~od!ridn~s
01 qrol Lrain3e a.ra.3 sa~druasiriam!pas ILE 'aseqil~cpuounuo.7 P ~ z ! ~ L I ! u3111
!I~~
'3!x01 l o l l alaM IOZ al!q.n (%I I) spod!~[dme 01 ~ ! r u lL l a i n ~ eala".% sardrues loaul!pas
'aruqcilcp i c g ?t~koss!gaqi
'spod!l[drlm 01 >!xu1 ,S)aln~oa m \ \ I
?111on[c.>!l!n a!(] Pu![,3a3ra sa~dcuesasoq, 30 a 6 a l 1 ~ a ~ ~all3
a dse {lam se ua.>!8 a x a n p n le3!i!rJ a*!~=adsa.~
at11 Pu!paa~rasanleh ~ u a ! ~ o aoaaru
b
I!A\
saldu~esJO ~ ~ q r n r naclt
! 'aZou.~@ ) b ~~ ~ J C IJ
'palv-ln3le3 aq p1n03 LlO!lCU!qUIOJ [u3!1~lal[~
.ze[ns!iJed aql q3!q.3i 103 aseqelsp ipea o! saldmes $0 raqulnu a1n sIJaUa.1 (spasq ooln[on an11 )I! saa>[&:.! 2,) 1111103 ;.
.a!xoi IOU a.ram ,592'2 al!qhl ( % ~ 1 spod!l[dlue
j
01 31x01 d(aln3e axas+ s?~dmeslualo!pas PSI. 'arcqctup SO.LCIDS
O&P 'YgL
'160.1
%$I 'zOP'I
%L5 '062
%IS ' S I I
6 5 9 '55
8 6 s '82
=U
= u
= U
=
"
=
1'
= 21
= 1'
(269'1 = TI)
pbmaua
D h t l
(269'1 =
(269'1 =
1')
z01,vua
.-E-
=
I)
=
71
=
71
=
=
?I
-
=
11
z,
4hP '0s 1
%S '861
%9 '80:
B L L 'L[
%LQ '9
9 5 ~ 9'E
= 'I
=
="
= rr
"
= z,
=v
I,
( ~ Z Z= "1
9013;.1s
( 9 i =
~ 10
zbwaa
"2
o
%
U
G
.-
2
?B
2
.3
.-
a P 'Osg
%6 'LSO'1
% I [ '~ZP'I
SO9 'E9Z
O,lL '6L
%SL 'SP
%&s 'LI
(269'1 =
11)
=
~ t )~aldurusk
%S '1 € 1
%P'OSI
%P 'OOZ
%E9 'gS
%5L 'OZ
%E8 '21
n*001 's
= ti
= z!
= ri
= U
= a
=
=
~1
=
"
"1
%E 'Sc9
Oh9 '096
% € I '{LE'I
8 9 5 '12C
%SL ' S I I
%S8 '65
%P6 'PC
=
5,
=
7,
=u
= <#
=
= !P
%6L'61 =
"
I'
= rt
0.9 ' ~ L & ' I= t8
068 'Is'[= 7,
c?&ZS ' L i \ = ts
%ZL '9,; = '1
%p9 'Sz'= I!
O ~ S
'961'1
%PC 'El = u
.i
=g
U
l
%p
%17'881
%s '00Z
%6L '61
BOO1 '6
BOO1 '9
%(o,r
'&
'I
"
"
=
lt
( ~ Z Z= ")
= u
= tI
=
=u
= fi
= t,
"
(612 = 10
9000s
107ad
%& 'LES =
Y ~ ' FZ Z ~= )I
%S 'St0'1 =
BOfl
= Id
y,g$ -911 = 1,
%69 '8s = 'Z
"
OhIL '82 =
U
'291 =
%,f '981 =
Oh$ 'EOz =
O ~ P 'L&z =
Oh98 'PI =
'L
I,
W
ti
"
= 1,
%OO[ 'P =
( ~ Z Z=
U
"
"1
(ZhY'I =
2,
.
%Z 'La* = I 8
%E'999=U
B L '608 =
%St. '9EI = 11
9 0 9 'OP =
*bog 'SZ = tc
"
"
,I
"
"
"
1.0 1 '6 =
!<
"
!,
"
Oh& '619 =
% L '9t.6 =
%&r ' 9 ~ E . l=
' 9 1 ~=
%9L 'LO1 =
%9L 'SS =
~,L!o'
'ZE =
( ~ 6 9 ' 1=
030s 30
= u
= 8r
=
=
=
=
11
=
1,
I1
71
11
oh*
ii
?I
u
t,
"
11
,J
' ~ Z =
L
=
=
=
=
=
=
968 '61#1
'LLt.1
895 'S[t
'%us 'LZI
y b ~ s'29
,33001 'ZE
kt
I
u
i
U
l,
11
8,
"
0
c
t
0:)
1bo0s
--
<
,a2
(7.69'1 = !!I
10
~0'30s
CLOS!,C~NIUS
"
% * ' ~ 9 1=
'E61 =
%< 'fiOZ= I t
%8L '81 = I,
BOO1 9 =
%OOI 'E = l'
0-21
Oh*
"
OhP 'E91 =
ohp ' 8 ~ 1=
%P 'E()z =
0 . t ~
'57. =
0 ~ '51
~ =8
BOO1 'L =
%OO1 't =
( 9 ~ s=
~boDs
bobs )o
%C '9sg
%f 'Sc6
Oh9 '19['1
%Sf 'SOZ
%9L '8s
%IS 'TS
%6L '61
"1
(9zz = 10
11)
~090s
DdL '9PL =
%P'LM)'I=U
%S '90t.I = 1,
%9P '9L1 = Ct
UhZL '8s =
%LL 'OE = I(
%EL '$1 =
"
=
= 7,
= M
= U
= '1
= 11
tbsOs
%P 'SsL =
%P 'L81 = I,
61.'EOZ = "
%t.L '&Z =
'91 =
BOOT 'L = 1'
%OOl '9 =
(922 '=
sb!~Os
="
%& 'PO9
%S 'LL6
'62v'l
Oh95 '592
Oh&g '59
%SP 'EC
%I~Z'El
"
=
= U
= tt
=U
=
= 1,
(269'1 = L')
$bobs
SD!IOS
( p g z ~= I,) einp L e g a u L ~ ~ s !Bn!sn
g
,L~!a!xo] o! spoI(1am ua!le[n3le3
ohp '90s
ObP '6f0'i
ObL '[EC'I
O&LP ,891
%LL '19
"St.8 ' I E
=
g ~uno.rj
' ~ elep c~oruotolpaz!tu!u!nt Bu!sn d i ! ~ ! r o ia1 sporliau oo!tzln>les
=
=
=u
= I!
=
"
%b 'OPL '
8 6 '€LO'< r 'L
% 2 f '010'1 = U
%LS 'ZsZ F 7,
%gL '86 = I!
'05 F u
ObZ6 '9Z =
(269'1 =
7,)
sbobs
ib%a
(,Z69'1
% b 'Z91
%t. '6S1
%b (ZOZ
O,[L'PZ
Oh98 'P 1
0668 '6
%(I01 't.
%E 'OPS = 1'
D$< '658 = t,
%El '9~z.I = U
%flS ' 9 i f = 11
oh29 'PIC = u
%9L 'ZI I = I,
0 4 ~ s'St = u
'600'1 = !!
'SPZ'1 = U
% L I '1St.l = Lg
%LC 'OIZ = "
oh59 'g = u
~ $ 'LE
9 = tr
%95 '81 = 11
O h f l
,%b '951 =
%>P'LS I =
%P ,Z[]z =
L!
L'
1,
,'
(1
8'
1.
i
<.#
o ; , l ~ ' + ~ = r r <i
UhCS '81 = I,
IS
%001 '6 = , I
5
e 0 o l ' 9 = t,
i ,
LC
$1
"
1'
(YZZ =
11)
z21obs
azt
")
but
1~90s
uos!jed1no3
BE 'SIC
965 '99L
Y 6 '6+6
'ESl
5$15 :6p
=
7,
=
1,
=
CJ
=
=
11
1,
O h ~'
EtZ =
Oh0 :Z[ =
-
,,
tr
oh2 ' 6 z t
%E '9E9
8 9 'Z6L
O$CP ' t g l
%OL '25
%99 :6Z
O ~ E S-LI
=u
=
3
=
=
11
at
?!
11
=
"
=
$I
oh@ '891
=
11
II ;
7.ZIl
'8 [ L ' I
%St. '9IZ
%8L 'OL
=
=
t
tt
i l j
=
81
EO
S
"
5'0
=
II I
F>SY' ~
= 7, i 5.1
-d,6S
--
61 =
"
F.T
-
(O
.Lu~.
L I Z ~ I ZU,
? ~ . 2001
quotient values for all 18 methods ranged from 0.01 to 100.
The unique combination most predictive of toxicity to amphipods was SQGQ1. This combination of mixed sediment
quality guidelines for mean quotient calculation resulted in the
greatest association with the incidence of toxicity to amphipods, (89% for values >2.3 and 88% for values >1.5; Table
6). Other derivation methods yielded reduced associations
(30%-81%) with observed toxicity for values grearer than 1.5
(Table 6). The incidence of toxicity invasisbly was less than
10% in samples with lesser mean quotient values (e0.5).
As with the BPTCP database, there was a sipificvnt difference in the total number of SEDTOX samples for which
mean quotient values could be calculated using different che~nical combinations. Totals ranged between 946 and 1,668 and
represented 35 to 60% of the available data. The method that
used the greatest number of chemicals (24 for ERMQ2) in the
mean quotient calculations yielded the largest number of calculated SQGQ values (Table 6 j but also demonstrated some
of the lowest incidences of associated toxicity. Inclodin:: the
large number of individual PAHS in the ERMQ2 calculation
was conducive to use of the data set but did not improve
associations with acute amphipod toxicity. The SQGQ1, which
incorporated only nine chemicals (no individual PAHs) and
most accurately predicted toxicity, could be calculated for
1,435 samples, representing slightly more than one half of the
available data.
Most mean quotient calculation methods provided good
predicrive ability in the Biscayne Bay data (Table 6). Both
SQGQ6 and SQGQ.1 predicted equally well based on the number and percentage of samples that were toxic (100%) when
mean qi~otientvalues were greater tllan 1.0 and 1.5, respectively. The incidence of toxicity was greater than 70% tor
SQGQ6 and SQGQl when values were grearer than 0.5, while
values less than 0.5 were toxic less than 5% of the time.
To demonstrate general relationships between acute toxicity
to amphipods and SQGQ values, data irom all three sources
were combined into a single data set. Reduction of the combined data set of 3,584 samples to only those samples that
could be used in all 18 quotient calculntion methods resulted
in a commondata set of 1,692 samples. Calcniated mean quotient values for all 18 methods ranged from 0.002 to 100. Tile
SQGQ1 resulted in the greatest associarion with tlie incideiice
of toxicit), to amphipods (100% for values >2.3 and 87% lor
values >1.5). A low incidence of toxicity (4%) was observed
when SQGQl valoes were less than 0.1 (Table 6). Table 7
identifies the percentages of samples that were ecutely toxic
for 12 ranges of SQGQl values. Also expressed is the magnitude of tile toxic response (as mean % survival) over the
same SQGQl runges. The SQGQ1 values were highly corre!sled with the incidence (r2 0 . 9 0 1 , ~< 0.001) and magnitude
(r2 = 0.913, p < 0.002) of acute tosicity (Fig. 1).
-
DISCUSSION
Mean SQGQs have been used previously to compare toxicity test response and observed benthic community response
to concentrations of a mixture of 16 chemicals in coastal sediments from Califoriia, USA 1211. Similnrly, Long et Li6]
I.
used mean SQGQs to compare observed toxicity test rzsponsr
with n mixture of 24 chemicals in Lest sediments from a oational database (11 = 1.068). Both these studies found o pattern
oii,?creasinzincideiice of toxicity in sediments wit11 increasing
mean SQGQ values and identified critical levels above which
Sioiogicai effecls could be predicted. Each of these smdies
Table 7. Acute amphipod losiciry associated with SQGQl rangeSli:
common samples from combined NOAA/BPTCPIBiscayn
database (n = 1,692)"
Ba!
% of
SQWl
range
Range
averaee
-.
Average Average
No. of Sampies % Fur- no ~ h
sanlples
toxlc
iilval
exceedq
,
.
\_
~NOAA = National Oceanic ond Atmospheric Administiniiao.
BPTCp = Bay Proiect~onand Toxic Cleanup Program; SQGQ;
scdimcnt quality guideline quotienr: ERM = effects range lnedialli
used effects range median values to derive mean SQGQs,but
cbemicals used in the quotient calculation methods were n a ,
identical. Long et al. [6] used 13 individual SQGs to reptem
the individual polycyclic aromatic hydrocarbon con~pout~dri,~
the average, while Fairey ct al. [21] represented the PANsb,
using two individual SQGs, one each for low and high n+,
lecular weight PAW classes. Another difference is that Loog
et al. [6] represented the organochlori~~epesticides
using ERMr
for p:p'-DDE and total DDT while Fairey el al. 1211 rcp~. 01
sh
sented organochlorine pesticides using ERMs for total chioi:
ob
dane, dieldrin, endrin, and an organic carbon normalized vaiuc
c!c
for DDT [333. These seemingly minor diffzrences bad tht
lilt
potential. of yielding dramatically different resulls. For exme
ample, there is poor correlation (Pezson's correla~ioa,r 2 =
me
0.119) between mean SQGQ vnlues calculated by these two
for
mean SQGQ calculation methods in the present paper (ERNlQl
[11] and EPGi4Q2 161; Table 2) using chenlical data from CciInel
ifomia's Bay Protection and Toxic Cleanup Program (BPTCC
inti
n. = 605). Investigation of outliers that grouped far irom pie. ,
dicted values indicated that chemical n~ixtureswere dominated
mus
by DDT or PAHs in those particular California sanlples. As
CPC
demonsuated in Table 3, the predictive abilities of ERM SQGs
lonic
for DDT and most i~ldividualPAHs were low using BPTCP '
rega:
data. Inclusion of these individual SQGs in the mean SQGQ
This
calclllation method reduced the overail predictive ability of
nun11
the lnean ERMQ2 values. Using ERMQI, 86% of the mar1
sedili
contaminated samples (ERMQI > 1.5) were observed to
Illat (
toxic to amphipods (R. nbl.onius and E. esrr~al-ius),vihiie !he
and r
incidence of associated toxicity drops to 46% for ExMQ~ : envln
values > l . 5 (Table 5). Also of note is that the mean quoijeni
~nosi
calculation method requiring the greatest number o i ' c h e ~ ~ c ~and
~ a
(ERMQ?) resulted in the fewest samples available for me3'
wide
quotient calculation. Missing chemical analytes over mlllti~~' Choicc
years of the prognro limnitcd the utility of the ERMQ? vaiucr
huwe~
as indicators of chemical mixture concentration in this'd~"
edge i
set. These results denlonstrate how selection of indi\'iduai
Uuaii[!.
SGQs to be included in the mean SQGQ, in this case paruL
ltilllon
ularly with DDT and PAAs. can iniluence the outcome ofan?
chemii
subsequent analysis.
ination
In the data analysis reported here. the ability of mcd"
Surrog;
SQGQs to accunxely predict the probability of observln.'"'
c,I It,~.
'
associated biological eifecc i c .'----.I-.-.
r "Y
4
ill.
CJICU
lation of mean sediment quality guideline quotients
"es fur
0w
w
11o
-
-
ne Ba?
90% -
En~~iron.
Tosicoi. Clretn. 20, 2001
.
2283
___---..___----
0
0
*raze
P.
ERMs
ceded
o.001
'.Oj
i.3
6
r.3
.4
:.
7
6
,
5
f
.4
.2
-sg .
.-
!'at,on:
:dians.
;
10%-I
I
.
s. but
e no! '
eselll
;ds io
by :
,,
~aog
RMs
$pie$101-
alue
thf
9.
., =
LV/O
r QI
:a!-
CP;
,<zted
.?3
ti5
""
"3
i?
#
: :1
.
0
0.5
1
1.5
2
2.5
3
3.5
4
4.9
5
Average of SOGOl Ranger Fig. I. Amphipod response over the range of average sediment quality guideline quotients (SQGQI)values. Values were calculated from 1,692
samples for which all 18 chemical combinations could be calculated from the combined National Oceanic and Atmospheric Adrninistration/Ba)i
Pmtection and Toxic Cleanup Program (NOAABPTCP), Biscnyne Bay databases. Logaritl~micfit ecluarion lor percent of samples toxic versus
SQGQ iu" 0.2352 in(.r) + 0.6487 (r2 = 0.901,.p c: 0.001). Linear fir equation for avenge % survival versus SQGQ is y = -0.1774(5) +
0.8277 (r' = 0.913, p C: 0.001).
or'cheinicals and SQGs used in the calculation. This.relationsllip beconles clear from the wide range of predictive success
observed amot~gderivation methods presented here. It is also
clear that use of differeht data to assess mean SQGQ calculation influences the estimates of predictive ability for seditncnt toxicity. This demonstrates that calculation methods for
mean SQGQs must be refined and standardized using a method
for which predictive ability and interstudy comparaKility are
not comproroised. This point is critical to our future Use of
menn SQGQs and will directly impact their utility as assessment tools.
To refine calculation of mean SQGQs, a major assumplion
must be made that chemical analytes used in the mean SQGQ
calculation are indeed representative of, or surrogates of.. the
'oxicolo~icallysignificant chemical mixture in the samples
re%iirdlessof which cl~emicaiswere quantified in the analyses.
This is asimplistic approach because of the seemingly infinite
"Umber of chemical combinations present in held-collected
"bents. 111addition. each research or monitoring program
Illat generales sediment chemistry data has its own objectives
""esultant
analyte lists, based on economic and regional
environmentalconsiderations. Use of chemicals that occur
mostcommonly across many pograms will maximize data use
"'jd allow development of mean SQGQs more applicable to a
"Ide range of environmental conditions and objectives. The
ciiniceuf toxicologically representative chemicals is limited,
k'""el'er by cunent a~laiytica!ntethodolog~esand our knowlL
'ge of [hose chemicals in the iiteralure for which sedilnent
have been established. This operatiotlal linl? " a i i tguidelines
~
"!lion
mandates the assumptioil d ~ arbose
t
few published SQG
COC
"'cal analytes are represen~abveof antlvopogenic contam'nai'r~" in the sample or that tbey serve well as covaryin~
'""u~ules. I( is construcdve. though. because. as demonstrati.-ore ufi,:, ,-the3. number of chemicals in the calculs~ion,.,11-,.,-
''.
vation that only 4% of samples were loxic when toxicity was
not predicted (i.e.. mean SQGQl < 0.1; Table 6) suggests that
chemicals for which analyses were not perfornled, or for which
there are no individual SQGs, infrequently occur at toxisologically significant concentration5 in otherwise uncontaminated samples.
An implication of the mean SQGQ approach is that toxicologica! mechanism(s) of the representative chemicals are
additive. Independent experimental evidence has demonstrated
that acute toxicity to amphipods by some substances, such as
PAHs, are similar and additive when known combinations Of
chemicals are spiked into clean sediment [ 3 4 ] . The research
presented here does not provide direct experinleiltal evidence
of additivity. The results are, however, consistent with the
hypothesis because sediments are toxic more irequently when
chenucal concentrations simultaneously exceed increasing
lumbers of individual SQGs. Table 7 demonstrates that, as
increasing numbers of effects range median SQGS are exceeded, the incidence and magnitude of toxicity correspondingly increases. An additional observation in the cuuent research is that associations between toxicity and mean SQGQs
improve by removing chenlicols with less predictive SQGs
(Table 3) from the quotient calculations. Tlris suggests that
effectively represenring toxicological modes of action m-~ybe
more impo~cant!o SQGQ predictive accurac)) than simply including additional toxic chemicals to the representative chemical matrix.
Consideration of rile issues discussed here and the results
presented here leads ro the recommendation that a standard
meli~odfor combination of chemicals and respective SQGs
should be adopted for the calculation of menn SQGQs to allow
comparability becween regions and surveys. The current yeA-psearch has demonstrated !hat. for the mi-tlmde
2284
&triror~.Toricol. Citem.20. 2001
CPIC
Table 8. Chemicals and sedimenr quality guidelines uscd in chemical
combinntion SQGQI'
SQGQl
chemicals
Cadmlum
Copper
Silver
Lend
Zinc
Total chlordane
Dleidrin
Total PAHs
Total PCBs
a
SOG concentration
Guideline type
4.21 ~ 1 dry
% wt
270 pglg dry wt
1.77 pglg dry wt
113.18 ILS/E dry wt
410 pBlg-d;y wt
6 nplg dry wt
8 ngig dry wt
1,800 pglp OC
100 nglg dry wt
PEL [I71
ERM [I61
PEL [I71
PEL [I71
Em1 [I61
ERM [151
ERM [I61
Consensus [?O]
Consensus [35]
SQGQ = sediment quality guideline quotients; PEL = probable
effect Leveb. ERM = effect range meC~an:PAH = polycyciic aromatic hydrocarbon; PCB = polychlorinated biphenyl; OC = organic
carbon.
parating a number of theoretical and empirical sediment quality guidelines (Table 8) from a variety of reliable sources
[16,17,20,35]. As evaluated with the criteria for identification
of optimal predictive ability, the incidence of toxicity incrensed
steadily and average survival decreased steadily as mean
SQGQl values increased. T11e incidence of toxicity was less
than 5% and average survival was over 90% when mean
SQGQl values were less than 0.1. The incidence of toxicity
was greater than 90% and average survival was less than 50%
when mean SQGQl values weregreater than 2.3. The SQGQl
calculation method incorporates SQGs for nine of the most
commonly measured chemicals and best utilizes availabie data
in the tested data sets. Until a more predictive method has
been demonstrated, SQGQi is recommended to serve as the
standard for calculation of mean SQGQs when comparing with
acute toxicity to amphipods. SQGQ1 is calculated as
X ([dieldtin]/S)([total PAH,,]/lSOO)([total
PCBI/~OO))
+ 9)
An additional recommendation conceriis sediments where
unique chemical conditions prevail or regional anthropogenic
activities w m a u t special consideration. In these situations,
investigators might adjust the mean quotient derivation method
to better fit local environmental or management concerns. A
unique chemical example would be in a large agricultural watershed. Where endrin has been found to be a primary chemical
of concern, managers may wish to incorporate the U.S ELIvironmentcl Protection Agency's EqP-based sediment quality
guideline for endrill into their mean quotient derivation method. This flexibility will ensure a major component in the watershed's chemical signature is not ignored when investigating
the relarionship between chemical mixtures and biological effects. Wit11 this type of specific objective. the derivation method that best demonstrates associations between chemicals of
mterest and biological effects will be the most effective tool
fur utilizin$ PI-ohabilitiesof tosicicy and focusin., local causality studies. It is important, however. that investigators report
differences in the mean SQGQ calcuiation methods. We reconlnlend use of the SQGQl method reported here for providin!
.
.
,
method for addressing any unique local chemical mixturesu,
specific management objectives. The ability to exercise flex,
ibility with derivation methods shouldencourage and enhan,, ,
the effective use of mean SQGQ values.
As with other empirical approaches to evaluating sedi,na,
contamination using sediment quality guidelines, iimitatialls
in the use of the mean SQGQs must be sch1owledged,
SQGQl approach is meant to serve as a central tendencyin,
dicator (i.e., as means of multiple individual quotients) ofcon,
tamination for a complex sediment matrix. The disadvnnta
of central tendency indicators is that they minimize the EP
tential for impact from any olle component. Therefore, w,,cll
performing a sediment toxicologic evaluation, it is prudellt
consider chemical exposure on an individual chemica\ ha$,;,
in addition to the chemical matrix basis described here. con.
sideration of individual SQGs may help in situations w]le,l
exposure to single chemicals is poorly represented by overall
contamination within a sample. The poitlt to emphasize is thal
mean SQGQs should not be used as the sole indicator of sed. t
iment contamination. They should be used as additio~l~l
toolS
in our efforts to better understand the relationships betweeti
chemical exposure and biological response. Use of me;m I
SQGQl should fit well in the conceptual framework of thr
sediment quality triad that emphasizes a weight-of-evidetlce
approach to sediment quality assessment [36].
The current research has focused on n~anipulationof mul.
tiple chemical constituents to illvestigate their relationshipr
with a single biological response. Critical values selected for :
classifying toxic bioloiical responses are a major factor in the
results presented here, and it should be noted that the critical
values used as biological benchmarks for comparisotl over
selected SQGQ ranges are themseives currently subject to illvestigatioc New methods for determining toxicity u e being
examined [30,37,3Sl wlule others are reviewing current n~eth.
ods with more extensive data sets than have previously been
assembled [1S]. Results may influence the selection of critical
values, such as r test and minimum significant difference determinates used in this study. As methods for deternlinalioli
of toxicity are revised. the need to revisit r e i ~ t i o r ~ s hbetweell
i~s
I
toxicity and chemistry also may follow.
The current research has focused on acute toxicity of sedinlent to marine amphipods as the sole measure of biological ,
response. There is a need to further this research and ae:ive
optimal chemical mixtures for correlating with other test SPc.
cies and response parameters. Long et al. [6].1eponed that the
incidence of toxic samples increased marlcedly when data froin
sublethal tests were added to those from the acute amphipoi'
bioassays. Evaluation of mean SQGQs using iarge databases
that include bioassays with reproductive, developmnental, and
chronic endpoints are needed to further validate use of meall
SQGQs. Sitbsequent research should continue to expand "Ivestigation of impacts of multiple contamillants to me as^'^^"
in other environments and to measures of individual and co!]'.
munity response in the fie!d. Work with freshwater species1:
being reported [32], and relationships similar to those reportcu
here are emerging between toxicity to several freshw3ter te:
species and mean SQGQs. Current efforts by Nyland
[ l i l , Lowe and Tl~ompson[XI], and Fairey ct al. (21)inciudC
investigatioils of marine benthic community response andllavz
demonstrated increased probubiiities of community irnp~c[j"
mean SQGQ valiies increase. These efforts 2nd the cu""'!
"'
~vio
1o0l
:lul,
lion1
;,,Id
Disc:
[oili
I. 1
I
I
'
?. I
3. !
I
4.
(
I
5, 1
I
"
6. 1
i
7. 1
i
8.
(
I
0
9. i
C
f<
I/
v
10. t
P
0
Y
II. P
S
I?.
c:
L
c:
13.
c
>I,
01
14.
ti
bt
ui
2:
l j L,
0,
Ti
N
1
1
: b I,.
9,
1,:
!?.
s;
i(
ie.
ill
I
ic
:I'
0
p,,lculationof mean sediment quality guideline quotients
E,tt?iro~~.
Toxicol.
Chem. 20. 2001
2285
.a,.
,,,idence supporting the
use
of
mean SQGQs as interpretive
tools.
4cltno,uicdgcmenr-Funding support for the current research was
Jitnen,
ratiot~~
d. The
lcy in.
>f con'zntage
.he po.
, when
dent to
11 basis
?. Con; wilril
o\'er2il
:is tlta!
of sed.
31 tools
between
i mean
oi the
vidcnce
of mulonships
cted for
,r in tl~e
crilical
on over
i t to in:e bei~~g
nt metlt;ly beet1
i ci:ticai
:txe dci~inatioll
betweso
Caiifornia's State Water Resources Control Board Bay Protection
Program. Data from the SEDTOX database and
B,,d~ ~ x Cleanup
i c
Biscayne Bay were graciously supplied by NOAA through Jay Field
pr,jee of Response and Restoration).
REFERENCES
1 , ~ r GE
n Jones-Lee A. 1996. Can chemically based sedimentquaiitg criteria be used as reliable screening tools for water quaiity
ilnoacts7 Learned discourses: Timely scientific opinions. SETAC
~ c ; v s 16:14-15.
?, ~ o n gER. 1996. Numerical sediment quaiity guidelines can be used as reliable interpretive tools. Learned discourses: Timely scientific opinions. SETAC News 16:IS-19. 3 sampson JR, Pastoral: RA, Ginn TC. 1996. ER-L and ER-M values should not be used to assess contaminated sediments. Learned discourses: Timely scientific opinions. SETAC News 16: 29-3 1. 4, WConner TP. 1999. Sediment qualily guidelines do not guide. I earned discourses: Timely scientific opinions. SETACNews 19: 28-3 5 , ~~~g
ER, MacDonald DD. 1999. Sediment guidelines do guide those willing to see-A reply to Dr. Tom O'Conner. Learned discourses: Timely scientific opinions. SETAC News 19:21-22. 6. Long ER, Ficld LJ, MacDanald DD. 1998. Predicting toxicity in lnaritle sediments with numerical sedimenl quality guidelines. E~>.,,ir.onToricoi Citenz i7:714-727. 7. Long ER. MacDonaId DD. Cubbage IC, Ingersoil CG. 1998. Predicting the toxicity of sediment-associated trace metals with simultaneously extracted trace metal: Acid volatile sulfide can-
centrations and dry weight-normalized concentrations: A critical compunson. Eitviroa Toxicol Clrcnt 17:972-974. 8 O'Conner TP, Dnskaiakis KD, Hyland JL, Paol IF, Summers JIC. 1998. Comparisons of sediment toxicity with predictions based on chelnical guidelines. Environ Tox'icol Chcm 17:468-471. 9 U.S. Er.vironmentai Protecrion Agency. 1993. Technical basis for cstablisi~ingsediment quality criteria for nonionic contaminants for the proteclion of benthic organisms by using eqt~ilibriumpar-
tilioning. EPA 822-R-93.011. Officc of Science and Technoiogy, '~Yashingron,DC. 10. Di 'r0vol.o DM, Mahoney ID, Hnnsen DJ. Scott ICJ, Carlson AR,
Anldey AR. 1992. Acid-volatile sulfide predicts the acute toxicity
of cad~niumand nickel in sediments. Environ Sci Teclznoi 26:
96-10],
1 : PTI
E~~vironmernai
Services. 1991. Pollutants of concern in Pugel
Sound. EPA 910/9-4 1-003. U.S. Environmental Protection AgenCY, Seattle. WA.
Lollg ER. MacDonald DD. 1998. Recommended use of empirically-derived sediment quality guidelines formarine and estuarine
n Risk Assess 4:1019-1039.
ccosYstcms. H ~ ~ s mEcol
Can'RS, Long ER, Windom HL, Chapmaan DC, Thursby G. SloGM.Woife DA. 1996. Sediment quality assessment studies
Of Tampa Bay. FL, USA. Envir-on Toxicol Clze,n 15:12lS-1231.
1 4 . H~landJL, Van Doiah RF, Snoots TR. 1999. Predicting stress in
bcnihic coinmunities of southeastern U S . estuaries in relation to
chenlical Contamination of sediments. Environ Toxicoi Cltcm 18:
2557-?jrjq,
IS
ER. Morgan LG. 1990. The potential for biologicai effects
ofsc[limellt-sorbed conlaminants tested in theNatiana1 Status and
Trends program, NOAA Technical Memorandum NOS OMA 52.
NaliOnalOceanic and Atmospheric Administration. Sealtle, WA.
USA. n
,,,
cu RS, Caldw FD. Long ER, Ingersoll CG.
Igu6. Derpeiopmcntand evaluation of sediment quality guidelines
18. I'U, Fioridn coastal waiers. Ecoroxiculog)' 5:253-178.
Envlronmentn~ Protection Agency 1995. Gi~idoncc.ioiol.Asi""iii~
Cliei~iieniConlaniinnar Doia ,for. l/se I,? Fish ild~,isor.~es,
I--Fish SilmplinS and Analysis. 2nd ed. EPA 833-R-95-007.
Orfice u i \~:lter, \ v a s i ~ i ~ > r ~DC.
on.
I ' MacDonnld DD,
19. National Oceanic and Atmospheric Administration. 1989. Standard analytical procedures of the NOAA Analytical Facility. 2nd
ed. NOAA Technical Men~orandumNMFS FINWC-92; 1985-86.
National Status and Trends Program, U.S. Department of Com' merce, Roclcville. MD.
20. Swartz RC. 1999. Consensus sediment quality guidelines for
polycyclic aromatic hydrocarbon mixtures. Eniliron Taricol
Chem 18:780-787.
21. Fairey R, et ui. 1996. Chemistry. toxicology, and benthic community conditions in sediments of the San Dicgo Bay Region.
CA State Water Resources Conlrol Board Finai Report. Sacmmento, CA. USA.
32. Anderson BS, el 91. 1997. Chemistry, toxicity and benthic coalmunity condirions in sediments of tile southern California bays
and estuaries. California State Water Resources Control Board,
Sacramento, CA. USA.
23. Downing I , et ai. 1998. Chemical and biological meapres of
sediment quality in the central coast region. CalifornivState Water
Resources Control Board, Sacramento, CA, USA.
24. Jacobi M. et ai. 1998. Chemical and bioioeicai measures of sed-
35. Phillips BM. et al. 1998. Sediment chemislry, toxicity, and benthic
community conditions in selected water bodies of the Santa Ann
region. California State Water Rcsoorces Control Board, Saciamento. CA. USA.
26. Anderson BS, et al. 1998. Sediment chemistry, toxicity, and benthic communitv conditions in seiected water bodies cf the Los
Angeles region. California Stale Water Resources Control Board,
Sacramento, CA, USA.
Hunt JW, et ai. 1998. Sediment quality and biological effects in
San Francisco Bay. Caiifarnia Stale Water Resources Control
Board, Sacramento, CA, USA.
Field LJ, MacDonald DD, Norton SB, Sever" CG, Ingersoll CG.
1999. Evaluating sediment chemistry and toxicity dirta using loeistic renression modeiinn. Envirori Toxicoi Chent 1S:1311-1322.
29. t o n g EK, et 31. 1999. ~ i ~ n i t u and
d e extent of chemicai cootarnination and toxicity in the sediments ofBiscayneBay and vicinity.
NOAA Technical, Memorandum NOS ORCA 141. National Oceanic and Atmospheric Administration, Silver Spcng. MD. USA.
30. Phillips BM, Hunt JW, Andcison BS. 2001. Statisticalsignificance
of sediment toxicity test results: Threshold values derived by the
detectable
significance approach. Ern,ir-oi?Toxicol Cltezn 20:371--373.
31. Thursby GB, Schieicat CE. 1993. Statistichi analysis o i lo-day
solid phase toxicity data for amphipods. Absrrncrs, 14th Annual
Meeting, Society of Environmental Toxicology and Chemistry,
Houston, TX. USA. p 99.
32. MacDondld DD, Ingersoll CG, Berger TA. 2000. Development
and evaluations of consensus-based sediment quality guidelines
for freshwater ecosystems. Arch Ensiron Corunat Toxicol 39:203i.
?X Swartz RC, Cole FA, Lamberson 10. Fernto SP, Schults DW,
DeBen WA, Lee U E,Ozretich RJ. 1994. Sediment toxicity, contamination and amphipod abundance at n DDT- and dicidrincontaminated site in San Francisco Bay. E!?.viro,i Toxicol Clieln
13:949-962.
34. Swartz RC. Kemp PE Schults DUT, Lamberson JO. 1988. Effects
of mixtures of sediment contaminants an the marine amphipod,
Rhepo.~ynius abrorxius. Environ Toxicoi Chenr 7:101?-1020.
35. MacDonald DD, DiPinto LM. Fields 1, Ingersoil CG; Long ER, Swartz RC. 2000. Development and evaiuatian of consensus-
based sediment effect concentrations for polychlorinated biphe- nyls. Environ Toxi$ol Cilent 19:1403-1413. 36. Chapman PM, et ai. 1997. General guidelines for using the sed- B L ~34368-372.
I
iment quality mad. Mor Poll~~r
37. Smith R. 2001. The use of random-model rolelance inter1i3ls in
environmental monitoring and regulation. J Agric Bioi L~tiair-oii
Star (in press).
38. Thursby GB, Heitshe J. Scott KJ. 1997. Revised approach to
toxicity test acceplabililg crileria using n statistical performance
assessment. Eji,jir.on Tosicai Chenl 16:133?-1339.
3 9 Lowe S. Thompson B. 1999. Results of the Xcnrhic Piloi Study.
1994-1 997, 11: Identifying benthic responses to conlnmiilation in
San Francisco Bay. San Francisco Estuary Regional Monirorino
Program lor Trace Substances. Technical Rcpon 39. San F~anc ~ s c oE s i u x y Insticute. Richmond. C.4.
APPENDIX
Summations used in sediment quality guideline quotient (SQGQ) chemical combinations: references are given for source of summal;n,
..",,
technique
Totd DDT [15.15] =
---.
([o~p'-~~~][p',p'-D~~l[o~p'-~~~]~!p'-~~~l[o~p'-DDTl~!p'~~~~l
2 ([cis-chiordme][rr11n~-~hlo1d~ne][~i~-nona~11101][t1-~~~~-nonachlol[oxycNordme])
Total PCB (191" = 2 ( [ P C B ~ I [ P C B ~ ~ ] ~ C B ~ ~ ] [ P C B I ~ ~ ~ ~ B ~ ~ ] [ P C B ~ ~ ] [ P C B ~ ~ ~ I [ P C B ~ ~ ~
Total clllordnne [I81 =
[PCB153][PCB170][PCB18O][PCB187][PCB195][PCB2061[PCB2091)2) Low molecular weight PAHs 1201 =
2 ([acenupthene][acenaphthyiene][antlir3~ene][flu0rene][napl1tha1ene][phenmthrenelj Hi& molecular weight PAHs [20] = ~ ( [ b e n z [ n l m t h i a c e n e ] ~ e n z o [ n l p y r e n e ] [ b ~ t b e n e ] ~ b e n z o [ k ] f l u o r a n r h e n e ] [ c h y r s ~ n e ] [fluoranthenel[pyrenell Total PAHs [20]
a
s
2 ([low molecular weigh1 PAHsl[higJ~molecular weight PAHsIj
-
This summarion is based on work of O'Conner 1191 for the National Oceanic aird Atmospheric Administration's Narionai Slatus and T ~ ~ ~
Program while developing comparability between individuai PCB congener summetions and historic Aroclor equivalents. PCB = polychlarinaled
bipllenyi; PAHs = polycyclic aromatic ihydrocnrbons; DDE = (1.1-dichloro-2,Z-bis[p-chloropbenyllethylene).