API Publ 4751:2005 pdf download

API Publ 4751:2005 pdf download.Evaluation of Water Quality Translators for Mercury.
The ratios summarized above were developed based on a literature review performed during the late I 990s, Numerous studies published since that time contain extensive additional data for freshwater. cstuarine. and marinc systems. Selected studies containing relevant data are listed in Appendix A. While the newer data for freshwater systems appear generally consistent with the infOrmation compiled by EPA (20(X)a.b), a detailed comparison was not conducted for this report. The large volume of data now available merits consideration by EPA or others, to develop a better understanding of BAF and fraction dissolved (I) values in estuarine and marine systems, and to farther rcfine the currcnt understanding of these parameters in freshwater systems.
In addition to the ratios shown in Table I. the ratio of methylmerctiiy to total mercury in fish is important to the development of mercury translators. The draft EPA guidance indicates that essentially all of the niercu.y in edible fish tissue can be assumed to be present as methylmercury (EPA 2004a). This assumption is based on a landmark paper by Bloom (1992). which argued that earlier estimates of a lowei proportion of methylmercury in fish muscle were duc to analytical artifacts. Sources of artit’actual results were identified as (I) contamination of samples with inorganic mercury, especially prior to development of “ultra-clean” sample handling niethods: (2) incomplete homogenization of samples combined with rounding down of excessively high values to l(Mf’i.; and (3) incomplete recovery in mncthylmcrcury methods versus near-complete recovery in total mercury methods. Subsequent research has generally confirmed this finding, at least for higher trophic level (carnivorous) fish in waters relatively uncontaminated with inorganic mercury (Lasorsa and Allen-Gil 1995; Wagemann ci al. 1997). Where slightly lower percentages of methylmnercurs have been measured in fish muscle e.g.. Kannan et al. 1998; Neff 2002), it is difficult to refute a possible role of incomplete analylical recovery of methylmercury. However, the results of Baeyens ci a]. (2003) provide a clear counter example. Specifically. methylniercury comprised only 58% ol’ the total mercury in commercial fish captured from the Scheldt estuary in Belgium. although the same species captured from the Belgian coastal zone and the Greater North Sea contained primarily methylmercury as expected (approximately 90% or greatert. The authors speculate that the Scheldt fish may Iced at a lower truph.c level due to the presence of large amounts of untreated sewage (Baeyens et al. 2003), Despite this apparently unexpected finding, the uncertainty associated with an assumption of 100% methylmercury in edible fish tissue can generally he considered low relative to the other uncertainties associated with criteria translation between fish and water.
Finally, while mercury in fisti consumed by humans is essentially all melhylmercury, this is not necessarily the case for fish consumed by wildlife. In contrast to humans, wildlife consume whole fish and typically consume smaller (lower trophic level) fish, Methylniercury constitutes less than 100% of total mercury in whole fisli’and in the muscle tissue of some lower trophic level fish (Bloom 1992; EPA 2000a). Comparison of the human and wildlife values developed by EPA as part of the Great Lakes Initiative suggests that piscivorous wildlife may actually be more sensitive to mercury than humans (EPA 1995). If waler quality criteria arc developed specifically for the protection of wildlile. then the ratio between mcthylmercury and total mercury in whole fish would merit further evaluation.
On a cautionary note, Sonesten 2()O3a.L argues that multiple linear regression, the method used by Moore el a). (2003) and Hakanson i2000. is mathematically inappnpriale for assessing mercury hoaccumublion. Due to intercorrelatiun among environmental variables. multiple linear regression may overstate the arnoum of variation esplained and could potentially lead to incorrect identification of key cnvironmcnai factt affecting hinaccumulation. Sonesien (2()03a.h) advocates the use of partial least squares regression. hut this mcthod does not produce a predictive equation that could he used for purposes of mercury translator development.
With these caveats, regionally applicable models for lake systems may pmvide a viable option for mercury criteria translation purposes, if available for a particular lake of interest. In addition to the models described above, models are also asailahle for reservoirs (Braga e al. 2000). utilizing esplanatory variables such as area flooded, reservoir age. and the ratio of catchmcnt area and runoff (French etal. l99 un ci aL 1999: Tbcrriault and Schneider 199S). However, accurate models are not ye available for river systems. and it appears that mercury hioaccumulation modeling has not been attempted in estuarine or marine systems (Braga et al. 2(WX). Thus. hioaccumulaiion models for criteria translation in these system’. will need additional research before being useful for NPDES or TMDL regulations.