|
<br />~
<br />
<br />DeForest, Brix, and Adams
<br />
<br />Debate/ Commentary
<br />
<br />for the studies we reviewed are also provided in our summaries for compari-
<br />son.
<br />
<br />TISSUE-BASED THRESHOLDS
<br />The primary effect of selenium on (ish populations is manifestcd via mater- ni li
<br />naltransfer from the ovaries to the eggs (Gillespie and 8aumann. I U86; Woock ;: ;:
<br />... .. ..
<br />el ai., 1987; Schultz and Hermanutz, I U90; Hermanutz el ai., 1992; Coylc el ai" Q) 1 .~ 1
<br />~
<br />1993), with the dictalY pathway ueing the most important exposuJ'C route for c.. 0.
<br />e ~ .5 .5 .13
<br />juvenile and adult (ish (Sandhohn, Oksanen, and l'esonen, 1973; Benram and 0 co- u u ~
<br />.:.c '" > >
<br />Brooks, 1986; Woock el ai., 1987; 8ess<:r, Canfield, and La Point, J 993; Coyle lI) '" III 'fl 'fl 'ij
<br />'tl :::- :> :> :>
<br />elai., 1993). ~ 'ij ] ] e
<br />Given the imponance of the dietary exposure route for (ish, and of the OJ 0. 0. ~
<br />-;- ~ u
<br />.. p::
<br />ovaries in transferring selenium to cmuIYos. the most relevant tisslles for ~ 0.
<br />C'l 2
<br />developing residue-based guidelines are ovaries and food items. Whole body C'l 0
<br />- "" 1l
<br />~ II>
<br />and liver thresholds have also been recommcndcd, although some authors -0 u
<br />i c: .5
<br /> u U ~ '0 00
<br />(e.g., Skorupa el ai" 1996) have suggestcd that liver residues may havc limited E
<br /> E -0 , .;. M
<br />j .:; <-
<br />cnvironmcntal relevance. Accordingly, we focuscd our revicw on published 0
<br /> ~ 0
<br />laboratory, mcsocosm, and lield studies in which selenium effects (or no ;..., p::
<br />..c
<br />effects) can be correlated with selenium residucs in ovaries, whole body tissuc, 'tl
<br />Q)
<br />and diet. The thresholds recommended by Lemly (1993a) and Skorupa el al. '^
<br />0 -0
<br />(1996) for these tissues are shown in Table 1. 0- f! :a e f!
<br />0 '"
<br />~ E .!! :> :>
<br />In reviewing these studies, we established guidelines for critiquing the 0- ~ 'c ] ~
<br />~ u
<br />studies and evaluating the residuc-based loxicity data. First, all studics had lo ~ u > u .~
<br /> ,~ .2, ,~
<br />be based on environmentally J'clcvant and important exposure pathways, e,g" U '- 0 u
<br />i ~ 1LI :> 0 i :>
<br />dietary exposure to organo-selenium or maternal transfer to the embryo. .... ~ ,q ~
<br />'"
<br />c- o. .. Q. D.
<br />Studies based on other exposure routes, such as water only exposures, wcre u t: e u
<br /> p:: 0 p::
<br />b
<br />reviewed but not included in the final data analysis. Justification for this is ~ E ;:;l
<br />u
<br />provided in the analysis section of this paper. Second, we only considered ~ ...J
<br /> '"
<br /> u
<br />toxicological endpoints with clear relevance to population level effccts, such Q) "d u
<br />'tl c: .5
<br />as reproduction, survival, growth and teratogenesis. This is consistent with the .~ u
<br /> e u ~ -.r ....
<br /> E '"
<br />USEPA's approach for deriving water quality criteria (Stephan el ai., 1985). ~A 0 ':;
<br /> ~ 0
<br />Third, again following USEPA guidance, chronic values wcre calculated as thc 'a (0
<br />geometric mean of the no obselvablc ellt:ct concentration (NOEC) alld lowest Q)0'l
<br />v~
<br />obselVable effect concentration (LOEC). lI)~ >>
<br /> -0
<br />The use ofNOECs and LOECs in developing chemical guidelines, however, .... .8
<br />is not a rigorous approach becanse these values are highly dependent 011 the u ~ u
<br />Q) " '0 U
<br />~ ::I ~
<br />study design. The NOEC and LOEC do not provide any information on the j:: 0 is
<br />magnitude of the effect, only that the effecL is statistically significantly greater r-<
<br />than that obselVed in the control organisms. This, in turn, is dependent on the
<br />variability in the treatment and control responses aud the amount of replica-
<br />tion (Gelber el ai., 1985). To evaluate the studies further, we lit statistical
<br />models (probit,logit) to the toxicity data for the various tissues and endpoints.
<br />Although the level of an "acceptable" effect is arguable and variable depend-
<br />ing on the site and species evaluated, this approach allows thresholds to be
<br />developed for dilTerelll effect levels. We estimated EC.o, EC~o' and EC50 values
<br />
<br />I hllll, F.l'UI. Risk Assess, VIII. fl. No, 6, IUUU
<br />
<br />1191
<br />
|