|
<br />DeForest, U.-lX, and Adams
<br />
<br />
<br />The use of a cnbic model for Iilling a curve lO biological rcsponsc dala is
<br />queslionablc, howcvcr, becausc il will prcdiu levels of cHecl grcalcl' lll<ln
<br />100%. Sigmoid models snch as probil and logil models thal asymplole al a
<br />100% response havc bcen gcnc....lly dcmollslralcd to appropriately dcscribe
<br />biological rcsponsc dara (Gelber el at" 1985). Acconlingly, we also lil probit
<br />and logit models lo the raw dala; bOlh appcared lo providc a good lit lO lhc
<br />dala (Figure I). As shown ill Figurc I, Ihc ECIO, EC~o' and EC50 1()J'lhc probil
<br />and 10giL models are 34, 47, and 7'1. and 37, 4!J, and 70 mg/kg dw, respcClively.
<br />The 15 Illg/kg whole body residue associaled with Icr,llugcnesis rcportcd hy
<br />Lemly (I ml3a) appcars lo be overly conscrvalive for this endpoinl givcn lhal
<br />cven the ECos from lhis sludy is '1.4 IlIg/kg dw using the probil model. In
<br />addilion, Lemly (1998) suggcsts thal if Ihc proponion of juveniles or adults
<br />wiLh lerala is Icss lhan '1.0%, Ihc allLicipalcd impacl on lish populaliolls is
<br />expecled lo be ncgligiblc. Cunsideriug this, wc would selccl ,III ECIO or 3'lmgl
<br />kg dw as a COllsclvalive eHecls 1I1I'cshold 1'01' Ihis study. Regardlcss, as dClllon-
<br />slraled later in this paper, leralogcncsis in juvcnilc alld adull /ish is a Icss
<br />sCllsilivc endpoinl lhall lalvalmorLalily.
<br />Lemly (1993c), Field-collcclcdjllvcnile hluegills were cxposcd in lhc lahora-
<br />lory for J8U days 10 a diclary SclcIIO-I.-mClhiollinc conccnlr,llion or 5,1 IIIgl
<br />kg dw and a walerborne J: I selcnalc:selcnile concelllralion of 4.8 ~g/L. In
<br />one group of lish, winter cOlldiLiolls wcre simulalcd by gradually rcducing thc
<br />water tcmperalure lo 10(; aud dccreasing Ihe I'holoperiod; in Ihe olhn group,
<br />the water tcmpcralllrc was mainlaillcd al '1.lloC. 111 the warlllwaler group,
<br />l110nalily (5,8%) was nol signi/icantly reduced relalivc lo Ihe wannwatcr
<br />conlrol fish (1.3%). In the coldwater group, mOl'lality was 33.8% cOlllpared to
<br />2.8% in Ihc coldwatcr conlrol. The whole body lissue conCl:nlralions iu the
<br />wannwatcr and coldwalcr groups werc li.ll ,ulll 7,~llIIg/l:.g dw, rcsl'cnively,
<br />.Ialvinen and Anklcy (I !J!J9) inlerpreted Ihe resulLs similarly, Whilc Ihesc dala
<br />provide interesting inlormatioll on Ihe pOlcnlial intel'active eflects 01' lelll-
<br />pcralure and selcuium, waler-qnalily crilcria guidelines arc nul cUlTenlly
<br />developed using studies with thermally stressed organisms (Slephan el ai.,
<br />1985). Consequently, wc did nol consider lhesc dala in our linal assessmenl.
<br />
<br />Slriped Bass (Morone saxatilis)
<br />
<br />Coughlan and Velle (1989). Slripcd bass wcre /Cd selenium-conlaminaled rcd
<br />shiners (Nolropis tUlrellsis) in the lahoralory for 7H days, The red shincrs werc
<br />collected from llelcws Lake and cOlllained all avcrage wholc hody sclcniulII
<br />concenlralion of 38.1i mg/kg dw. AfLer 78 days, all slriped bass had dicd ur
<br />were killcd because they were near dealh. The dielary concen Iralion of 3tl.1i mg/
<br />kg dw is comparable lo the diclary concenlralion of 35 mg/kg dw reponcd hy
<br />Lemly (I !J93a).
<br />
<br />ANALYSIS AND RECOMMENDATIONS
<br />
<br />Most of the sludy resuhs slnnlllarizcd abovc arc discussed in Icrllls of
<br />whethcr the rcsponses inlhe Ircatlllelll groups ,IrC signilicalllly dilJerenllllan
<br />
<br />Debate/Commentary
<br />
<br />lhe rcsponses in Ihe conlrol groups. We also cvalualed the concentralion-
<br />responsc dala using regression lIIodels whcu sullicicnt data were availablc,
<br />This approach allows dillercnl clrecl levels to bc cslimaled alld allows data
<br />frolll lIIuhiple slUdics wilh similar endpoints lo be pooled. , ,
<br />Ovwies. In many of the swdies in which advcrsc eflects were assoClaled wllh
<br />ovalI' residucs, the rangc of ovalY selenium conccnlralions measured was
<br />insuflicienl for dcvelopillg an adequale concelllralion.response rclatiouship.
<br />The Coyle el at, (1~1!J3) swdy is Ihe only one lor which a concentration.
<br />responsc rehllionship could be dcveloped based 011 dala from a single study,
<br />Thc ov;uy-hased chronic value from lhis sludy for lalvalmonality at day 5 was
<br />'1.6 mg/kg dw. Filling probit and logil models 10 the day 5 laIVal morlality dala
<br />(following Abhull's cOlTection2) resulLs in ECIO, EC~o' and EC~o values of21, 27,
<br />and :S3 IlIg/kg dw and '1. Ii , '1.9, and 34 mg/kg dw, respeclively (Figurc '1.). .Th~
<br />ECIIl and EC211 valucs Imlckel Ihc chronic value ('1.1i IIIg/kg dw). Of the IIldl-
<br />vidual s1udies, this sllldy scems to be the most appropriate for recommending
<br />an ovall'-bascd selcniulll guideliue allhis lime because iL was based on a wide
<br />range of exposure couccnlralions (j,e., a suflicielll number lo brackel Ihe
<br />cf/CClS thrcshold), it evalualcd cnvironmenlally rcalislic exposure palhways
<br />(i.e" organoselcniulII ill Ihe diel ali(I inorgauic selcnium in lhe waler). and il
<br />evaluated lhc crilical exposure palhway for lalvac (i.e" maternalLTansfcr Irom
<br />lhc (lVall' to thc egg). However, we did not wanl lo disregard lhe dala ~rolll
<br />olher sludies lhal evalualed lalvalmortalily frolll parents exposed to selcnllun,
<br />so we pooled all thc relevant data.
<br />The sludics pooled werc Ogle and Knighl (l9H!J) , lllyson el at, (1984, 1985
<br />a,b), Ilcrmalllllz el at, (I ~m'1.), and Coyle /il at, (1993). Using Abbott's corrcc-
<br />tion on the (;uval monalily dala, thcse data lcnd 1.0 lollow a characterislic
<br />sigmoid dislrilllllion whcn plolled (similar lo lhe (;oyle el at. data in Figure 2).
<br />!lolh thc pwhil. and logilmodels lil the raw dala quite well (Figure 3), The
<br />ECIIl, EC~lI' and ECslI values arc 17, 20, and 27 mg/kg dw based on the probil
<br />model, and I H, 21, and '1.7 mg/kg dw bascd on thc logit model. The prol)ll
<br />ECIIl is 70% grealer lhan the ovary guidcline of 10 mg/kg dw recommended
<br />hy Lemly (I !J!J3a) which, as discussed above, appears lo he derived from lhe
<br />Ilermauulz elai, (I 9~1'1.) slUdy hy associaling the lowesl measured ovary selc-
<br />niulII cOllcenlralion in pareul /ish with significanl obsclVed laIVal morlalilY.
<br />Thcsc dala suggesllhallhe OV.111' chronic valnc of 26 mg/kg dw from Coyle el
<br />ai. (I !1!1:1) lIIay bc nonconsclv,llive as the EC511 is of similar magniludc. Accord-
<br />illgly, wc recolI\lI\clllllhallhe ECIO or 17 mg!kg dw hc IIsed as lhc ovall'-bascd
<br />lhrcshold for hllval morlalilY,
<br />The slatislical regression approach allows dilferent levels of prolectioll lo
<br />be cvalualcd ,md docs nol lIudllly placc all the wcight 011 one sludy. A key
<br />ullcertainly in this approach is the applicabiliLy of lhese models lo coldwalcr
<br />lish (all relevalll ovary dala currenlly available arc for wannwaler species), As
<br />
<br />Corrcclcd % MlIrlalilY = (Ohsclved % Monality - Conlrol % Monalily)/(IOO-
<br />CIIJllwl % Munalil.Y)
<br />
<br />I hllll, E<:ol. Risk Assess. Vol. 5, No, Ii, 1999
<br />
<br />1211
<br />
|