United States        Office of Water        February 1985
           Environmental Protection    Regulations and Standards (WH-553) EPA-440/4-85-023
           Agency          Washington DC 20460
           Water
v>EPA     Cadmium Contamination
           of the Environment:
           An Assessment of
           Nationwide Risk

-------
                                    DISCLAIMER
This document has been  reviewed in accordance with the procedures of the Office of Water
Regulations and  Standards of the U.S. Environmental Protection Agency, and approved for
publication. Such approval does  not signify that the contents necessarily reflect the views and
policies  of the U.S. Environmental Protection  Agency. Nor does mention of trade names or
commercial products constitute endorsement  or recommendation for use.

-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing}
1. REPORT NO. ' 2 — 	
EPA-440/4-85-023
4. TITLE AND SUBTITLE
Cadmium Contamination of the Environment:
an Assessment of Nationwide Risks
7. AUTHORIS) 	
Charles G. Delos
9. PERFORMING ORGANIZATION NAME AND ADDRESS
12. SPONSORING AGENCY NAME AND ADDRESS
Monitoring and Data Support Division
Office of Water Regulations and Standards
U.S. Environmental Protection Agency
Washington, DC 20460
3. RECIPIENT'S ACCESSION NO.
5. REPORT DATE
February 1985
6. PERFORMING ORGANIZATION CODE
8. PERFORMING ORGANIZATION REPORT NO.
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
13. TYPE OF REPORT AND PERIOD COVERED
Final Report
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES " 	 —
  The incidence of cadmium  induced  harm  to  human  health  and aquatic life was evaluated
  and linked to pathways  of exposure  and means of disposal.   The  population's median
  exposure appears to be  12% of the kidney  toxicity threshold.   Statistical extrapola-
  tions suggest that  1-2 persons/million might exceed a toxic threshold among nonsmokers,
  and somewhat more among  smokers.   For the  potential of lung cancer, the median ambient
  inhalation exposure could  be  projected  to yield a  lifetime upper-bound risk of 4xlO~6.

  Data indicate that  tobacco smoking and food contribute most of the population's total
  cadmium burden,   and that  ambient  air  inhalation  and drinking water  contribute much
  less.  The cadmium content of  food  and tobacco  is believed  to be  related to the cad-
  mium content of  topsoil.  Most  of  the cadmium handled by man is likely to be disposed
  of by  landfill  burial.   Nevertheless, there  are some  pathways  for the  addition  of
  cadmium to cropland topsoil.  These include phosphate fertilizer,  sewage sludge land-
  spreading, emissions deposition,  and  irrigation  water.   Modeling  suggests a  very
  gradual increase in population  exposure due to these pathways.
  The extent of any cadmium induced impairment of aquatic life remains uncertain.
  source discharges  of cadmium are estimated to be decreasing, however.
Point
7.
1.
Cadmium
Exposure
Risk
Water pollution
Air pollution
Effluents
Waste disposal
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
Sludge disposal
Phosphates
Food contamination
Toxic diseases
Urologic diseases
Carcinogens
Population (statistics)
18. DISTRIBUIION STATEMENT
Release to oublic


b. IDENTIFIERS/OPEN ENDED TERMS
Pollutant pathways
Soil contamination
19 SECURITY CLASS /This Report,
Unclassified
20. SECURITY CLASS (This page)
Unclassified
c. COSATI Held/Group
13J
06T
06F
21 NO. OF PAGES
77
22. PRICE
$11.50
EPA Form 2220-1 (Rev. 4-77)   PREVIOUS ECITION is OBSOLETE

-------
                                        EPA-440/4-85-023
                                        February  1985
      CADMIUM CONTAMINATION OF THE ENVIRONMENT:

          AN ASSESSMENT OF NATIONWIDE  RISKS
                  Charles G. Delos
    Monitoring and Data Support Division  (WH-553)
      Office of Water Regulations and Standards
         U.S.  Environmental  Protection Agency
               Washington, D.C.  20460
      OFFICE  OF WATER REGULATIONS AND STANDARDS
         U.S.  ENVIRONMENTAL PROTECTION AGENCY
               WASHINGTON, D.C.  20460
230 South Dearborn Street
Chicago, Illinois 60604

-------
Envfronmenta, Protectlon

-------
                                   FOREWORD
     Effective regulatory action for toxic chemicals requires an understanding
of the human and ecological  risks associated with the manufacture, use, and
disposal  of such chemicals.   Assessment of risk requires scientific judgement
about the probability of harm to the environment resulting from measured or
predicted environmental  concentrations.  The risk assessment process integrates
data on harmful  effects  and  information on exposure.  The components of an
exposure assessment include  evaluations of the environmental sources, exposure
pathways, ambient levels, and exposed populations.

     This assessment was initiated as part of a program to determine the
environmental  risks associated with current use and disposal patterns for
65 chemicals or classes  of chemicals (expanded to 129 individual  "priority
pollutants") named in the 1977 Clean Water Act.  It includes an assessment
of risks to humans and aquatic life, and is intended to serve as a technical
basis for identifying unacceptable risks and developing the most appropiate
strategy for their mitigation.


                            Michael W.  Slimak, (former) Chief
                            Exposure Assessment Section
                            Water Quality Analysis Branch
                            Monitoring  and Data Support Division (WH-553)
                            Office of Water Regulations and Standards

-------
                               TABLE OF CONTENTS


     ACKNOWLEGMENTS 	 iv

     EXECUTIVE SUMMARY 	  v

     PROGRAMMATIC CONSIDERATIONS 	 1 x

1.0  INTRODUCTION 	  1

2.0  HUMAN EXPOSURE AND EFFECTS 	  3

     2.1  Health Effects 	  3

     2.2  Cumulative Exposure of Population 	  4
          2.2.1  Estimating the Incidence of Exceeding a Threshold ....  5
          2.2.2  Predictive Uncertainty 	  7
          2.2.3  Exposure Trends 	  8

     2.3  Exposure Routes 	  9

     2.4  Potential Cancer Risks 	 17

3.0  ENVIRONMENTAL SOURCES AND PATHWAYS 	 19

     3.1  Use and Environmental Release 	 19

     3.2  Pathways for Contamination of Food via Topsoil 	 25
          3.2.1  Phosphate Fertilizer 	 25
          3.2.2  Emissions Deposition 	 28
          3.2.3  Irrigation Water 	 28
          3.2.4  Sewage Sludge Landspreading 	 29

     3.3  Long Term Implications of Contaminating Topsoil 	 32
          3.3.1  Current Concentration in Topsoil 	 32
          3.3.2  Forecasting Changes in Soil Concentrations 	 32
          3.3.3  Influence of Topsoil Contamination on Human Exposure  . 36
          3.3.4  Predictive Uncertainties 	 41

     3.4  Other Pathways of Exposure 	 46
          3.4.1  Bioconcentration in Shellfish 	 46
          3.4.2  Landfilled Cadmium 	 47

4.0  ECOLOGICAL CONSIDERATIONS 	 49

     4.1  Aquatic Life Exposure and Effects 	 49

     4.2  Controlling Key Sources 	 52

     REFERENCES 	 57
                                    TM

-------
                               ACKNOWLEDGMENTS


     Numerous people contributed to and influenced this work, which was
performed intermittently over a six year period.  The author gratefully
acknowledges the program management, encouragement, and critical support
provided by a supervisory team comprised at various times of Michael Slimak,         '.
Michael Callahan, Alexander McBride, Donald Ehreth, Frederick Leutner, and           ,
Edmund Notzon.  Some of the information reported was provided by Rod Cam,
Robert Westin, Larry Davies, Bruno Maestri, Justine Alchowiak, and others            '
at Versar, Inc.  During the initiation of the project, Charles Cooper and
others at Arthur D. Little, Inc. contributed effort.  Some aspects of the            ]
probabilistic approach were initially inspired by Charles Cook of EPA.               I
Donald MacGregor of Environment Canada also influenced the thinking on
some elements.                                                                       i

     Substantial  help was provided by individuals who reviewed the first
draft of this document (December 1983):  James Ryan, Steven Weil, Paul Brown,        ,
Thomas Gleason, Ray Morrison,  Robert Kellam, Sandra Lee, Charles Spooner,            I
Charles Stephan,  Nelson Thomas, and members of the Cancer Assessment Group
and Exposure Assessment Group.   By reviewing the second draft (November 1984),
considerable additional help was provided by James Ryan and Donald MacGregor.        1
Such review does  not necessarily imply concurrence.  The shortcomings of this        <
assessment, errors in fact or  in judgement, are the sole responsibility of
the author.                                                                          I

-------
                              EXECUTIVE SUMMARY


Introduction

. Cadmium is a metal used primarily for corrosion resistant plating, for
  plastic and pigment formulation, and for batteries.  In addition, substantial
  quantities reach the environment as impurities in other materials, partic-
  ularly fossil fuels and phosphate fertilizers.

. While cadmium has been studied extensively for several  years, and numerous
  regulatory actions have already been taken, there are some remaining questions
  and regulatory issues.  This assessment has focussed on human, non-occupa-
  tional exposure, and aquatic life exposure.  The results suggest that in the
  long term, the most potentially bothersome problem may  be the possibility of
  gradually increasing dietary exposure due to accumulating contamination of
  cropland topsoil.
Human Health Concerns

  Exposure and Effects

. The human health effects of concern are kidney dysfunction and (for inhaled
  cadmium) probable carcinogenicity. Preventing kidney dysfunction has been
  the basis for existing EPA criteria and regulations.

. A detectable (but not necessarily clinically significant)  effect on kidney
  function is expected to occur when cumulative exposure has raised cadmium
  concentrations in the renal  cortex above approximately 200 ug/g.

. Current median exposure of the population appears to be 12% of the kidney
  effect threshold.  Extrapolations suggest that possibly 1-2 persons per
  million may exceed this threshold among nonsmokers and somewhat more among
  smokers.  The accuracy of such extrapolations is  not certain,  however.

. Some evidence suggests that  population  exposure levels may have been increas-
  ing at a rate of more than 1% per year  during this century.  Other evidence
  suggests that exposure may not have been increasing.  Whether  any increase
  is now occurring is not known.

. Assuming that inhaled cadmium is  carcinogenic, and using the EPA Cancer
  Assessment Group (CAG) evaluation of lung cancer  potency,  the  median inha-
  lation exposure to ambient urban  air would be projected to result in an
  upper bound risk of 4xlO-6.   Exposure to indoor air contaminated with
  cadmium from cigarette smoke would pose additional  risks.

. The CAG has found no compelling evidence for carcinogenicity of ingested
  cadmium.

-------
Key Pathways of Human Exposure

Tobacco smoking and food are the major exposure routes for the population.
Ambient air inhalation and drinking water are minor routes.  Median exposure
via food appears to be 50 fold higher than the combined median exposures via
ambient inhalation and drinking water.  Vegetables, potatoes, and grains are
the largest contributors of cadmium in the average diet.

Topsoil appears to be the medium potentially most sensitive to contamination
because it appears to link directly to the critical food and tobacco exposure
routes.  As topsoil is also difficult to decontaminate, accumulated contami-
nation of topsoil  merits concern.

Phosphate fertilizer constitutes the largest pathway to topsoil.   Emissions
deposition and irrigation water are of lesser but somewhat uncertain impor-
tance in this respect.  In terms of total  quantity of cadmium, the land-
spreading of sewage sludge is also of lesser importance; however, the hazards
of such landspreading have been a concern due to a potential  for  causing
very intense contamination of small areas.

Phosphate minerals mined to produce fertilizer naturally contain  elevated but
variable levels of cadmium.  Commonly used Eastern phosphates are much lower
in cadmium than the less commonly used Western phosphates.  Any increase in
the market share of Western phosphate would increase the size of  this large
but low intensity  pathway.

RCRA/CWA regulations controlling the landspreading of municipal  sewage sludge
are rather effective in reducing the risks associated with growing food on
such disposal  sites.  With these regulations in place, the most significant
overall effect of  pretreatment for important sources (such as metal  finishers)
may be to reduce the sludge disposal costs for those municipalities  that
want to landspread their sludge but are constrained by the cadmium limits
imposed by the regulation.  For municipalities where sludge is landfilled
(or possibly ocean dumped or incinerated)  rather than landspread, the benefits
of pretreatment are less.

The sale and give-away of sludge is not controlled under current  regulations,
but is of concern  particularly for home gardeners.  Contamination of equiv-
alent areas of commercial agricultural  plots and home garden  plots have
different results.  The former results in  low level exposure  for  large numbers
of people; the latter results in high level  exposure for a few people.

Although the direct inhalation of cadmium in ambient air contributes little
to total  exposure, the widespread low-intensity deposition of airborne cadmium
onto agricultural  soils may contribute some additional  exposure.   The major
sources of atmospheric cadmium appear to be coal  and oil  combustion  and
some types of metals smelting.  Zinc/cadmium smelting no longer appears to
be a major source.
                                  VI

-------
Irrigation water appears to be only a minor source of topsoil  contamination
overall  in this country, although reliable data on many waters are not
available.

Based on currently estimated rates of cadmium deposition onto  soils (from
phosphate fertilizer, sewage sludge, emissions deposition, and irrigation
water),  it has been theoretically estimated that population exposure may
increase perhaps 70% over a period of several centuries.  Such an increase
in exposure could be projected to increase the estimated number of persons
at risk  for cadmium induced kidney dysfunction from the current 1.5 persons
per million to 120 persons per million.  These projections involve consider-
able uncertainty, however.  Indeed, it is possible that no significant
increase may occur.


Other Pathways of Human Exposure

Recent data show drinking water to be an insignificant exposure route.  In
the rare cases where the 10 ug/L standard has been violated, it has generally
been attributed to corrosion of the distribution system rather than contam-
ination  of the raw water by point or nonpoint sources.

Exposure to waterborne cadmium also occurs by ingestion of shellfish, many of
which greatly bioconcentrate metals.  While shellfish are too  small a portion
of the average diet to significantly affect population exposure as a whole,
individuals who eat unusual amounts of shellfish may be exposed to 2-4 fold
more cadmium than average.

Cadmium  uses in 1981 were as follows:

   Metal Plating           34%
   Pigments                27%
   Batteries               16%
   Plastics                15%
   Other                    8%

The cadmium incorporated into end products is believed to have relatively
little exposure potential.  Only if such products are disposed of to the
5-10% of municipal refuse that is incinerated, then the cadmium escaping
emissions controls would be released to air.

Nearly all cadmium handled by human society is mined or extracted from rather
stable geological formations incidently with other materials (zinc and lead,
fossil fuels, and phosphate ores).  The bulk of cadmium handled by man is
thought  to be disposed of by landfill burial.  While this matrix is probably
not as isolated from the biosphere as that from which the metal  was originally
extracted, it appears to be substantially safer than the other major modes
of release (i.e., to air, water, or topsoil).  Nevertheless, while significant
exposure to cadmium (and several other toxic metals) via groundwater appears
to be rare, the long term behavior of landfilled cadmium is not known with
confidence.

-------
Ecological  Concerns

.  The effects of cadmium on aquatic life have been  a  concern  for EPA's  water
  program.   Nevertheless, it is  difficult to ascertain  confidently  how  wide-
  spread are problems of cadmium induced ecological  impairment.   The  newly
  revised criteria for protection of aquatic life have  the  same  general
  magnitude as the human health  criteria.  The freshwater criterion appears
  to be exceeded at most a few percent  of the time.

.  Best Available Technology (BAT) and Pretreatment  Standards  for Existing
  Sources (PSES) seem likely to  substantially reduce  the  point source discharge
  of cadmium from levels thought to be  discharged in  the  late 1970's.

.  Effects of cadmium on terrestrial  ecosystems have not been  a factor in
  regulatory efforts.  The overall  concern has been low.
                                     vni

-------
                        PROGRAMMATIC CONSIDERATIONS
Current Regulatory Status

     Several EPA programs have regulations affecting cadmium.  The most
significant of these include (Kayser et al. 1982):

     RCRA/CWA (Resource Conservation and Recovery Act; Clean Water Act)
     sludge landspreading rules for disposal sites.

     RCRA standards for designating certain wastes as hazardous, thereby
     requiring special precautions for landfill ing.

     RCRA regulations for generators and transporters of hazardous wastes;
     standards for storage and disposal.

     CWA (Clean Water Act) pretreatment and effluent standards for a number
     of industries, including metal finishing.

     CWA water quality criteria for protection of aquatic life and human
     health (advisory).

     CWA ocean dumping regulations.

     CWA control of discharged pollutants through NPDES permits.

     CAA (Clean Air Act) New Source Performance Standards and State Imple-
     mentation Plans controlling particulate emissions (thereby indirectly
     affecting cadmium).

     SDWA (Safe Drinking Water Act) maximum contaminant level for drinking
     water:  in ug/U

     SDWA underground injection control requirements.

     FIFRA (Federal Insecticide, Fungicide, and Rodenticide Act) rebuttable
     presumption against registration (RPAR) of cadmium pesticides.
Issues on Problem Assessment and Regulatory Strategy

     The single most important goal of a regulatory strategy for cadmium would
appear to be to prevent any substantial  increase in the accumulation of cadmium
in topsoil.  While other goals are also apparent, the agency's regulatory
actions to date have been compatible with each other.  Generally the safest.
means of disposal appears to be landfill burial.  Fulfilling the need to use
sewage sludge as a soil enhancing resource is thus contingent on continuing
actions to prevent serious contamination of sewage sludge.

-------
     Although a number of possible new regulatory initiatives could be iden-
tified, the most important regulatory questions appear to be the following:

  1) Should EPA regulate the sale and give-away of severely contaminated sewage
     sludge?  The existing RCRA/CWA landspreading regulations apply to disposal
     sites, not to marketing and distribution programs.  Developing additional
     CWA 405 regulations intended to control this potentially hazardous pathway
     continues to be the first priority of the EPA Sludge Task Force.

  2) Should EPA regulate cadmium under CAA Section 112?  This decision, called
     for by Section 122(a), has not yet been made.  Some factors affecting
     the decision include:  (a) Cadmium has been judged to be a lung carcinogen.
     (b) Widespread, low-level deposition of emissions onto cropland soils can
     be expected to increase dietary exposure slightly; nevertheless, imminent
     hazards may not be possible to identify,  (c) Many formerly important
     cadmium emission sources are now tightly controlled through limitations
     on particulate emissions; the potential for substantial  additional
     reductions has become more limited.

  3) Should EPA take any action to encourage the substitution of low cadmium
     Eastern phosphate in place of the small volume of high cadmium Western
     phosphate used for fertilizer?  A significant reduction  in amount of
     cadmium reaching topsoil  would result from such an action; however,
     demonstrating an immediate hazard by this pathway would  probably be
     dubitable.  Rather, the concern is long term, and may merit further
     scrutiny in future decades.

  4) Should EPA take any action to ban uses of cadmium in pigments, plastics
     batteries, or metal  plating (under TSCA), or in pesticides (under FIFRA)?
     While such actions have been considered for some time, an unambiguous
     need for them seems yet to be established.

A counterpoint to the above possibilities is to undertake no  new initiatives
to regulate cadmium at this time.  Many controls are either already in place
or else under development.   Consequently, while on-going work affecting
decisions on the first two issues (1 and  2 above) is proceeding, the Agency is
not undertaking regulatory work involving the latter two issues (3  and 4).

     In addressing the above regulatory issues,  the following analytical  issues
seem most important:

     In estimating the risks of kidney dysfunction,  only the  long term total
     exposure by all exposure  routes is of interest.   Assesments of single
     exposure routes, unless they provide information on the  general  magnitude
     of other exposure routes, tend to occlude the perspective needed to  reach
     an appropriate risk management decision.  On the other hand, for the risks
     of cadmium induced lung cancer, inhalation  exposure alone is relevant.

     For consistency, an interoffice consensus is needed on some issues:
     (a) Cadmium's renal  toxicity threshold (in  terms of kidney concentration
     and in terms  of daily  uptake)  is not completely resolved,   (b)  For

-------
     cross-media comparisons, the means of comparing exposure by
     different routes (i.e., absorption efficiencies)  is  not  uniformly  estab-
     lished, and (c) the means of assessing plant  uptake  (and resulting human
     exposure) of cadmium reaching topsoil  by different pathways is  uncertain.

Based on the work described herein, it appears that  cadmium contamination  of
the environment may remain under scrutiny over a long  period  of time.

-------
                                SECTION  1

                               INTRODUCTION
     The environmental  hazards of cadmium, a metal  being dispersed into the
biosphere through several  pathways, have been a concern for many years, particu-
lary after it was linked with the occurrence of Itai-Itai  disease in post-war
Japan.  In the U.S. it  appears that the margin of safety between the current
population exposure and the kidney effect threshold is not particularly large.
Furthermore, a possibility of carcinogenicity has also been a concern.

     The Office of Water Regulations and Standards  of the  Environmental Protec-
tion Agency began a multi-media assessment of cadmium in 1978 as part of a
program to evaluate the exposure and risk of the 129 priority pollutants.
Through this effort several reports (referenced herein) on selected aspects of
cadmium use and environmental dispersal were produced by consultants.  An
assessment of exposure  and risk, performed in-house in 1979, resulted in
recommendations for an  agency course of action.

     In autumn of 1983  the Deputy Administrator requested  that the Office  of
Water Regulations and Standards assess the problems of environmental cadmium,
and determine the needs for regulatory action and inter-office coordination.
This assignment provided an opportunity to update and expand the previous
assessment and prepare  this technical  report of findings.

     The purpose of this document is to summarize the key  information on cadmium
use, dispersal, exposure,  and risk.  The intent is  to quantify, to the  extent
possible, the effect of human activities on the incidence  of cadmium-induced
human disease and ecological  impairment.  The assessment is focussed primarily
on contamination of food,  ambient outdoor air, and  drinking water (as well  as
ambient surface waters); contamination of the workplace is not considered
here.  The presentation has been kept as brief as possible; no attempt  has
been made to reference  all relevant literature.

     The organization of the material  presented was selected in order to
expedite relating exposure to health risks, and to  focus sharply on the impor-
tant exposure pathways.  It may be noted that the critical  analysis of  data is
concentrated on cadmium releases to and dispersal  through  the environment, the
resulting exposure, and (given a particular toxic potency)  the resulting
incidence of harm.  By  contrast, the information an cadmium's toxicity  (and
carcinogenicity) is rather uncritically accepted from other sources.

     Section 2 describes the human health effects of concern, assesses  the
     current exposure of the population, and predicts the  resulting disease
     incidence.

     Section 3 presents cadmium uses,  estimates release to all  environmental
     media, and assesses the key pathways contributing to  human exposure.

     Section 4 discusses concerns about cadmium effects on  aquatic ecosystems.

-------
     Although cadmium has been one of the most intensely studied toxic  pollu-
tants, uncertainties in its assessment remain.  Many of the uncertainties  in
the levels of exposure and in the quantities discharged, emitted, or  disposed
of are caused by the difficulty of detecting cadmium at its usual trace concen-
trations.  It is common for cadmium to be undetected in a substantial portion
of samples.  Other uncertainties arise in extrapolating from the limited data
available.  Still  other uncertainties apply to cadmium's toxicity and especially
its carcinogenicity.  Thus, while this analysis has  attempted to be quantita-
tively definitive, the results and conclusions are not  indisputable.  Indeed,
substantial differences in opinion exist  on many important issues.

-------
                                 SECTION 2

                         HUMAN EXPOSURE AND EFFECTS
2.1  Human Health Effects

     Damage to the kidney's ability to reabsorb blood protein is the known
(non-carcinogenic) effect having the lowest exposure threshold.  Increasing
degrees of cadmium induced renal tubular dysfunction are manifested in
62-microglobulin proteinuria, general proteinuria, aminoaciduria, and gly-
cosuria, in order of increasing severity.  Effects on bone and mineral metabo-
lism have accompanied kidney damage in severe cases as found in the Itai-Itai
or "Ouch-Ouch" disease in Japan (EPA 1979, 1980).

     Elevated g2-|7licr°9^ODul'in excretion is not equivalent to clinically
significant proteinuria.  Without continued high exposure to cadmium, there is
little evidence of either a progession of severity of kidney dysfunction, or a
significant shortening of life expectancy.  Nevertheless, while some elevation
of S2-microglobulin excretion appears to be a relatively benign condition,
it is usually taken as the threshold health effect in setting ambient cadmium
criteria (EPA 1979, 1980).

     Emphysema is another clearly documented health effect, having resulted
from chronic occupational (inhalation) exposure to cadmium.  However, as its
threshold is higher than for kidney dysfunction, it is not used as a basis for
environmental standards.

     Inhaled cadmium also appears to be carcinogenic.  EPA's Carcinogen Assess-
ment Group (CAG) has suggested that cadmium be considered "probably carcinogenic
in humans," based on lung cancer rates observed in smelter workers and based
also on animal studies (EPA 1984c).  Carcinogenicity is generally assumed to
be a nonthreshold effect, with probability of occurrence proportional  to the
total cumulative dose.  For lung cancer, inhalation may be deemed the only
relevant exposure route.

     While several other health effects, for example, immunosuppressive and
other carcinogenic effects, have sometimes been ascribed to cadmium, the data
indicating such effects have been considered too weak or conflicting to support
regulatory standard setting.  Thus, the effects considered in this document
are kidney dysfunction and (for inhaled cadmium) lung cancer.

     Whether the effect to be prevented involves a threshold (kidney dysfunction)
or no threshold (cancer), the cumulative exposure to cadmium is of primary
interest.  Body burdens of cadmium (about one-half of which are concentrated
in the liver and kidneys) steadily increase from very low levels at birth to
maximum levels at age 40-50 years.  Thus, cadmium depuration is very slow; once
absorbed into the body, its half-life is estimated to be 18-38 years (EPA 1979,
1980).

-------
     Of the total quantity of cadmium brought into the lungs or digestive
tract, only a portion actually crosses membranes into the blood stream.  This
percentage, the absorption efficiency, varies with the type of exposure
typically as follows:

                     Inhalation            25%
                     Ingestion            5-6%

Such efficiencies may vary substantially from individual  to individual  and
from time to time within the same individual.  Some studies have found absorp-
tion of cadmium from cigarette smoke to be substantially higher than 25%.

     32-Microglobulin proteinuria may occur when the concentration in kidney
cortex reaches approximately 200-400 ug/g (wet weight), although individual
susceptibilities may fall  outside this range.  At the present time a kidney
concentration of 200 ug/g  is the most widely accepted estimate of the critical
threshold (Ryan et al.  1982).  This concentration is estimated to result from
a daily retention (absorption) rate of 10-15 ug/day over a 25-50 year period
(EPA 1979, 1980).  If 12 ug/day is taken as the absorbed dose that will produce
a kidney cadmium level  of  200 ug/g over a 25-50 year period, then at a  6%
absorption efficiency this corresponds to a gross ingestion of 200 ug/day,
a value sometimes cited as a threshold (Commission of the European Communities
1978).  Thus, the WHO (1976) recommended tolerable gross  intake from food,
57-71  ug/day, apparently incorporates an additional  margin of safety; the
appropriateness of this value has been questioned (Page et al. 1984).  Ryan  et
al. (1982) and Logan and Chaney (1984) review some of the difficulties  in
establishing the dose-response relationship.  Individual  variations in  protein-
uria susceptability and cadmium absorption and depuration confound the  estab-
lishment of a firm threshold.

     The Cancer Assessment Group (EPA 1984c) considers the carcinogenicity of
cadmium to vary depending  upon exposure route.  While the available data do
not indicate that ingested cadmium is carcinogenic,  there is some evidence,
provided primarily by studies of smelter workers and of rats,  to support clas-
sifying inhaled cadmium as being "probably carcinogenic in humans."  EPA (1984c)
has estimated the probable upper limit for the cancer potency  of inhaled cadmium
as  follows:   lifetime (70  year)  exposure to 1  ug/m3  in ambient air may  give
rise to risks as high as 2.3xlQ-3 for lung cancer.  This  estimate involves
considerable uncertainty;  other data, other assumptions,  or other extrapolation
procedures can yield different results.  The Cancer  Assessment Group considers
its procedures to provide  the best estimate for the  upper limit on cancer
potency; they do not consider it feasible to estimate either the most probable
potency or a lower limit for potency.
2.2  Cumulative Exposure of Population

     Long term total  exposure can  be  indicated  by  the  cadmium  concentration
in kidney cortex.   Autopsy studies in Dallas, involving  93  men of  greater than
30 years age,  revealed an approximately  log-normal  distribution  of exposures,

-------
 with  a  geometric mean  of  24 ug/g, as shown in Table 1  (raw data from Johnson
 et  al.  1978).   These men  had no  known occupational exposure to cadmium.
 Table  1:   Statistics  for Cadmium in Kidney Cortex  (Dallas autopsy study, wet
 weight, males  aged at least 30 years).
 Category     Number

 Nonsmokers    21

 Smokers       72

 Combined      93
                   Geometric              Geometric
Arithmetic        Mean (ug/g)              Dispersion
Mean (ug/g)  [95% Confidence Range]  [95% Confidence Range]
   17.2

   30.9

   27.8
15.0 [11.7-19.3]

27.9 [25.0-31.2]

24.2 [21.6-27.1]
1.74 [1.53-2.19]

1.60 [1.50-1.75]

1.73 [1.61-1.89]
The age groups 30-39, 40-49, 50-59, and greater than 60 years, did not differ
significantly and consequently were combined together here.  It is apparent,
however, that the smoking and nonsmoking groups differ substantially:  the
arithmetic mean concentration for nonsmokers was only 56% of that for smokers.
(The tabulated geometric dispersion, a geometric or multiplicative standard
deviation, is through its definition the ratio of approximately the 84-th
percentile concentration to the 50-th percentile concentration in a log-normal
distribution.)


     2.2.1  Estimating the Incidence of Exceeding a Kidney Effects Threshold

     Figure 1 displays the Dallas autopsy data.  The distribution for non-
smokers appears to be log-normal  (the number of runs of consecutive data points
above and below the straight line falls within the "runs test" 95% confidence
range (Remington and Schork 1970)).  In a log-normal  distribution the fraction,
Q, of the population exceeding a  particular concentration, x, is given by:
            Q(Z) =  (1//7F) /°°exp(-z2/2)  dz
                                                   (2-1)
where Z is the log-normal  deviate, or relative distance between the particular
concentration, x, and the geometric mean concentration, JL:
               Z = ln(x/Xg)/ln sg
                                                  (2-2)
where sg is the geometric dispersion (i.e.,  geometric standard deviation).
For this work Equation 2-1 was evaluated with the numerical  integration  function
of a pocket calculator (Hewlett-Packard 1984).   Alternatively, however,  the
relationship between Q and Z can be obtained from extended  tables  of  the normal
distribution, such as provided by Meyer (1975).

-------
                                                                CONCENTRATION  (ug/g)
                                         o
                                         •
                                         o
                                                          CT>   CO  O
                                                           en
                                                           o
oo  o
o  o
                             cr>

                             70
                          -n o
                          o>
                          1-3:
                          m
                          coo
                          try o
                          m m
                          m o
cr>
m
70
o
                          3CO
                          O25
                            o
                            o
                            m
                            x
                            73
                            m
                            JO
                            o



                            a
                            »-H
                            CO
                            CO
                                         en
                                         ro
                                         o
                                         en
                                         o
     oo
     o
                                         en
     <£»
     10
                                         vo
                                         •

                                         vo
                                                           \    I   I
                                                                             (£>
                                                                             UD

                                                                             WD
                                                                                                                 vo
                                                                                                                 to
                                                                             IO
                                                                             en
                                                                             oo
                                                                             o
                                                                            en
                                                                            o
             ro
             O
                                                                                                                en
                                                    I      i    I   I
                                                           I    I   l\
             O
             •

             o

-------
     Mathematically extrapolating the log-normal  nonsmoker distribution upward
in this manner, it would be estimated that 1.5 persons per million (in this
age group) exceed a 200 ug/g kidney effects threshold.  The validity of such
an extrapolation of course hinges on the assumption that log-normality holds
in the extreme upper tail  of the distribution.  Thus, it involves considerable
uncertainty, like cancer risk extrapolation.  Unlike cancer extrapolations,
however, it was not intended here to increase safety by overestimating the risk.

     Unlike the distribution for nonsmokers, the Figure 1 distribution for
smokers does not appear to be truly log-normal.   The lack of log-normality
may be an artifact of the way Johnson et al (1978) defined smoking/nonsmoking
status in the autopsy study.  Perhaps more likely, however, is an inherent
lack of log-normality in the distribution of exposure among the smoking
population.  If it were assumed that the smoker  distribution were also log-
normal, extrapolation would indicate that 14 persons per million (in this
group) would exceed the threshold.

     (Assuming that cadmium were also carcinogenic, the potential cancer risks
cannot be estimated solely from data on kidney concentrations.  Additional
consideration of cadmium intake is required and  will be taken up later.)


     2.2.2  Predictive Uncertainty

     The results of the Dallas autopsy study appear consistent with other
studies (presented by Ryan et al. 1982).  While  the samples were not randomly
selected from the nation's population, a clear justification for expecting
it to differ significantly from the rest of the  nation was not apparent.

     The 95% confidence limits for the geometric  mean and geometric dispersion
were presented in Table 1.  The confidence interval for the geometric dispersion
of the nonsmoker distribution, because it is based on only 21 samples, is
rather large and may overstate the true uncertainty in this parameter.  An
independent study of kidney cadmium, by Indraprasit et al. (1974), also appears
to indicate values for the geometric dispersion  the range 1.7-1.8.  To produce
these values the arithmetic mean, *a, and standard deviation, sa, reported
by Indraprasit et al. (1974), were converted to  a geometric dispersion, Sg,
by the expression:
                    sg = exp /ln(l + (sa/xa)^                         (2-3)

Ellis et al.  (1979), measuring  the total  quantity of cadmium in  kidney,  found
a geometric dispersion of 1.7 for smokers and 2.0 for nonsmokers.   The  vari-
ability they found might be expected to be somewhat higher since it involves
the variabilities of both kidney size and cadmium concentration.  Overall,  the
results of these other studies, and the weight of evidence from  the additional
72 samples of smokers taken by  Johnson et al. (1978), suggests that the  confi-
dence interval  tabulated for the combined smokers and nonsmokers,  approximately
1.6-1.9, might  better represent the uncertainty in the geometric dispersion.

-------
     The uncertainty in estimating the number of persons exceeding a 200 ug/g
threshold is quite sensitive to the uncertainty in the geometric dispersion.
If the confidence interval  for this parameter were taken to be 1.6-1.9,  and
the mean taken to be 15.0 ug/g, then the corresponding range for the number of
persons exceeding the threshold would be 0.018-27.2 persons per million.

     The predictions are less sensitive to the uncertainty in the geometric
mean.  The confidence interval for the nonsmoker mean, 11.7-19.3 ug/g,  when
coupled with the geometric  dispersion of 1.74, corresponds to a range of
0.15-12.1 persons per million exceeding the threshold.

     The incidence of kidney dysfunction is quite sensitive to the concentration
threshold assumed.  If the  threshold were increased from 200 ug/g to 300 ug/g,
the number of persons exceeding the threshold would drop from 1.5 persons per
million to 0.032 persons per million, nearly a 50 fold decrease.

     In any case it must be emphasized that the projected incidence assumes a
log-normal distribution of  kidney cadmium levels.  If log-normality does not
hold in the extreme upper tail of the distribution, the true incidence  may be
quite different.  The validity of applying log-normality to the extreme  upper
tail has not been established.


     2.2.3  Exposure Trends

     Relative to other toxic metals, relatively little margin of safety  exists
between typical exposures and the threshold effect level.  On the other  hand,
based on the extrapolation  from the Dallas autopsy study, cadmium induced
32-microglobulin proteinuria appears to be very uncommon in the general  popu-
lation.

     Perhaps a more appropriate concern is whether exposure may be gradually
increasing.  Blinder and Kjellstrom (1977) (and Kjellstrom 1Q79) report  that
cadmium levels in preserved kidney specimens collected between 1880-1899 are
nearly four fold lower than levels for nonsmokers who died in 1974.  This
represents an annual compound rate of increase of 1.6 %/yr.  Interestingly,
Kjellstrom et al. (1975) report a corresponding rate of increase, 1.4 *,/yr,
in cadmium concentrations in a series of fall wheat specimens collected  between
1920-1970.  The rate of increase estimated for cadmium in spring wheat  specimens
was less than that for fall wheat and was not statistically significant.
Kjellstrom et al. (1975) and Purves (1977) discuss reported cadmium increases
in natural vegetation over  many years, apparently due to emissions deposition.
MacGregor (1981) has reviewed trends in ambient levels; the general tendencies
appeared to be increases in some cases and plateaus or stable levels in  other
cases.

     Despite the apparent agreement between the rate of increase in cadmium
levels for fall wheat (Kjellstrom et al. 1975) and for human kidney cortex
(Elinder and Kjellstom 1977), the existence of and magnitude of any long term
increase in exposure is uncertain.  Orasch (1983) measured the cadmium  content
of preserved liver and kidney specimens collected between 1897-1939 and  compared

-------
them to autopsy samples taken in 1980.  He found a 47 fold increase in kidney
cortex cadmium but a negligible increase in liver cadmium.  Such results are
difficult to reconcile; while the levels in both kidney and liver are considered
to be good indicators of cumulative exposure, levels in liver may be the better
indicator (Cherry 1981).  Overall, it might be concluded that there is some
evidence for a significant long term increase in cadmium exposure; however,
such evidence may result from artifacts of sampling.  Consequently, it is also
possible that no significant increase in exposure has been occurring.  In any
case, it is not known whether any such increase is now occurring.

     In linear nonthreshold cancer projections, small percentage increases
in (arithmetic) mean exposure produce equally small  percentage increases in
population risks.  For threshold effects, on the other hand, small percentage
increases in mean exposure may more sharply increase the population risks.  To
illustrate, a 1 %/yr increase in mean exposure to cadmium would yield a 22%
increase in mean kidney concentrations, when compounded over 20 years.  Pro-
jecting from the nonsmoker distribution in Figure 1, a 22% increase in geometric
mean (with no change in geometric dispersion) would  result in more than a
5 fold increase in the number of persons exceeding the 200 ug/g threshold
(i.e., raising it from 1.5 persons per million to about 8 persons per million).

     Increasing the variability of exposure would even more sharply increase
the risks.  For example, increasing the geometric dispersion of the nonsmoker
distribution from 1.74 to 2.0 (with no change in the geometric mean)  would
increase the incidence from 1.5 persons per million  to 93 persons/million,
assuming a log-normal distribution.
2.3  Human Exposure Routes

     The concentrations of cadmium associated with airborne particulates  have
been measured at numerous urban and rural  stations belonging to National  Air
Surveillance Networks.  Detectable concentrations occur far more frequently  at
urban than at rural sites.  Figure 2a illustrates the overall  distribution  of
urban concentration statistics provided by Rhodes et al.  (1979) for all
quarterly composite samples collected in the period 1970-1976.   Assuming  a
log-normal distribution, the geometric mean might be estimated  to be perhaps
1.6 ng/m3.

     The variability (slope) shown in Figure 2a represents  the  variability  among
quarterly composite samples.  Consequently, it has components  of both temporal
and spatial  variability.  Since only the spatial  variability is of interest  in
the assessment of long term exposure, the  distribution may  somewhat overestimate
the true variability of long term ambient  inhalation exposure.

     The arithmetic mean, x , of a log-normal  distribution  can  be estimated
from the geometric mean, Xn, and geometric dispersion, sn,  as  follows
(Meyer 1975):             y                             9

                   *a = exp [(In xg) + 0.5(ln sg)2]                   (2-3)

-------
                                                                 CONCENTRATION   (ng/m3)
o
                         o
                         rt-
                         fD
                         CO
                         o
                         fD
O-


O
0>
rt-
fD
-*
O)
3
UD
m

o
-h

O.
fD
rt-
fD
O
rt-
ro
a.

o
o

o
a>
3
rt-

Oi
rt-

O
3
to
        C?

        70
        rn

        ro
                  o
                  *
                  o
                                                                                                           o
                                                                                                           o
                              JO
                              |o

                              O -H
                              CO J»
                              m o
                                 70
                              m i— i
                              70 70
                              OT m
t>
m
o
m


i
m
CO Z
i—I O
—I -<
m
co a
V  H-<
   CO

«0 TO

O 03
 i  cr
                              *^j o
                              en 2=
                                 o
                                 PO
                                           U1
                                           ro
                                           0
                  tn
                  o
                  oo
                  o
                                           IO
                                           (71
                                           vo
                                           10
                  to
                  to

                  to
                                           
-------
Using this relationship an arithmetic mean of 3.5 ng/m-3 was estimated for the
air data.

     Concentrations of cadmium in drinking water almost never exceed the stan-
dard of 10 ug/1.  A 1969 nationwide survey showed 63% exceeding 1 ug/1 and
0.15% exceeding 10 ug/1 (Battelle 1977).  While this frequency distribution is
plotted in Figure 2b, it is important to note that the measurements were gener-
ated using crude flame atomic absorption sgectroscopy procedures, now considered
to be unreliable due to their strong overestimating bias when applied to cadmium
at its usual  ambient levels.

     A nationwide survey using a sensitive graphite furnace AAS method has
apparently not been done in this country; however, such work has been done in
Canada.  As the complete data sets could not be obtained, rudimentary summaries
of these data (Meranger et al. 1981) were used to construct the estimated dis-
tribution.  As shown in Figure 2b, the constructed distribution has a geometric
mean of 0.04 ug/L, a geometric dispersion of 2, and an arithmetic mean of 0.05
ug/L.  To compensate for the uncertainty in portraying this distribution,
a high bias was intentionally incorporated.  It is estimated that the mean of
this distribution over-estimates the true Canadian mean by a factor of two,
based on the discussion of HWC and EC (1983).  In the absence of reliable data
for the U.S., the distribution constructed from the Canadian data will be
assumed to provide the best estimate for U.S. nationwide exposure via drinking
water.  The variability shown applies to grab samples.

     Dietary intake of cadmium may be estimated from either food or fecal
analyses.  Estimates of cadmium gross intake based on measurements of concen-
trations in foodstuffs coupled with diet information (FDA data, as analyzed
and reported by Yost et al. 1978) result in a log-normal distribution of
cadmium intake with geometric mean of 12 ug/day and geometric dispersion of
4.6.  Estimates of cadmium gross intake based on measurements of fecal concen-
trations in 477 persons in Chicago and Dallas (as reported by Kowal  et al.
1979) indicate an approximately log-normal  distribution having a geometric
mean of 11.4 ug/day and geometric dispersion of approximately 1.9-2.1.  These
are plotted in Figure 2c.   Both studies provide a population distribution of
cadmium intake on a single day.  Yost's analysis, however, applies to a teenage
male diet.  Kowal's fecal  data includes both sexes and all age groups from
0-70 years, but its variability applies primarily to the cadmium concentration,
and only partially to the  fecal volume.

     Although the geometric mean intakes obtained from both the food/diet and
fecal methods are virtually identical, the much greater variance obtained
using food data results in a significant difference in arithmetic means between
the two.  Yost's analysis  of food data indicated an arithmetic mean  of 39 ug/day
(which is identical  to the mean reported by Pahren et al.  (1979) for 7 years
of FDA data).  The fecal data has an arithmetic mean of approximately 14 ug/day.

     As indicated by the previously mentioned difference in kidney cadmium
levels between smokers and nonsmokers, tobacco is an important source of cadmium
exposure.  Cigarettes are  estimated to result in the inhalation of 3 ug/pack
coupled with  a 25% absorption efficiency (EPA 1980).  Smokers, of which there
                                     11

-------
      99.99 99.9     99     95      80      50      20      5
      10
                                                                    1      0.1    0.01
o>
     0.1
<_>
o
    0.01
   0.001
                                I   T  I  I I  I   I   I
                             I   I    I   I  I  I	1	I
                                                             I. . .   L	 I .
       0.01   0.1      1
                                    20      50      80

                                        PERCENTAGE
95     99    99.9  99.99
         FIGURE 2b:   DRINKING  WATER DATA  FOR  CADMIUM:   FREQUENCY DISTRIBUTION OF
                     GRAB SAMPLES  AT TAP  OR WITHIN DISTRIBUTION SYSTEM.
                                          12

-------
      99.99 99.9      99    95      80       50       20
                                    1     0.1   0.01
    1000
-o

en
     100
to
o
a:
CD
UJ
t—I

O
      10
               I       I
I   I  I  I  I  I  I   I    II
                                                                    L..... .  I
       0.01   0.1      1      5      20       50       80      95     99   99.9   99.


                                        PERCENTAGE



         FIGURE 2c: DIETARY INTAKE OF CADMIUM:  FREQUENCY  DISTRIBUTION FOR GROSS

                    QUANTITY INGESTED ON A SINGLE  DAY.
                                                  99
                                           13

-------
were 54 million in 1978, each consume an average of 1.56 packs/day (Richmond
1981, as reported by Gilbert 1982).  As smoking is not done over an entire
lifetime, 1 pack/day might approximate an effective average consumption of
cigarettes over the first 50 years of a lifetime for the smoking subpopulation.

     It should be noted that cadmium has been detected in air contaminated
by tobacco smoke at levels as high as 100 ng/m3 (Brodie and Matousek 1974,
as reported by MacGregor 1981).  Such levels are higher than normally found
in outdoor urban air.

     Comparisons between the contributions of air (with outdoor quality),
water, and food to human uptake of cadmium are shown in Figure 3.  These results
have been summarized and compared with cadmium exposure via tobacco smoking in
Table 2.  Both Figure 3 and Table 2 were generated from the following data
and assumptions:

     Air:  Figure 2a concentration distribution, assuming 20 m3/day inhalation
           rate, 25% absorption efficiency.

     Water:  Figure 2b Canadian data, assuming 2 L/day ingested, 6% absorption
             efficiency.

     Food:  Figure 2c fecal  data, assuming 6% absorption efficiency.

     Tobacco:  3 ug/pack cigarettes, 1-3 packs/day, assuming 25% absorption
               efficiency.

It is apparent from Figure 3 that population exposure via food greatly surpasses
the exposures via drinking water and ambient air inhalation.  With a  long term
exposure threshold (for kidney dysfunction)  of 10-15 ug/day, it is also
apparent that hazardous ambient exposure is  almost not possible unless dietary
exposure is elevated.

     It should be noted that the variabilities (slopes) in Figure 3 may not be
strictly comparable due to the different time frames for sampling.  Furthermore,
in comparing to a long term exposure threshold (10-15 ug/day), it must be noted
that the short term exposure variability illustrated is inherently larger than
the long term exposure variability, which was discussed in the previous section
and illustrated in Figure 1.  It should also be noted that air, water, and
food exposures may vary independently of each other.

     The Table 2 exposure route estimates seem consistent with the Table 1
kidney cadmium levels.  In Table 1 the nonsmoker arithmetic mean is 56% of
the smoker arithmetic mean,  while in Table 2 it is 54%.  In Table 1 the non-
smoker arithmetic mean is 8.6% of the 200 ug/g kidney cadmium threshold,
while in Table 2 it is 5.8-8.7% of the 10-15 ug/day absorbed dose threshold.
As Table 2 is based on dietary intakes predicted from fecal  measurements, it
appears that these produce reasonable exposure estimates.  The FDA food data,
on the other hand, would seem to overestimate the current exposure when compared
to the threshold.  (Nevertheless, the FDA food data can produce equally con-
sistent results if the ug/day cadmium level  producing the 200 ug/g kidney
                                     14

-------
   99.99 99.9
   10
I  1  I  I  I   I   1
0.001 -
    0.01  0.1
20      50      80

    PERCENTAGE
95     99    99.9  99.99
      FIGURE   3:  COMPARISON OF DISTRIBUTIONS OF ABSORBED DOSE VIA INHALATION
                 AND VIA  INGESTION OF FOOD AND WATER.
                                      15

-------
Table 2:  Comparison of cadmium exposure routes.
                       Cadmium Absorbed into the Body, ug/day (a)

                      Typical  Exposure
Route
Air
Water alone (c)
Food + Water
Total Ambient
Cigarettes
Median
0.008
(0.005)
0.66
0.67

Arithmetic Mean
0.018
(0.006)
0.85
0.87
0.75
Elevated Exposure (b)
0.4
(0.04)
5.6
6.0
2.3
(a) Origins of data are described in text.
(b) For air, water, and food individually,  elevated implies 99.9^.
(c) The Kowel  et al. (1979)  fecal  data  provides estimate of combined food
    plus drinking water exposure.   Consequently, "water alone"  is not
    added into the "total  ambient" exposure.
                                     16

-------
level is approximately doubled and if the absorption efficiency for tobacco
smoke is doubled.)  Thus, the data and assumptions on exposure and toxic
threshold used in this assessment seem to fit together in a consistent manner
(although other data and assumptions are also capable of such consistency).
2.4  Potential Cancer Risks

     The incidence of cadmium induced kidney dysfunction has already been
discussed in terms of cumulative body burden in Section 2.2.  As cancer risks
are estimated from cadmium intakes rather than body burdens, such risks can be
dealt with here.  EPA (1984c) has estimated that lifetime exposure to an air
concentration of 1 ug/m3 would yield (as an upper bound) a lifetime incremental
risk of 2.3xlO~3 for lung cancer.  The upper-bound risk corresponding to the
0.0016 ug/m3 estimated median concentration in urban air would thus be 3.7xlO"6.
The highest 0.1% of the quarterly composite samples shown in Figure 2a exceeded
the median by nearly 50 fold; however, it is not known by what factor the
highest 0.1% of individual lifetime exposures exceed the median.

     For the arithmetic mean concentration, estimated to be 0.0035 ug/m3 in
urban air, the upper-bound risk would be S.lxlO"6.  This would correspond to
an upper-bound incidence of 25 cases/year nationwide (assuming that indoor air
is similar to outdoor air, and overlooking the fact that the urban mean concen-
tration is higher than the rural mean).  It must be recognized that this
prediction involves many uncertainties and is intended to be an upper bound.
The actual incidence of cadmium induced lung cancer could be much different.

     The contamination of indoor air with cadmium from tobacco smoke (or
other sources) could further increase the risks; however, the amount of this
increase, which depends on concentration and duration of exposure, has not
been estimated.

     For cigarette smokers, the risks from direct inhalation of tobacco cadmium
can be estimated by noting that smoking of 1 pack of cigarettes containing
3 ug/pack is equivalent to inhaling air having concentration 0.15 ug/m3 for
1 day.  The upper-bound cancer risk (due to cadmium alone) associated with
averaging 1 pack/day over a 70 year lifetime would thus be 3.5xlO~4.  The
upper-bound cancer incidence associated with 54 million smokers averaging
1.56 packs/day would be 415 cases/year.  As the actual  nationwide incidence
of lung cancer is around 90,000 cases/year (based on U.S. rates tabulated by
Schottenfeld 1975), it appears that inhalation of cadmium is not a major
contributor to the nationwide incidence of lung cancer.

     Cancer risks (if any) associated with exposure via food and drinking
water cannot be expressed quantitatively but may be considered to be low,
since EPA (1984c) found no evidence for carcinogenicity of ingested cadmium.
                                     17

-------
                                 SECTION 3

                     ENVIRONMENTAL SOURCES AND PATHWAYS
3.1  Use and Environmental Release

     Cadmium is obtained as a by-product primarily during the smelting of
zinc but also during the smelting of lead and copper.  Total domestic use of
imported and domestically produced cadmium has fluctuated between around
3000 and 6000 mt/yr (metric tons per year) in the period 1970-1982, depending
in part on the overall level of of economic activity (U.S. Bureau of Mines
1982).  Table 3a shows the total domestic use for recent years (U.S. Bureau of
Mines 1982).  Table 3b identifies the distribution of total use for the year
1979 (U.S. Bureau of Mines 1980, as reported by Kayser et al. 1982) and the
year 1981 (U.S. Bureau of Mines 1981).


Table 3a:  Total Domestic Use of Cadmium.

           Year                Use (mt/yr)

           1978                   4510
           1979                   5099
           1980                   3534
           1981                   4378
           1982                   3707


Table 3b:  Distribution of Domestic Use in 1979 and 1981.

                                Percentage of Total
      Uses of Cadmium             1979       1981

      Metal  Plating                51         34
      Batteries                    22         16
      Pigments                     13         27
      Plastics                     11         15
      Other                         3          8


The largest users of the cadmium-bearing materials from the above catagories
are the transportation equipment and defense industries.

     In addition to the commodity volumes shown, a significant amount of
cadmium is handled and released as an impurity in other materials.   Of partic-
ular importance are cadmium impurities in phosphate fertilizer,  in  fossil
fuels,  and in other metals, especially zinc.

     An integrated multimedia materials balance for cadmium release to the
environment is presented in Table 4, derived from data  from several  sources
                                     19

-------
                          Table 4:  Materials Balance for Cadmium  (mt/yr)


Zn/Pb Mining & Beneficiation
Zn/Cd Smelting
Electroplating
Batteries
Pigments & Plastics
Pesticide
Other Cd Products
Impurity in Zn Products
Iron & Steel Industry
Primary Non-Ferrous/Non-Zinc
Secondary Non-Ferrous
Printing/Photography
Other Manufacturing Activity
Coal Mining
Coal Combustion
Oil Combustion
Gasoline Combustion
Lubricating Oil
Tire Wear
Phosphate Detergent
Phosphate Fertilizer
Urban Runoff
Culturally Hastened Erosion
Natural Weathering
Potable Water Supply
POTW Effluent
POTW Sludge
Municipal Refuse
Totals

POTW
.
- (W) w
82 (U) v
5 (U)
7 (V)
-
N
N
10 x
- (W) w
1 (W)
11 (V)
45 (U) u
- (W)
- (w)
-
'
-
-
10 (V)
-
2 c
-
_
2 P
MUNICIPAL
REFUSE

-
817 d,e
672 d,f
1801 d,n
-
341 d
N b
-
-
-
N
N
-
-
-
-
N
N
-
-
_
-
_
-
-
175/485 j,t
3631 t

AIR
- (V)
7 (G)

1 (V)
13 (V)

N
N
14 (G)
218 (G)
2 (V)

N
-
202 (G)
363 (G)
13 (E)
1 (V)
5 (E,V)

-
_
N
N
-

14 (G)
38 (G)
891

WATER
8 (W) a
1 (W) w
4 (U)
13 (U)
1 (V)

N
N
10 (U)
1 (W) w
- (V)
- (V)
12 (U) u
45 (A) i
28 (W)
_
_
N
_
_
_
19 c
182 q
170 r
-
76 j
- m
-
570

LANDFILL
250 (V)
- (v)
370 y
9 (V)
17 (V)

N
N b
400 (Y)
_
20 (V)
N
N
N
429 (V)
_
_
N
_.
_
_
_
_
_
-

211 j,k
3593 g
5299

LANDSPREAD

_
_
_
_
9 (S) h
N
N
_
_
_
_
_
_
_
_
_
_
_
_
400 (X) S
_
_
_
-

123 j,k
-
532
Symbol  "N" signifies "no data"; symbol  "-" signifies "insignificant,
See following page for notes a-y and references (A)-(Y).

-------
Table 4 (Continued):  Explanatory notes.

The nominal reference year for the table  is 1981,  although many values  are taken  from studies  applicable
to other recent years.  The table recognizes four  ultimate dispositions,  to air,  water,  landfill,  and
landspread.  The totals contributed to POTWs and municipal refuse are ultimately  redistributed (near
the bottom of the table) to air,  water, landfill,  and landspread.

a.  Alternative estimate for Zn/Pb mine discharges:   114 mt/yr  (active  mines)  plus  90 mt/yr (inactive
    mines  (Yost 1978).
b.  The Cd impurity in zinc products has  been alternatively estimated as  173 mt/yr  (Yost 1978) or  as
    2371 mt/yr (JRB 1980).  Disposition of this  is not known but may be mostly toward landfill.
c.  Urban  runoff Cd:  concentration roughly 1 ug/L (EPA 1983)  for 21 trillion  liters/yr  (Sullivan  1977),
    flowing either t'o POTWs or directly to surface waters.  This quantity may  include material corroded
    from Cd and Zn plating as well  as Cd  emissions washout.
d.  The quantity ultimately disposed of in municipal  refuse is  estimated  to be the  entire amount of
    virgin Cd used in the industrial  catagory in 1981, minus the quantities identified to be released
    to the other media shown.  Material quantities have been balanced assuming that there is no  change
    in the accumulation within the technosphere  (socio-economic system);  that  is, quantities entering
    the technosphere are balanced by quantities  disposed of or  released to the environment.  Where
    population numbers and wealth are increasing,  such a simplifying assumption is  not strictly  accurate;
    nevertheless, it is used here to facilitate  understanding  of cadmium  movement.
e.  Quantity in spent electroplated products is  estimated as in Note d, minus  an  additional 216  mt/yr
    cycled as scrap to the iron and steel  industry (Yost 1978).
f.  Recycle of battery Cd is estimated to be 89  mt/yr (Yost 1978 and Versar 1980).   The  demand for
    virgin Cd in 1981 is assumed  to be required  to replace battery Cd which is not  recycled.  That
    Cd which is not recycled is assumed to be disposed of or released to  the environment.
g.  Refuse Cd entering landfill is  assumed to be the  total municipal  refuse Cd minus the municipal
    incinerator emissions.
h.  Alternative estimate for landspread pesticide  quantity:  Versar (1980) notes  an additional
    280 mt/yr of imported Cd fungicide/nernotocide. Such Cd is  used on  non-agricultural  turf.
i.  From active coal mines Versar (1984)  estimates 15 mt/yr to  water.

-------
Table 4 (Continued):

j.  Two independent estimates of POTW influent Cd are as follows:  175 mt/yr is sum of identified
    contributions; 485 mt/yr is estimated from a flow-weighted mean concentration of 13.4 ug/L for
    39 of 40 POTWs sampled by EPA (1982) (rejecting one outlier), multiplied by a total nationwide
    POTW flow of 26.2 bgd (EPA 1979b).  The 485 mt/yr influent Cd estimate almost balances the
    76 mt/yr total effluent Cd estimate (from the same EPA (1982) data, mean effluent concentration
    2.1 ug/L), added to 425 mt/yr total sludge Cd quantity independently estimated from a production-
    weighted mean concentration of 85 ppm for 353 POTWs listed by EPA (1979a) multiplied by 5 million
    mt/yr sludge production (EPA 1979a).
k.  The disposition of 5 million mt/yr sludge is taken as follows (EPA 1979a):
      Air (incineration):  21%, less quantity captured by emission controls.
      Ocean dump:  18%.  This quantity does not appear on the table.
      Landfill:  32%, plus captured emissions.
      Landspread:  29%.  (More recent data (EPA 1980) indicate landspreading of 31% of all  sludge.)
    These quantities do not take into account RCRA/CWA regulations limiting the rate of landspreading
    sludge cadmium on disposal sites.
m.  No sludge is expected to be released to fresh or estuarine waters.   The table does not include
    77 mt/yr sludge Cd ocean dumped.
n.  The pigment and plastic refuse Cd quantity compares favorably with  the combustable refuse Cd
    quantity independently estimated from the data of Campbell  (1976):   100 million mt/yr combustable
    refuse having Cd concentration 14 ppm.
p.  Background concentration of drinking water (measured within the distribution system)  is estimated
    from Meranger (1981), and assumed to apply to 26.2 mgd (EPA 1979b).   It might also be noted that
    since fecal measurements indicate excretion of 10-20 ug/person-day  (Kowel  et al.  1979), the 150
    million people served by POTWs would generate only about  1  mt/yr  of the POTW influent Cd.
q.  The culturally accelerated soil  loss is 70% of the roughly  1 billion mt/yr soil  carried to major
    streams (Wischmeir 1976).  Cd concentration of the eroded soil  is estimated to be 0.26 ppm (Carey
    1979).  Much of this release may be carried as a stream's particulate load.
r.  Natural  background concentration is highly uncertain but  is assumed here to be 0.1 ug/L;
    continental U.S. streamflow 1200 bgd (Miller et al.  1963).

-------
          Table 4 (Continued):
ro
CO
s.  Versar (1979b) projection for 1980.
t.  Totals in POTW influent and municipal  refuse are redistributed to the four basic media (air,
    water, landfill, and landspread).   The total quantity passing through the technosphere is the
    sum of the totals for the four basic media.
u.  Other manufacturing applies to industries evaluated by Versar (1984):  Aluminum and Non-Ferrous
    Metals Forming, Coil Coating, Copper Forming, Electrical  and Electronics, Foundries, Inorganic
    Chemicals, Leather Tanning, Pesticides Manufacturing, Petroleum Refining, Pharmaceuticals,
    Plastics Molding, Porcelain Enameling, Pulp  and Paper, and Textiles.
v.  The electroplating (metal finishing) contribution to POTW, 82 mt/yr,  assumes that 18.8% of the
    industry discharges raw wastes, and  81.2% has installed pretreatment  meeting EPA's PSES standards.
    As the raw waste before pretreatment is estimated to be 313 mt/yr, the recent installation of such
    pretreatment has greatly reduced this formerly large source.
w.  Alternate estimate for the entire  non-ferrous metals smelting industry:  10 mt/yr to POTW and
    10 mt/yr to surface water (Versar  1984).
x.  The Iron and Steel contribution to POTW is considered to be somewhere between the Versar (1984)
    raw waste estimate of 37 mt/yr and pretreatment estimate of 1 mt/yr.
y.  The quantity landfilled is assumed to be the quantity removed from wastewater prior to discharge:
    456 mt/yr raw waste (Versar 1984)  less 86 mt/yr discharged to water or POTW.
          REFERENCES:

          (A) Arthur D. Little (1979)
          (E) EEA (Coleman et al.  1978)
          (G) GCA (1981)
          (S) SRI (Casey 1979)
                                               (U)  Versar (1984)
                                               (V)  Versar (1980)
                                               (W)  Versar (Alchowiak and Maestri  1980)
                                               (X)  Versar (1979b)
                                               (Y)  Yost (1978)

-------
identified therein.  The table identifies four basic media (air, water, subsoil
or landfill, and topsoil) and two submedia (municipal wastewaters and refuse)
which in turn feed into the four basic media.  The table represents a closed
balance for commodity cadmium, assuming no change in the accumulation within
the socio-economic system.

     Publicly owned treatment works (POTW) receive wastes from households and
industries discharging to sewers.  It may be noted that the quantity estimated
to reach POTWs (based on POTW sampling) exceeds the contributions estimated
from industries and households.  It is not known whether the tabled values
underestimate some sources (such as electroplating wastes, which are projected
to be undergoing a substantial reduction), or whether the POTW influent data
are no longer representative.

     Municipal refuse is loosely defined to include junked end products, after
completion of their useful life.  For example, the entire volume of plated
cadmium, less plating wastes (liquid and solid) and metal recycle, is assumed
to be disposed of in the manner of municipal  refuse.  Thus, most of the values
in this catagory are not based on waste sampling.

     The major atmospheric releases have been estimated from U.S. EPA Air
Program information.  Such releases exclude those associated with windblown
soil  or other natural emissions.  Fossil fuel combustion and nonferrous metals
(excluding zinc) industries appear to constitute the largest sources to air;
zinc smelters, formerly a large .source, now appear to be tightly controlled.

     The releases directly to water involve discharges and runoff to surface
waters, including estuarine and coastal waters; they exclude dumping in open
ocean waters.  With implementation of national BAT (Best Available Technology)
standards, it appears that the formerly large industrial sources, such as elec-
troplaters, are being substantially reduced.   Consequently, it appears that
nonpoint sources, containing essentially background levels of cadmium, may
become the dominant sources when aggregated nationwide.  The total estimated
discharge of cadmium expected under BAT requirements for various industries is
presented later in Table 12; the values presented in Table 4 may or may not
represent BAT.

     The landfill catagory is loosely defined as any solid waste that is
disposed of on land but not spread thinly over the surface.  Most of these
values are highly uncertain, and no data are  available for some potentially
large waste volumes.  Under the assumption that junked end products are dis-
posed of in manner resembling that of municipal refuse, most of the commodity
cadmium would end up in landfill.

     The landspread catagory, on the other hand, involves material spread thinly
into the the topsoil, generally with the intention of increasing vegetative
productivity.  This catagory will be more fully discussed below.
                                     24

-------
3.2  Pathways for Contamination of Food via Topsoil

     The most critical environmental pathway for cadmium exposure involves the
contamination of topsoil.  Terrestrial plants uptake and concentrate cadmium
from the root zone.  The largest component of human exposure results from
consuming the edible portions of terrestrial plants (and for tobacco, inhaling
the combustion emissions); the other major component involves consumption of
animals (or animal products) fed with terrestrial vegetation (diet data from
Yost et al. 1978 and Pahren et al. 1979).  The consumption of fish and other
products of aquatic rather than terrestrial origin is too small a component in
the average American diet to account for a significant part of the overall
population exposure.

     Major pathways to agricultural topsoil are summarized in Figure 4.  The
potentially important pathways appear to be phosphate fertilizer, emissions
deposition, irrigation water, and sewage sludge landspreading.


     3.2.1  Phosphate Fertilizer

     Phosphate fertilizer, projected by Versar (1979b) to be 400 mt/yr for
1980, appears to be the largest pathway.  It is believed to affect nearly all
of the 130-150 million hectares cropland.  This contamination stems from cad-
mium's natural association with phosphate minerals.  The cadmium content of
phosphate rock is variable and depends on the origin of the phosphate.  Western
phosphates carry high concentrations of cadmium and thus result in more intense
contamination than the more commonly used Eastern phosphates (Versar 1979b).
The cadmium concentration in phosphates has been estimated as follows (Versar
1979b):

               Region                     Cd Concentration (ug/g)

               Florida                             10
               North Carolina                      20
               Tennessee                            3
               Western                            100

     The production and use of phosphate ore is summarized in Figure 5.
Western fertilizer has only about 6% of the market; however, its high cadmium
content makes makes it responsible for around 40% of the total  quantity of
cadmium in fertilizer.  Significant increases in the Western market share
would result in significant increases in cadmium application to cropland.
Although Versar did not anticipate large changes in the relative proportions
of Eastern and Western market shares before year 2000, reserves in many Eastern
mines are dwindling.  While some shifts in supply will  have to occur, ample
world-wide reserves make it difficult to predict which sources will  be tapped
(Stowasser 1975, Emigh 1972).
                                     25

-------
      EASTERN FERTILIZER
                             APPROXIMATE PERCENTAGE
                              OF CROPLAND AFFECTED

                                       94
   WESTERN FERTILIZER
     EMISSIONS DEPOSITION
                                      100
        IRRIGATION WATER
    POTW SLUDGE (IF UNREGULATED)
 0
FIGURE 4a:
 10
             20          30
            FLUX  (g/ha-yr)
                                       16
                                        0.15
850
ESTIMATED MEAN FLUX (APPLICATION OR DEPOSITION RATE)
OF CADMIUM TO CROPLAND TOPSOIL
       EASTERN FERTILIZER
                   WESTERN FERTILIZER
                   t	
                     EMISSIONS DEPOSITION
             IRRIGATION WATER
                    LANDSPREAD POTW SLUDGE (IF UNREGULATED)
                                     4-
             100
                                     400
500
FIGURE 4b:
             200         300

         TOTAL QUANTITY  (mt/yr)

NATIONWIDE QUANTITY OF CADMIUM REACHING CROPLAND TOPSOIL
                           26

-------
UNITED STATES
    48.774
   EXTORT
    12370
    26.4%
AGRICULTURAL
    31.333
    64.2%
 INDUSTRIAL
    4.571
    9.4%
   FLORIDA AND
   N. CAROLINA
     41,415
84.9% OP TOTAL US.
                        EXPORT
                         11310
                         283%
                     AGRICULTURAL
                         29314
                         703%
                       INDUSTRIAL
                                                               WESTERN STATES
                                                                    5.671
                                                              113% OP TOTAL U3.
                291
                0.7%
TENNESSEE


1388
33% OP TOTAL U3.

1 INDUSTRIAL

"11.688
J100%
EXPORT
AGRICULTURAL
INDUSTRIAL
2392
45.7%
1.060
18.7%
2,018
35.6%
          FIGURE  5  PHOSPHATE ROCK SOLD OR USED BY PRODUCERS - 1978
                                (THOUSANDS OF METRIC TONS)
                                      (VERSAR  1979b)
                                         27

-------
      3.2.2   Emissions Deposition

      To  varying degrees emissions deposition also affects essentially all of
the  130-150  million hectares of cropland.  Deposition fluxes are not accurately
known, however, and have been roughly estimated here to average 1 g/ha-yr in
rural areas.  Deposition fluxes appear to be much higher in urban areas than
in rural areas, with measurements often in the range 3-30 g/ha-yr (Nriagu
1980).   Urban soils are also known to have higher cadmium concentrations than
suburban soils (Carey et al. 1980).  Near smelters deposition rates have
been  measured as high as 100-500 g/ha-yr in the 1970s (Nriagu 1980), although
the more recent installation of pollution control technologies is believed
to have  greatly reduced such high rates.

      It might be noted that the 1 g/ha-yr estimate is consistent with what
would be calculated conservatively assuming that all U.S. emissions were depos-
ited  on the  area of the continental U.S.  Despite such a conservative assump-
tion, however, measured cadmium deposition rates in rural areas are often
several fold higher than the above estimate (Nriagu 1980).   It is not clear
whether such measurements simply reflect the wind blown movement of soil (with
its associated cadmium) from one area to another, or whether they indicate
actual enrichment of soil  with new cadmium.  Only the latter process is of
concern.

      The major emissions sources contributing to this pathway are believed to
be coal and  oil  combustion and primary nonferrous metal  industries (excluding
zinc/cadmium smelting, formerly an important source, now apparently tightly
controlled)  as shown in Table 4.  Incineration of municipal  refuse is apparently
a minor emissions source (as shown in Table 4), although refuse cadmium can be
volatilized  and potentially emitted at the relatively low temperature of 765°C.
However, as  only 5-10% of all  municipal  refuse is incinerated, the disposal  of
products with cadmium pigments,  plastics,  and batteries appears to have little
potential for causing human exposure.  (It should be noted,  however, that Yost
et al. (1980) postulate that a considerable amount of uncontrolled open burning
of trash is occurring; under this assumption they identify  such products as
having significant exposure potential.   The basis for expecting much uncon-
trolled trash burning is not clear, however.)

     Although emissions deposition appears to be a potentially sizable  route of
exposure, its intensity is generally so  low that it  is  not  obvious exactly how
it can cause specific individuals to exceed the kidney  effects threshold.
Even  in the worst case situation of homegardening on soil contaminated  with
poorly controlled smelter  emissions, exposure does not  appear sufficient
to induce kidney dysfunction in  otherwise  typical  people.


     3.2.3   Irrigation Water

     While contaminated irrigation water was  the cause  of the Itai-Itai  problem
in Japan, irrigation water does  not appear to be a major factor  for  cadmium
contamination in  the U.S.   Unreliable data generated by  the  insensitive  analyt-
ical  methods commonly employed make estimations difficult;  for Figure 4  the
                                     28

-------
99 billion gallons/day irrigation water consumed on 23 million hectares irri-
gated cropland (Versar 1979a) was assumed to have an average concentration of
0.5 ug/1.  It was also assumed that the cadmium content of the consumed water
was captured in the soil.  (Water consumption equals water use minus return
flow.  It was thus assumed that the cadmium concentration in the return flow
was the same as that originally in the irrigation water.)  JRB (1980)  has
identified a few areas where a possibility for irrigation contamination exists:
(a) the Snake River Valley in Idaho, (b) the Coeur d'Alene Valley in Washington,
(c) Central Florida, and (d) Central California.


     3.2.4  Sewage Sludge Landspreading

     Landspreading of sewage sludge differs from the above pathways in that it
can cause extremely intense contamination of very small areas.  Of the 5 million
mt/yr sludge produced (EPA 1979a), 31% is estimated to be landspread,  16?, on
food chain crops, 3% on non-food chain products, and 12% by distribution and
marketing, (EPA I980a).

     Concentrations of cadmium in sewage sludge range from less than 1 ug/g to
greater than 1000 ug/g.  POTWs serving small non-industrial communities tend
to have much lower sludge cadmium concentrations than do POTWs serving large
industrial cities.  Most POTWs are small and have relatively low cadmium
concentrations.  In the EPA (1979a) listing of 353 facilities, one-half of
the facilities had concentrations not greater than 17 ug/g; in a more  recent
listing of 511 facilities (Booz-Allen 1982) one-half had concentrations not
greater than 13 ug/g.  (The latter more recent data set is considered  more
reliable; the former is thus considered to have a slightly high bias.)  Neither
data set may be representative of the sludge quality that might occur  after
implementation of pretreatment standards in some important industries.

     The above medians are strongly influenced by many small POTWs that together
produce relatively little sludge.  Most of the sludge is produced by medium
or large sized facilities and is somewhat more contaminated.  For various
cadmium concentrations Table 5 shows the associated cumulative (a) percentage
volume of sludge, and (b) the percentage quantity of sludge cadmium.  (Table 5
does not show the percentage of POTWs.)  The production weighted mean  concen-
tration is 85 ug/g.  These statistics are based on an analysis previously
performed on the EPA (1979a) data set.  Such statistics are not available for
the Booz-Allen (1982) data set.

     RCRA/CWA regulations will limit the rate of cadmium application on sludge
disposal  sites to 500 g/ha-yr.  Sludge spreading rates for food chain  crops are
often in the neighborhood of 5-20 mt/ha-yr or more (CAST 1976, Yost et al.
1979, LaConde et al. 1978, EPA 1979a).  Economics often may not favor  sludge
spreading at rates much lower than the bottom of this range.  At a 5 mt/ha-yr
sludge spreading rate, the sludge cadmium concentration could not exceed 100
ug/g without exceeding the RCRA/CWA 500 g/ha-yr cadmium application limit.
Table 5 indicates that 73% of the sludge in the EPA (1979a) data set has concen-
tration less than 100 ug/g and could thus be spread at a rate of 5 mt/ha-yr
or greater.  It also indicates that only 38% of the total  quantity of  cadmium
                                     29

-------
Table 5:  Cadmium concentrations for cumulative nationwide percentages  of
          (a) the sludge production volume and  (b) the quantity  of  Cd in
          sludge.  (From analysis of data for 353 POTWs listed by EPA 1979a.)


       Cadmium            % of Sludge Volume        % of Sludge  Cadmium
    Concentration         with Concentration         at Concentration
       (ug/g)                 Less than                 Less than

         10                       8                        0.5

         25                      28                        4

         50                      42                       10

         75                      55                       20

        100                      73                       38

        125                      75                       39

        150                      79                       46

        175                      87                       61

        200                      90                       67

        250                      98                       90

        300                      99                       92

        500                      99                       93

       1200                     100                      100


Note:  The above statistics are generated tabulating the concentration, c-j,
and sludge annual production volume, S-j, for each of 353 facilities.  The
facilities were then rank ordered by increasing c-j, while adding up the
cumulative sludge volume,  E S-j, and cadmium quantity, SS-jC-j.
                                     30

-------
represented by this data set is contained in such sludges.  Likewise, only 42%
of the sludge volume and 10% of the sludge cadmium is contained in sludges
having concentrations of less than 50 ug/g, the concentration limit for
spreading sludge at the rate of 10 mt/ha-yr or greater.

     The overall effectiveness of pretreatment regulations in reducing the
concentration of cadmium in sewage sludge has not been estimated; however,
pretreatment is believed to be bringing about substantial reductions in impor-
tant industrial segments such as electroplaters.  The general effect of such
pretreatment on sludge landspreading can be illustrated as follows.  Using the
Table 5 statistics, it could be assumed for illustration that (a) 5-10 mt/ha-yr
were the mimimum economically feasible application rate, and (b) pretreatment
were to bring about a 60% reduction in sludge concentrations across-the-board.
In this case 98% of the sludge volume (and 36% of the current cadmium quantity)
in the data set could be spread at more than 5 mt/ha-yr; likewise, 75% of the
sludge volume (and 16% of the current cadmium) could be spread at more than
10 mt/ha-yr.  This hypothetical illustration indicates the general effects of
pretreatment when coupled with RCRA/CWA landspreading regulations: pretreatment
allows more communities to choose food chain landspreading as an economical
disposal alternative.  In so doing it is unlikely to bring about a substantial
reduction in the total amount of cadmium thereby landspread.

     The above described sludge landspreading regulations apply to disposal
sites only.  They do not apply to sludge which is either sold or given away to
individual farmers or gardeners.  This potentially critical  exposure pathway
is currently not regulated.  It should be noted that the contamination of
equivalent areas of commercial  agricultural  plots and home garden plots have
inherently different results.  As the contaminated commercial  produce would be
widely dispersed to many individuals, the result would be widespread low-level
exposure.  However, as the contaminated home garden produce could be a substan-
tial portion of the diet of a particular household, the result would be high
level exposure confined to a few individuals.

     Ryan et al..(1982), using (a) crop uptake factors for sludge amended
soils, and (b) Loma Linda University and adjusted FDA average diet data,
evaluated a scenario of home gardeners growing the entire vegetable,  grain,
and fruit component of their diet on sludge  amended soil.  For example, on
neutral  pH soils they estimated that a cadmium addition of 8,000-12,000 g/ha
would increase the gross ingestion of cadmium to 200 ug/day, a level  often
considered to be a hazard threshold.   For acidic soils, plant  uptake  of cadmium
is more pronounced, and the hazardous application rate somewhat lower.

     EPA (1979d) coupled similar reasoning with slightly different data and
assumptions.  They considered home gardeners growing some portion of  the  vege-
table component of their diet on sludge amended soils.   The  diet component
included legume, leafy, and root vegetables, garden fruits,  and potatoes;
it excluded grains and regular  fruits.  Based  on this analysis,  EPA (1979e)
promulgated cumulative application limits of 5,000-20,000 g/ha depending  on
pH and cation exchange capacity.   EPA (1980a)  calculated the sludge concentra-
tions corresponding to a  cadmium addition of 5,000 g/ha.   For  example,  at  a
sludge spreading rate of 4 mt/ha-yr,  application  of sludge containing  54  ug/g
                                     31

-------
cadmium would  result  in a cumulative addition of 5,000 g/ha after 25 years.
As  shown in Table 5,  much of the sewage sludge exceeds this concentration.'

      In conclusion, the total amount of sewage sludge cadmium expected to be
landspread can be summarized as follows.  Of the estimated 425 mt/yr sewage
sludge cadmium, 51 mt/yr (12%) is believed to be disposed of by unregulated
distribution and marketing, and will here be assumed to be landspread on crops,
If  unregulated, another 68 mt/yr (16%) could be landspread onto food chain
crops at disposal sites; however, under RCRA/CWA sludge disposal regulations
only about 7-26 mt/yr of this seems likely to be landspread.  Industrial pre-
treatment is expected to reduce sludge cadmium concentrations; however, this
change might be partially offset by increasing quantities of sludge landspread,
3.3  Long Term Implications of Contaminating Topsoil

     The total quantity of cadmium that could reach cropland topsoil may
thus be summarized roughly as follows:  phosphate fertilizer 400 mt/yr, emis-
sions deposition possibly as much as 140 mt/yr, irrigation water possibly as
much as 70 mt/yr, and sewage sludge (under current landspreading regulations)
possibly as much as 70 mt/yr.  The annual  additions would thus appear to total
as much as 680 mt/yr.


     3.3.1  Current Concentrations in Topsoil

     Figure 6 illustrates the measured distribution of cadmium in soils of
three states, Kansas, Montana, and Texas (data from Carey 1979).  The median
is 0.20 ug/g, the arithmetic mean 0.26 ug/g.  The total  amount of cropland is
around 130-150 million hectares (CAST 1976, USDA 1981);  they can be assumed to
contain 2.2 million kg topsoil per hectare (EPA 1979d).   The quantity of cadmium
in cropland topsoil would thus be around 74,000-86,000 mt.  The estimated
annual  addition of 680 mt/yr thus represents the addition of nearly 1% of the
accumulated amount per year.

     Other studies have found soil  cadmium levels in this same general  range.
For example, Pierce et al.  (1982) also happened to obtain an arithmetic mean
of 0.26 ug/g for biodiagenetically available cadmium in  soils of Minnesota.
Of particular significance  is their finding that the parent subsoils averaged
only 0.13 ug/g, indicating  cadmium enrichment of the topsoil  by some means.
It is not known whether such enrichment involves a natural  concentrating
process or whether it involves the deposition of cadmium onto topsoil.


     3,3.2  Forecasting Changes in Soil  Concentrations

     Cadmium, being a stable element,  has  no half-life for destruction.
Nevertheless, contaminated  topsoil  can eventually be removed from cropland by
erosion.   Based on the average U.S.  agricultural  soil  loss rate of 18.7  mt/ha-yr
(Center for Environmental  Reporting  1979),  the half-life of a six inch  layer
                                     32

-------
CONCENTRATION  (ug/g)




"TI
t— t
C

rn
*
o o
(/) a
— 1 3
70 »-t
1-1 C
—1 0
>— « O
O Z
z o
m
70 TO
CO — J
CO 1 — 1
«^j CD
z
C/)
" 3Z1
CO ^Z
C— 3»
m en
CO ?0
H-l
~n c~>
73 C
o r-
3C §
po r""
m
m co
o
^H P
3> to
— l ••
rn
• *n
m
c:
m
o
"*







o
o


o
t-*


•""^

tn



o

-o
m
m en
1 °
m
co
o

to
in


to
to


to
10
to
£









—

.

-

.
-
""
-

•
\

r
i
•


.

<0
to
o
0 ^.
o • "~^
H-» I-* 1— • CD CO
to
III 1


^


1_


.

•

-

:
'l. i

'i.
| —
. '
» •

\ *

•
',


1 1 1 1
to

to
i°
to

to
to

to
en



00
o

en
o
ro
0

en



>->



o
^
o
O
»— »

-------
of topsoil  completely mixed by tillage would be estimated to be 80 years.
Soil  loss  rates of different regions of the country differ considerably,
however.   Regional average half-lives may range from 29 years in the Lower
Mississippi area to  1270 years in California.  Site-to-site variability would
be even greater.

      As erosion may  act to remove the cadmium applied or deposited on topsoil,
it is apparent that  the cadmium concentration will not build up indefinitely.
Given a steady cadmium application flux, F (g/ha-yr), and soil  erosion rate,
E (mt/ha-yr), the cadmium concentration will eventually attain a steady-state
concentration, css (ug/g), readily calculable by recognizing that a well
tilled 6 inch layer  of topsoil  is analogous to a completely mixed flow reactor;

                           css = Cb + F/E                                (3-1)

where Cb is the cadmium concentration in the parent subsoil  (beneath the
topsoil).   Table 6 shows the results of such calculations assuming that
Cb=0.13 ug/g (as found by Pierce et al . 1982), and assuming various cadmium
application rates and soil erosion rates.

      It can be seen that long term build-up of cadmium levels can be quite
pronounced  at low erosion rates.   At high erosion rates, on the other hand,
the potential for accumulation  of cadmium is limited.   The table also presents
calculations for the nationwide arithmetic and harmonic mean rates of soil
erosion.  Because the plateau concentration css is related to the reciprocal
of the erosion rate E, the nationwide arithmetic mean  of css is related to
the harmonic (not arithmetic) mean of E (as indicated  by Meyer  1975).

      It must be noted that a substantial  period of time may be  needed to
approach a  steady state concentration.   The time behavior of this type of
system is given by O'Connor (1979):
            c(t) = [cb + F/E][1 - exp(-tE/M)] + [c0 exp(-tE/M)]          (3-2)

where c0 is the initial  concentration in topsoil, M is the quantity of soil
(2200 mt/ha), and t is time period in years.  Symbols F, E, and Cb are as
previously defined.  For a system where the topsoil  concentration begins at
the subsoil background concentration, given a value of 0.13 ug/g, and where
E is 11.8 mt/ha-yr, and F is 5 g/ha-yr, a period of over 300 years would be
needed to attain 90% of the steady-state concentration.  (Attaining 100% of
steady does not occur until t becomes infinite.)  The lower concentration
plateaus associated with rapid erosion rates can be approached more quickly,
in as little as a few decades.  The high concentration plateaus associated
with slow erosion may require a few millennia to approach.

     Finally, it is worth questioning whether the available data lend any
support to the validity of analytical framework applied above (i.e., Equations
3-1 and 3-2).  For Minnesota soils Pierce et al. (1982) found 0.26 ug/g to be
the current average concentration in topsoil, and 0.13 ug/g to be the average
concentration in subsoil.  Assuming that the topsoil  initial  concentration,
c0, was equal to the subsoil background concentration, c^, a prediction of the
                                     34

-------
Table 6:  Steady-state Plateau Concentrations Predicted for Various Cadmium
          Application Rates and Soil Erosion Rates.
Regional      Soil Loss    Calculated
Soil Loss      Rate, E   Soil Half-life
Catagory      (mt/ha-yr)      (yr)
                          Steady-state Concentration, Css (ug/g),
                           for Various Cd Application Fluxes, F

                           2 g/ha-yr    5 g/ha-yr    20 g/ha-yr
Low               1.2

Harmonic mean    11.8

Arithmetic mean  18.7

High             52
               1270

                129

                 81

                 29
1.8

0.30

0.24

0.17
4.3

0.55

0.40

0.23
16.8

 1.8

 1.2

 0.51
Notes on Derivation of Tabled Values:

Soil erosion rates taken from data presented by Center for Environmental
     Reporting (1979).  The harmonic mean is given by  n/ Z 1/x.
Half-life  t1/2 = 0.693/k
     rate in mt/ha-yr and
            where  k = -ln(l - E/M).  E is the soil erosion
            M = 2200 mt/ha (soil  quantity in 6 inch layer).
Steady state concentration  css = Cb + F/E  where  c& = 0.13 ug/g (the assumed
     concentration in the parent material), and F is the cadmium application
     flux in g/ha-yr.  (If the total flux of deposited material  (mt/ha-yr)
     (of which cadmium is a part) is significant compared to the soil  erosion
     rate, then E is the sum of soil erosion rate plus total material  deposition
     rate.  For example, in sludge application this accounts for dilution of
     the cadmium in the sludge itself.)
F = 2 g/ha-yr
F = 5 g/ha-yr
is an application rate typically expected for Eastern phosphate
fertilizer by itself.

is the estimated nationwide average application rate from
fertilizer, emissions deposition, irrigation water, and sludge
(assuming 680 mt/yr Cd on 140 million hectares cropland).
This value may be biased high.
F = 20 g/ha-yr is an application rate typically expected for Western phosphate
               fertilizer by itself.   It is also in the range of air deposi-
               tion rates found in  large cities.  It is substantially less
               than often occurs during sludge application,  however.  Sludge
               application is not considered in this table because it would
               not usually be done  for a sufficient period of time to attain
               a steady state.
                                     35

-------
current value of c(t) can be attempted using Equation 3-2.  Lacking historical
values for topsoil concentrations, cadmium deposition or application rates,
and soil  erosion rates, a true verification of the analytical  framework is
not possible.  Nevertheless, it can be noted that the observed topsoil  concen-
tration of 0.26 ug/g is consistent with what would be calculated for parameter
values that seem reasonable for Minnesota: an application rate of 4-5 g/ha-yr,
an erosion rate of 5-15 mt/ha-yr, and a time period of 50-100 years.  It must
be cautioned, however, that other unrelated processes could account for the
observed differences between topsoil  and subsoil  concentrations.


     3.3.3  Influence of Topsoil  Contamination on Human Exposure

     The intent of the following  analysis is to estimate the dietary intake
corresponding to growing food on  soils of various cadmium concentrations.
Ryan et al. (1982), for a similar purpose, coupled crop uptake factors  (ug/g
in crop versus ug/g or g/ha in soil)  with FDA data on diet composition.  They
discerned between acidic and neutral  soils, and average and vegetarian  subgroup
diets.  Due to lack of data on many crops, the needed uptake factors are some-
what uncertain.

     To be consistent with the current study's preference for using EPA fecal
data rather than FDA food data to estimate cadmium exposure (as discussed in
Section 2.3), this analysis will  approach the problem by a somewhat different
route.  It rests on the following basic premises:

(a)  Crop uptake of cadmium is taken  to be directly proportional to the topsoil
     cadmium concentration (although  the constant of proportionality may differ
     from crop to crop and from soil  to soil).

(b)  The nationwide arithmetic mean dietary exposure can be estimated from
     fecal cadmium measurements (Kowal et al. 1979); its value is taken to be
     14 ug/day gross quantity ingested or 0.85 ug/day absorbed dose.

(c)  The nationwide arithmetic mean soil  concentration can be estimated from
     available data (Carey 1979); its value is taken to be 0.26 ug/day.

(d)  An arithmetic mean exposure  would result from consuming an average diet
     grown on soils having average cadmium concentration (and average other
     properties).  That is, the population's sampled dietary exposure is
     the result of consuming foods grown on soils having the sampled concen-
     trations.

     The above premise (a) is technically sensitive.  One problem is that the
assumed linearity seems unlikely  to extend over a wide range of concentrations.
This may not be critical, however, since this analysis seeks only to apply it
over a fairly narrow range, with  mean topsoil concentrations increasing perhaps
2-3 fold over present values.  Another problem is that it implies that  cadmium's
bioavailability does not depend on the pathway by which it entered topsoil
(for example, via emissions deposition, sewage sludge, or parent subsoil).
While it has been found that initial  bioavailability varies with the chemical
                                     36

-------
form being added (as discussed by Logan and Chaney 1984), it is not known
whether such differences persist through the long time frames being considered
here (i.e., after a long period of equilibration).  Finally, the above premise
implies that it is the cadmium content of the topsoil, not the subsoil, that
effectively determines the cadmium content of the crop, even though the root
zone of many crops may extend somewhat below the topsoil.

     This approach does not assume that all combinations of crops and soils
have the same cadmium uptake factor (ug/g in crop verses ug/g in soil), but
only that such a proportionality would exist for each crop and soil type.
This approach is used here only for approximating average relationships between
soil contamination and dietary exposure.  It is not believed to be suitable
for discerning the effect of certain characteristics, such as site-specific
pH, differing markedly from the average.

     It is desirable, in any case, to consider how various portions of the
average diet are affected by cadmium enriched soils.   For example, in evalu-
ating the risks of home gardening scenarios, it is customary to assume that
only certain classes of food are garden grown.  Consequently, it is necessary
to discern the importance of each food class in the average diet.  Such infor-
mation is provided by Pahren et al. (1979) and Ryan et al. (1982).  Table 7
summarizes the results in terms of relative contribution of each food class to
total dietary exposure to cadmium in the average diet.  It can be seen that
the responsibility for cadmium exposure is spread broadly across food classes.

     Three cases have been considered in this analysis.  In Case 1 all food
classes have been assumed to be affected.  This may be most relevant in
evaluating the effects of raising nationwide average  soil concentrations.
In this case total dietary exposure is estimated to increase in proportion
with the increase in soil concentration.  There is some question, however,
about whether the cadmium contents of all foods, for  example, those of animal
origin, depend on soil cadmium levels (Logan and Chaney 1984).
    Consequently, in Case
fruits, potatoes, grains,
in soil cadmium.  As
67% of the average dietary exposure.
estimated to increase in proportion to the
     _2 only leafy, legume, and root vegetables,  garden
     and fruits have been assumed to respond to  changes
shown in Table 7, these food classes currently contribute
                 Exposure through these food classes  is
                      increase in soil  cadmium.   Exposure
through the other food classes is assumed to remain at current levels.

     In Case 3 only leafy, legume, and root vegetables, garden fruits,  and
potatoes have been assumed to be affected.  This is intended to correspond to
a scenario of serious gardeners home growing their entire consumption of these
food classes.  As shown in Table 7, these food classes currently contribute
38% of the average dietary exposure.  The other food classes are assumed to
remain at current levels.

     The total dietary absorbed dose, D (ug/day), is calculated as follows:

                     D = (fcDc/cc)c  +  (l-fc)Dc                        (3-3)
                                     37

-------
Table 7:  Relative Contributions of Food Classes to Cadmium  Ingestion
          in the Average Diet (adapted from Pahren et al. 1979, and
          Ryan et al. 1982).

            Food Class             % of Total Dietary Exposure

            Dairy products                      11
            Meat, fish, poultry                 11
            Grain and cereal products           27
            Potatoes                            24
            Leafy vegetables                     6
            Legume vegetables                    1
            Root vegetables                      3
            Garden fruits                        4
            Fruits                               2
            Oils, fats, shortenings              3
            Sugars and adjuncts                  2
            Beverages	6

            Total diet                         100
Table 8:  Dietary Absorbed Dose Corresponding to Consuming Food Grown
          on Soils of Varying Cadmium Concentration.

   Topsoil               Total Dietary Absorbed Dose (ug/day)
   Cadmi urn
   (ug/g)             Case 1           Case 2           Case 3

    0.26               0.85             0.85             0.85
    0.55               1.8              1.5              1.2
    2.0                6.5              4.7              3.0
    5.0               16.3             11.2              6.7
   10.0               32.7             22.2             13.0

Case 1 assumes that all  food classes respond to changes in soil cadmium;
       total dietary exposure is thus proportional  to soil concentration.

Case 2 assumes that leafy, legume, and root vegetables, garden fruits,
       potatoes, grains, and fruits respond to changes in soil cadmium.
       Exposure through  other food classes remains at current levels.

Case 3 assumes that leafy, legume, and root vegetables, garden fruits,
       and potatoes are  grown on the affected soil.  Exposure through
       other food classes remains at current levels.
                                     38

-------
where fc is the fraction of cadmium currently coming from the affected food
classes (from Table 7), Dc is the current average absorbed dose via diet
(taken to be 0.85 ug/day), and cc is the current average concentration of
cadmium in soil (taken to be 0.26 ug/g), and c is the concentration of cadmium
in the soil on which the affected food classes are grown.

    Table 8 shows the calculated exposure level corresponding to various soil
concentrations.  If the hazardous exposure level for kidney effects were taken
to be 12 ug/day absorbed dose, the corresponding hazardous soil concentration
would be calculated to be 3.7 ug/g for Case 1 (all foods affected), 5.4 ug/g
for Case 2 (leafy, legume, and root vegetables, garden fruits, potatoes, grains,
and fruits affected), and 11.5 ug/g for Case 3 (leafy, legume, and root vege-
tables, garden fruits, and potatoes affected).

     It is of interest to compare the results of this analysis with the results
of Ryan et al. (1982), who evaluated the equivalent of Case 2 by a different
method:  an "integrated response curve" crop uptake model, coupled with FDA
diet data.  In order to make such a comparison, the units of Table 8 were con-
verted to express ug/g gross ingestion per g/ha increase in soil cadmium (over
current levels).  To account for different assumptions about current exposure,
the Table 8 and Ryan results were then normalized to yield the ug/day increase
in gross ingestion per g/ha increase in soil cadmium.  The comparison indicated
that the results of the two methods differed by not more than 20%.  That the
results were this close must be viewed as somewhat fortuitous, since the two
methods incorporate different sets of errors and uncertanties which must be
expected to exceed 20% percent.  Nevertheless, that the two methods produce
nearly the same result lends confidence to the validity of both.

     These results were also compared with the results of EPA (1979d), which
evaluated the equivalent of Case 3 by a method somewhat related to that of
Ryan et al.  (1982).  Compared to the nationwide average increase in exposure
predicted by the present approach, EPA (1979d) predicted lesser increases for
neutral  soils and greater increases for acid soils.  The differences were
often roughly two fold.

     In coupling the results of Tables 6 and 8, in order to estimate the
dietary exposure resulting from various cadmium application rates, it should
be noted that Cases 1 and 2 of Table 8 are probably appropriate for coupling
with average application and erosion conditions of Table 6.  Case 3, on the
other hand, may be appropriate for a home gardening scenario, which might
under worst case conditions have high application rates and low erosion rates.
It must be noted,  in any case, that the time period needed to approach the
steady-state concentration plateau is perhaps 2-3 fold greater than the
erosional  half-life of the soil.   For typical  erosion rates this time period
is thus on the order of centuries.  Such a planning horizon is greater than
ordinarily applied to pollution control  problems.  However, as soil  contamina-
tion by metals is  often considered to be irreversible within ordinary planning
horizons (Purves 1977), such a long range perspective may be appropriate.

     In  order to predict the effect of soil  contamination on kidney cadmium
levels  and the incidence of toxicity, some assumptions in addition to the
                                     39

-------
 previously  stated  premises  (a)-(d) must  be  invoked:

 (e)   The  nationwide  distribution  of  kidney  cadmium concentrations can be
      estimated  from  available data.   For nonsmokers aged greater than 30
      years,  the distribution is taken to be log-normal, with geometric mean
      15.0 ug/g  (Johnson et  al. 1978).  The  geometric dispersion of 1.74 is
      assumed to be unchanging.

 (f)   The  kidney cadmium concentrations measured in the population can be
      taken to be the linear result of the measured exposures via air, food,
      and  drinking water.

      Considering premises (a)-(f) together, it can be seen that this method
 projects  nationwide  statistics from the statistics of the measured samples.
 The  kidney cadmium levels are assumed to be a function of the exposure via
 air,  food, and  drinking water.  The exposure via food is assumed to be a func-
 tion  of the  soil cadmium concentrations.  Exposure via air is assumed not to
 change with  changes  in soil cadmium levels.  Exposure via drinking water is
 negligible.

      The  method  thus does not explicitly employ measured soil-to-crop uptake
 factors.  However, an overall  uptake factor is implicit in linking the Carey
 (1978) soil  data with the Kowel  et al. (1979) dietary exposure data.   Likewise,
 the method does  not  explicitly state a hazardous threshold for absorbed dose.
 Nevertheless, by linking the Kowel dietary exposure data (with minor additional
 exposure  via  air) with the Johnson et al. (1978) kidney cadmium data, a long-
 term  absorbed dose of approximately 10 ug/day is implicitly made to correspond
 with  with a  kidney cadmium level  of 200 ug/g.

      The  method  can  be applied to values derived in Table 6.  For a cadmium
 deposition flux  of 5 g/ha-yr,  considered here to be a slightly high biased
 estimate  of  the  nationwide cropland average, coupled with the erosion rate
 of 11.8 mt/ha-yr, the estimated  nationwide harmonic mean, the average cadmium
 concentration in the topsoil is  predicted to attain a steady-state plateau
 of 0.55 ug/g  (Table  6), 2.1 fold higher than the current mean.   The dietary
 absorbed  dose corresponding to this concentration,  shown in Table 8,  is  1.5
 ug/day for Case  2.   After adding the exposure through air,  contributing  only
 0.02  ug/day, the factor increase in arithmetic mean exposure by all  routes
 (excluding tobacco) would be 1.74.  Assuming no change in the geometric  dis-
 persion,  this same factor increase must apply to the geometric  mean.   As  long
term  exposure is reflected by  kidney cadmium concentrations, the geometric
mean  kidney concentration for  a  nonsmoking population aged  greater than  30
years would  be increased to 26 ug/g under Case 2 assumptions.

      For  any particular geometric mean concentration  of cadmium in  kidney,  the
 fraction  of the population exceeding the  200 ug/g  kidney cadmium threshold can
 be calculated from Equations 2-2 and 2-1  (in Section  2.2.1)  for a  log-normal
distribution.  Assuming no change in the  geometric  standard deviation,  the
above increase in geometric mean  would be predicted to increase the  incidence
 from  the currently  estimated 1.5  persons  per million  up to  120  persons  per
mil 1 ion.
                                     40

-------
     Figure 7 shows the general relationship between topsoil cadmium concen-
tration and the incidence of exceeeding a kidney dysfunction threshold of
200 ug/g in the nonsmoking population aged greater than 30 years.  Table 9
shows the mathematical formulation of the model.  Incidence in the smoking
population should exceed that in the nonsmoking population; however, quantita-
tive predictions have not been attempted, primarily because the distribution
of kidney cadmium does not appear to be log-normal in the smoking population.

     Finally, to complete the analysis of the risks of soil contamination,
it should be noted that there is always some possibility that cadmium may at
some future time be judged to be carcinogenic via ingestion.  Predicting the
potential increase in cancer risk resulting from increases in soil cadmium is
straight forward, given the results presented above.  Under linear nonthreshold
assumptions (or multi-stage assumptions approximating linear nonthreshold), the
increase in cancer incidence is directly proportional to the previously shown
increase in mean exposure.


     3.3.4  Predictive Uncertainties

     In summary, the above analysis predicts the ultimate result of practices
pursued over a long period of time.  For a cadmium deposition rate and soil
erosion rate estimated to correspond to the nationwide means, the population
mean exposure by all  exposure routes is projected to increase by nearly two
fold before attaining a plateau.  Such an increase in mean exposure is projected
to increase the number of persons at risk for cadmium induced j32-micr°91obljlir)
proteinuria by 80 fold.

     The importance of the parameters used to predict changes in the mean
accumulated exposure among nonsmokers was determined by sensitivity analysis.
In such an analysis the value of each parameter is varied while the other
parameters are held constant.  The long-term factor increase in the mean
concentration of kidney cadmium (K/KC) was most sensitive to potential  errors
in estimating the current mean concentration of cadmium in topsoil (cc), the
flux of cadmium onto cropland (F), and the erosion rate (E).  It was less
sensitive to potential errors in the fraction of current intake from affected
food classes (fc) and the mean concentration in subsoil  (cb).   It was insensi-
tive to the current air and dietary intakes (Ac and Dc).

     The propagation  of the combined uncertainties of the important parameters,
cc» F» E, fc, and Cb, was evaluated by first-order error analysis (a procedure
described by Reckhow and Chapra 1983, and Meyer 1975).   In such an analysis
the variance, s£, in  the predicted values of a  dependent variable, y, is
related to the variance in reasonable estimates for the each parameter  or
independent variable.  For this application the dependent variable is the '
factor increase in mean kidney cadmium levels (y=K/Kc).   Assuming that  estimates
for cc and cb are correlated, and estimates for the other parameters are not
correlated, the uncertainty in y is approximated by the expression:
                                     41

-------
o

o

LU

I

>-

i—i
O

X
o
   10,000
    1,000
      100
to
X
LU
LU
O

LU
O
       10
      0.1
     0.01
                        I
                  0.2      0.4       0.6      0.8       1.0       1.2

                         MEAN CADMIUM CONCENTRATION IN TOPSOIL (ug/g)
                                                                           1.4
         FIGURE  7:   PROJECTED RELATIONSHIP BETWEEN NATIONWIDE MEAN CONCENTRATION
                     OF CADMIUM IN CROPLAND TOPSOIL AND INCIDENCE OF EXCEEDING
                     A KIDNEY CADMIUM CONCENTRATION OF 200 ug/g.

         Note:   The  incidence is  for  a nonsmoking  population  aaed greater than
                30 years.   The method of calculation is  presented in Table 9.
                                         42

-------
Table 9:  Formulations Setting forth a Relationship between Soil  Cadmium Levels
          and the Incidence of Exceeding a Kidney Dysfunction Threshold.
                 D = (fc DC/CC) c  +  (1 - fc) Dc

                 K = Kc (D + AC)/(DC + Ac)

                 Z = ln(Kt/K)/ln sg

                 0 = (1//TF)  J~exp(-z2/2) dz
Where:
 c = Future arithmetic mean concentration (ug/g) of cadmium in cropland top-
     soil.  This is the independent variable in Figure 7.  Its value could
     also be predicted as shown in Table 6.

cc = Current arithmetic mean concentration of cadmium in cropland topsoil
     (estimated to be 0.26 ug/g from Section 3.3.1).

fc = Fraction of current cadmium dietary intake from affected food classes
     (fc=1.0 for Case 1, and fc=0.67 for Case 2, as noted in Section 3.3.3).

 P = Projected arithmetic mean dietary absorbed dose (ug/day) resulting from
     consuming food grown on soils having mean concentration c.

Dc = Current arithmetic mean dietary absorbed dose; i.e, mean dietary absorbed
     dose resulting from consuming food grown on soils having mean concentration
     cc.  (Dc is estimated to be 0.85 ug/day from Table 2.)

Ac = Current arithmetic mean absorbed dose via air (estimated to be 0.02 ug/day
     from Table 2.)  Exposure via drinking water is assumed negligible.

 K = Projected geometric mean concentration (ug/g) of cadmium in kidney for a
     nonsmoking population aged greater than 30 years.  This population would
     have an arithmetic mean absorbed dose of D via food, and Ac via air.

Kc = Current geometric mean concentration of cadmium in kidney for nonsmoking
     population aged greater than 30 years; i.e., geometric mean kidney concen-
     tration occurring in a population having an arithmetic mean absorbed
     dose of DC via food, and Ac via air.  (Kc is estimated to be 15.0 ug/g
     from Table 1.)

K£ = Toxicity threshold for cadmium in kidney (taken to be 200 ug/g).

Sg = Geometric standard deviation kidney cadmium concentrations (estimated
     to be 1.74 from Table 1).

 Z = log-normal deviate.

 0 = Projected incidence of exceeding the threshold Kt in a population aged
     greater than 30 years, with kidney cadmium levels log-normally distributed
     with geometric mean K and geometric standard deviation sg.  Q is the
     dependent variable in Figure 7.


                                     43

-------
 s2 = (dy/dF)2 s2: + (dy/dE)2 s§ + (dy/dcb)2 s2b+ (dy/dcc)2 s2 + (dy/dfc)2 s2-

         + 2(dy/dcb)(dy/dcc)sc sc rc c                                   (3-4)
                              D  C  D C
where dy/dx is the partial  derivative of y with respect to each parameter "x"
(i.e., the sensitivity of y to potential errors in that parameter), s2. is
the variance in a set of reasonable estimates for the value of each parameter,
and r is the correlation coefficient between two parameters.

     A rigorous determination of the variance in parameter estimates seems
infeasible for this application, in part because the data used to derive the
estimates of cc, F, E, fc,  and c& are not based on a random sampling.   In place
of a rigorous determination, the level  of uncertainty was illustrated  using
parameter variance estimates that seem subjectively reasonable (as suggested
by EPA 1984b).  For estimates of cc, ct>, and F, the coefficient of variation
(s/X) was taken to be 0.5;  for estimates of fc (Case 2) and E the coefficient
of variation was taken to be 0.2.  Parameters cc and c& were  assumed to have
a correlation coefficient of 0.75.

     It was previously noted that the use of the "best estimates" for  F, E,
Cb, cc, and fc (for Case 2) indicated that, over a long period of time, kidney
cadmium would plateau at levels 1.74 fold higher than at present.  For the
subjectively selected variances, the results of the analysis  suggest that the
95% confidence interval  for this predicted factor increase, y, might range
from less than 1 to almost  4, assuming that the predictions of y are roughly
log-normally distributed.  The probability of no long-term increase in mean
levels of kidney cadmium (i.e., y £ 1)  was estimated to be 17%.

     First-order error analysis does not include the uncertainty in the validity
of the model  framework itself.  In this respect one important question is the
validity of assuming that the effective depth of the roots' nutrient uptake
zone is not substantially greater than  the depth of cultivation.  If a substan-
tial  portion of a crop's cadmium burden is due to uptake a background  cadmium
in the subsoil, then the crop burden would not be proportional  to the  quantity
of cadmium in the topsoil.   In that case the above analysis would overestimate
the increases in dietary cadmium.

     In addition, this first-order analysis has not been extended to the
uncertainty in the incidence of exceeding a toxic threshold (or the uncertainty
in the factor increase in the incidence).  As shown in Section 2.2.2,  the
predicted incidence is very sensitive to uncertainty in the geometric  mean and
geometric dispersion of the kidney cadmium distribution.

     Finally, it should also be reemphasized that the Figure  7 relationship
between mean topsoil  cadmium and kidney toxicity incidence assumes that the
kidney cadmium distribution is log-normal and that its geometric dispersion
(or arithmetic coefficient  of variation) is unchanging.  If,  however,  the
relative variability of soil  cadmium were also to increase in response to
contamination, then some type of increase in population variability of kidney
cadmium concentrations might also be hypothesized to occur.  Figure 8  presents
                                     44

-------
                                              Population variability
                                           of g?-microglobu1 in excretion
                                 #
                                  Population variability
                                   of kidney Cd levels
                            Population  variability
                            of other  physiological
                              factors affecting
                               kidney function
                    Population variability
                   of  Cd long-term exposure
               Population variability
                 of Cd absorption,
               depuration,  and  other
               physiological  factors
    Population variability
        of Cd  dietary
      long-term exposure
              ^
Population variability
of long term exposure
   via other routes
                                                 Day-to-day variability
                                                 of individual exposure
                        Population  variability
                    of  single-day dietary exposure
        Portion of variability
        of single-day exposures
        produced by variability
        of Cd content of  foods
         of the same catagory
      #
      i
       Variability of Cd  content
       within each food catagory
 Variability
 of soil  Cd
concentration
    ##
       Portion of variability of
     single-day exposures produced
       by individual  differences
        in dietary preferences
     for food catagories inherently
      having different Cd content
                        Variability  of  other
                       soil  factors  (e.g.,  pH)
                     as well  as cultivar traits
                    -».  Indicates link contributing
                         to variability.

                      #  Indicates a parameter  for
                         which measurements  are
                         available.

                     ##  Indicates a parameter
                         calculable  by  the methods
                         of Yost and Miles  (1979).
Figure 8:   Hypothetical  Relationships between Variability in Kidney Function, Long
           and Short  Term Exposure,  Food Concentrations, and Soil Concentrations.
                                         45

-------
a hypothetical  framework  relating the variabilities of cadmium concentrations
and exposures.   Such a framework can be used in a qualitative sense only,
however, because  some of  the relationships between parameters cannot be
described quantitatively  by existing data.
3.4  Other Pathways of Exposure

     In addition to the above mentioned critical exposure pathways involving
topsoil there are some minor pathways.  Direct inhalation of airborne cadmium
is of minor significance, as discussed in a previous section.  Emissions
deposition may constitute a larger portion of topsoil (and thus food) contami-
nation than emissions inhalation contributes to total exposure, as illustrated
by comparing the importance of air in Figures 3 and 4b.

     Direct ingestion of waterborne cadmium likewise was shown to be generally
insignificant (Figure 3).  The major releases to surface water would thus
appear to have little potential for bringing about hazardous exposure through
drinking water.


     3.4.1  Bioconcentration in Shellfish

     Of somewhat greater concern, however, is that waterborne cadmium may
enter the diet after bioconcentration by shellfish.  A seafood survey by Zook
(1976), as cited by Drury and Mammons (1979), found the cadmium concentration
of commercially caught Atlantic coast shellfish to average 0.575 ug/g, and
Pacific and Gulf coast shellfish to average 0.077 ug/g.  Data from a subsequent
survey indicated an average of 0.55 ug/g in 11 types of shellfish (Meaburn
et al. 1981).

     While shellfish consumption is too small  a portion of the average diet
for this pathway to have much influence on average exposure, the pathway can
be important for individuals if a significant portion of their diet consists
of shellfish.   Selected data on U.S.  per capita fish consumption is shown in
Table 10 (Javitz 1980).
Table 10:  Selected Data on U.S.  Per Capita Consumption of Fish and Shellfish,
     Catagory

  Total  U.S.
  Oriental  racial  identity
  Large central  city residents
  New England residents
  Mid-Atlantic residents
  Higher income  family members
  Consumption (g/day)
Mean"      Upper 95tTf Percentil e
14.3
21.0
19.0
16.3
16.2
16.7
41.7
67.3
55.6
46.5
47.8
49.0
                                     46

-------
     It can be seen that individuals in the upper 95th percentile of the
Oriental  subgroup consume 67 g/day of fish and seafood.  In the unlikely
situation that all  of this were in the form of Atlantic coast shellfish,
averaging 0.575 ug/g cadmium, this exposure pathway would increase absorbed
dose by 2.2 ug/day (assuming 6% absorption and subtracting the cadmium content
of the diet component replaced by such seafood).  Such exposure is somewhat
above the median absorbed dose of 0.67 ug/day but well below the probable
kidney hazard level of 10-15 ug/day.


     3.4>2  Land.fi,lied Cadmium

     The bulk of the cadmium handled by man (both as a commodity and as an
impurity) is believed to end up in municipal  and industrial landfill.  A survey
of eight municipal  landfills, however, suggests that such landfills do not
usually result in groundwater contaminated with hazardous levels of cadmium
(Fielding et al. 1981).  Concern about industrial solid waste disposal has
been greater, although a survey of 50 such sites did not implicate cadmium as
one of the metals of concern (U.S. EPA 1977).  Although cadmium may be mobile
in leachate through some soils (Fuller et al. 1979), such that precautions could
be appropriate for some solid wastes, it may be noted that exposure through
drinking water in general appears to be very minor (as noted in Table 2).
Consequently, as cadmium cannot be destroyed, an appropriate overall  strategy
might thus aim to steer cadmium away from air, surface water, and topsoil, but
rather toward the deeper soil strata, where its potential for return to the
biosphere is more restricted.
                                    47

-------
                                 SECTION 4

                         ECOLOGICAL CONSIDERATIONS


     The toxic effects of cadmium on terrestrial  ecosystems have not been
widely studied, apparently because the level  of concern has not been high.
Such ecosystems will  not be covered here.  Cadmium toxicity to aquatic eco-
systems, on the other hand, has received considerable attention.  Some aspects
are discussed below.
4.1  Aquatic Life Exposure and Effects

     The criteria for protection of aquatic life are being revised; the proposed
freshwater values depend on water hardness and apply to both acute and chronic
exposures (EPA 1984):

                   Hardness                     Acute-Chronic
                 (mg/L  CaCOs)                 Criterion (ug/L)

                      10                            0.31
                      50                            2.0
                     100                            4.5
                     200                           10.
                     400                           22.

     The proposed criterion appears to be fully protective of sensitive fish
species, despite being about 200 fold higher than the 1980 chronic criterion.
The saltwater criteria are 38 ug/L for acute exposure and 12 ug/L for chronic
exposure.  Both fresh and saltwater criteria are now expressed in terms of
"active" cadmium, signifying dissolved plus readily desorbed metal.  Recent
ambient data, however, are often only available for "total" cadmium, of which
"active" cadmium is only a portion.

     The levels of cadmium (and several  other metals) in natural  waters tend to
be low relative to the sensitivity of the analytical approaches commonly used
and relative to the level of laboratory contamination possible if meticulous
procedures are not used.  When sensitive methods are applied with care, typical
levels of cadmium found in fresh waters appear to fall in the range 0.01 - 0.1
ug/L (Martin et al. 1981).

     Frequency distributions of STORET cadmium data are shown in Figure 9.
The distribution of all STORET data (available before 1979) suggests a geometric
mean of perhaps 0.1 ug/L (by extrapolation), and a large geometric dispersion
of around 17.  It further suggests that the freshwater criterion may be exceeded
around 10% of the time.  These data are not considered to be reliable, however,
since the level of noise in the analytical methods may exceed the typical
                                     49

-------
       99.99 99.9     99     95     80      50       20      5      1     0.1   0.01
     100
      10
UJ
C_>

o
     0.1
    0.01
               p       i      i   i    i   i   i  i  i  i  i
                      j	i
i  i  i  i   i    i
i  .
        0.01   0.1     1      5      20      50      80       95     99    99.9  99.99


                                        PERCENTAGE



         FIGURE  9:   CADMIUM CONCENTRATIONS IN AMBIENT SURFACE WATERS:  FREQUENCY
                    DISTRIBUTIONS OF STORET DATA.
                                          50

-------
ambient levels by one to two orders of magnitude.   Data checking by the Office
of Toxic Substances and the Office of Water Regulations and Standards has
revealed systematic errors that produce a large positive bias in major portions
of the STORE! cadmium data base.  In addition, methods analyses indicate a
substantial  potential for positive bias when measuring cadmium below or near
the detection limit with the older procedures.  The frequent reporting of
cadmium in the ug/L range may thus have little significance (Delos 1981).

     In an attempt to circumvent the bias stemming from high detection limits,
EPA Regional Office data obtained using the sensitive graphite furnace (GFAAS)
method were extracted from STORET.  Such data were associated with Region  10
from May 1978 to the time of the study (March 1981).  The Region 10 GFAAS  data
show less variability than the nationwide data, even though the geometric
means appear to be almost identical.  Furthermore, despite representation  of
the Coeur D'Alene cadmium mining and smelting area in the data set, none of
the 73 values exceeded 10 ug/L.  At the hardness levels most commonly found in
the U.S. as a whole (50-300 mg/L) such a distribution might exceed the fresh-
water criterion about 2% of the time.  At the low hardness levels prevalent in
Region 10, however, perhaps 7% of the samples might exceed the criterion.
Either case, however, involves the worst case (and questionable) assumption
that all cadmium was in the active form.

     Based on single-species laboratory tests (used to derive the criterion),
moderate violations of the criterion would affect several species of trout and
salmon as well as some other fish and invertebrates.  Without ancillary field
data, the field applicability of laboratory tests may always be questioned,
particularly with regard to the toxic form of the pollutant, physiological
acclimation, genetic adaption, and community interactions (NRC 1981).  While
field validations of the cadmium criteria have not been done, the weight of
evidence supporting these criteria seems persuasive, particularly in light of
the number of species affected at levels near the criteria, the severity and
speed of the observed effects, and the applicability of the criteria only  to
readily available forms of the metal.  Thus, persistent violations of the
criteria seem likely to produce noticable changes in many biological com-
munities.

     In an attempt to further identify the magnitude of the aquatic cadmium
problem, the EPA fish kill files were searched.  For the period 1970-1978 only
one small fish kill was attributed to cadmium.  Nevertheless, cadmium's contri-
bution to some fish kills could be easily overlooked by field investigators
due to its presence at low levels relative to other metals such as copper  and
zinc.  It must also be noted, however, that in the year 1976 only 1.3% of  all
fish kills (and 0.1% of all killed fish) were attributed to metals released
during industrial operations (EPA 1979c).  While the lack of fish kills
strongly suggests the lack of widespread cadmium problems, an argument could
possibly be put forth that the effect of continuous discharges might be mani-
fested in a reduction of species diversity or an impairment of ecological
function rather than in discrete fish kill incidents.
                                     51

-------
 4.2   Controlling  Key  Sources

      Nationwide mapping of  STORE! data has not been found useful discerning
 associations  between  elevated cadmium levels and geographical areas having
 heavy concentrations  of metal handling industries or high population densities
 (Delos  1981).  Because STORET data are contributed by many different agencies,
 differences in analytical methods are Relieved to mask true geographical dif-
 ferences  in cadmium levels.  Likewise, while State agencies have identified
 river basins  that they consider to have cadmium contamination problems  (as
 shown in  Table 11), they apply differing criteria for selecting cadmium as a
 pollutant worthy of mention.  There is no certainty that cadmium is actually
 impairing the integrity of  aquatic ecosystems in the waters identified  in
 Table 11.  Likewise,  of course, it cannot be said that cadmium is not a problem
 in waters not identified in Table 11.

      Nevertheless, when data from a single laboratory using GFAAS were exam-
 ined, a relationship  between cadmium levels and point source discharges could
 be discerned.  A study of 60 sites in the Pacific Northwest indicated that
 stations downstream of point sources (of any type) had a geometric mean concen-
 tration two fold higher than stations unaffected by point sources (Delos 1981).
 It must be noted, however, that important industrial  catagories are in the
 process of reducing their cadmium discharges.  It might also be noted that
 some  of the most severe metals pollution problems have been associated with
mine  drainage (Perwak et al. 1980).

      Regulatory controls on cadmium and other metal  releases to ambient waters
 have  often been directed toward mitigating the effects of point sources on
 aquatic life.  Such controls have been instituted on  the basis of technology
 (for  example, industry-wide Best Available Technology, BAT)  or on the basis of
water quality (site-specific waste load allocations).   Table 12 summarizes
the total  discharge of cadmium estimated to be allowed under BAT for industrial
catagories.  The table also shows the estimated discharge allowed under PSES
 (Pretreatment Standards for Existing Sources) for industries connecting to
the sewer systems of POTWs (Versar 1984).   Based  on  these data,  BAT/PSES
controls could be expected to substantially reduce the total  point  source
discharge of cadmium from levels thought to be discharged in the late 1970's.
A dramatic example is the electroplating industry,  where the estimated BAT/PSES
discharge (Versar 1984) is more than an  order of  magnitude lower than  the
previous discharge (Alchowiak and Maestri  1980).

     Although the overall  importance of  cadmium in  impairing aquatic life  is
not known with any certainty, it is  one  of several metals for  which  aquatic
life concerns have been high.  In many situations,  however,  concerns about
cadmium cannot be readily isolated from  concerns  about  other metals.  Wastewater
discharges, as well  as the affected  water  bodies,  are  often  contaminated with
numerous metals.   Even after ecological  impairment  is  demonstrated  by  field
surveys of fish  and macroinvertebrates,  and after  contamination  is  documented
by chemical  analysis of water and sediment,  identifying  which  of the pollutants
in a complex  mixture are  the culprits  is  a formidable  problem.   There  are  two
general  approaches most likely  to find  use:  (a) pollutant-by-pollutant  analyses
and (b) whole effluent toxicity  analyses.
                                     52

-------
Table 11:  Summary of River Basins  Having Cadmium Contamination Problems,
           as Identified in 1980  State  305(b) Reports.
Basin No.
Basin
Sources Implicated*
0109
0202
020301
020302
020401
0207
030702
030801
030802
030901
030902
031001
031002
031101
031102
031200
031300
031401
031402
031403
0410
0411
041202
0502
0504
0505
0506
050702
0508
051002
0511
051302
051401
051402
070801
0710
080202
100902
1018
1019
102001
MA-RI Coastal
U. Hudson
L. Hudson
Long Island
U. Delaware
Potomac
St. Mary's
St. Johns
E. Fla. Coastal
Kissimmee
S. Fla.
Peace
Tampa Bay
Auci11a-Waccasassa
Suwannee
Ochlockonee
Apalachicola
Fla. Panhandle
Choctawatchee
Escambia
W.L. Erie
S.L. Erie
L. Erie
Monongahela
Muskingum
Kanawha
Scioto
Big Sandy
Great Miami
Kentucky
Green
L. Cumberland
L. Ohio-Salt
L. Ohio
U. Miss-Skunk Wapsipinican
Oes Moines
St. Frances
Powder
North Platte
South Platte
Mid. Platte
   VARIOUS
   NAT, MUN, IND
   VARIOUS
   VARIOUS
   MIN, MUN, IND
   MUN, IND
   IND
   UR, MUN, IND
   UR, MUN
   AG, UR
   AG, UR
   IND, AG, UR
   AG, MIN, IND, UR
   AG, SILV
   VARIOUS

   MUN, IND, AG, SILV
   VARIOUS
   MUN, AG, SILV
   MUN, AG, SILV
   UR, MUN, IND
   VARIOUS
   VARIOUS
   UR, MUN, IND
   CON, IND, MUN
   MUN, AG, CON
   IND, UR, MUN, CON
   .VARIOUS
   VARIOUS
   VARIOUS
   MUN, UR, AG, IND
   MUN, UR
   AG, MUN, IND
   VARIOUS

   NAT, UR, MUN
   MUN, AG
   MUN
   IND, LS, UR, MUN
   IND, LS, UR, MUN
   MUN, LS
                                       53

-------
 Table 11 (Continued)
Basin No.
Basin
                                                      Sources  Implicated
1025
1102
1105
1106
110701
111003
111203
120301
120302
120401
120402
1205
120701
121004
121102
1301
130201
130202
130301
1401
1402
1403
140401
140801
1502
150501
150502
150602
150701
2101
2102
Republican
U. Arkansas
L. Cimarron
Arkansas, Keyston
Verdigris
L.N. Canadian
North Ford Red
U. Trinity
L. Trinity
San Jacinto
Galveston Bay-Sabine Lake
Brazos H.W.
L. Brazos
Central Texas Coastal
S.W. Texas Coastal
Rio Grande H.W.
U. Rio Grande
Rio Grande-Elephant Butte
Rio Grande-Cabal!o
Colorado H.W.
Gunmi son
U. Colorado-Del ores
U. Green
U. San Juan
Little Colorado
M. Gil a
San Pedro-Wilcox
Verde
Aqua Fria L. Gil a
Puerto Rico
Virgin Island
AG, LS
IND, AG, LS
UR, AG, MUN
AG
AG, MUN
AG, UR
AG
MUN, IND, UR
MUN, IND, AG
MUN, IND, AG, VES
VARIOUS

MUN
MUN, AG
MUN
MUN, IND
MUN, AG, UR
MUN, UR
MUN, UR

IND

IND, MUN, LS
MUN, AG
CON, MUN, UR
*Sources are those named by the State for metals in general, not just cadmium.
 The order of appearance has no significance.
          AG  = Agriculture
          CON = Construction
          IND = Industrial
          LS  = Livestock
          NAT = Natural
          MIN = Mining
                    MUN     = Municipal
                    SILV    = Silviculture
                    SW      = Solid waste
                    UR      = Urban runoff
                    VARIOUS = Several  categories
                    VES     = Vessels
                                       54

-------
Table 12:  Waterborne Discharges of Cadmium Allowable  from  Regulated  Industrial
           Categories under Proposed Best Available Technology  (BAT)  and  Pre-
           treatment Standards for. Existing Sources (PSES)  (Versar  1984).
Industrial Catagory

Aluminum Forming
Battery Manufacturing
Coal Mining
Coil Coating (I)
Coil Coating (II)
Copper Forming
El ectri cal/Electronic Mfg
Foundries
Inorganic Chemicals (I)
Inorganic Chemicals (II)
Iron & Steel
Leather Tanning
Metal Finishing/Plating
Non Ferrous Metals
Non Ferrous Metals Forming
Ore Mining
Organic Chemicals
Pesticides Mfg
Petroleum Refining
Pharmaceuticals
Plastics Molding/Forming
Porcelain Enameling
Pulp & Paper
Textiles
Direct
Number
of
Plants
42
15
10375
29
3
37
83
287
114
35
738
17
2800
79
49
515
1082
42
164
80
565
28
355
229
Dischargers
Catagory
Total
Ib/day (a)
0
0
91
0
0
1
0
0
40
0
15
0
24
3
0
40
NA (b)
0
1
0
0
1
0
5
Indirect
Number
of
Plants
64
134
0
39
80
45
244
327
21
18
160
141
10200
85
147
0
535
39
47
392
1006
50
261
1047
Dischargers
Catagory
Total
Ib/day (a)
0
0
0
0
0
1
1
0
0
0
6
0
53
0
0
0
NA (b)
0
0
0
0 (c)
2
0
9
     (a)  Values are approximate.
     (b)  Not available.
     (c)  Excluded from PSES regulation
due to minimal importance.
                                     55

-------
      In the poTlutant-by-pollutant approach each wastewater constituent is
evaluated separately.  The traditional method is to compare water column
concentrations with effect levels determined by single-species laboratory tests
of pure substances.  Other methods, for example, utilizing sediment concentra-
tions or utilizing concentration-effect correlations obtained from field data,
could be developed in the future.

      In the general toxicity approach, on the other hand, control alternatives
may be tested at the bench, directly using toxicity tests as the measure of
effectiveness.  For some types of control alternatives the identity of the
culprit pollutant(s) might remain unknown using this approach.

     The pollutant-by-pollutant approach indicates the allowable concentrations
of pollutants in the effluent.  The general toxicity approach indicates an
allowable dilution of the whole effluent.  Neither approach, however,  indicates
the allowable frequency of exceeding the target level  (criteria).  Data indi-
cating the likelihood of observing ecological  changes  at various exceedance
frequencies are not generally available for any pollutant.  In conventional
practice, the allowable exceedance frequency is set arbitrarily, usually at
a low frequency.

     Reductions in wastewater concentrations of cadmium can often be readily
accomplished by sedimentation with or without  chemical  addition, by biological
treatment, and by some other methods (EPA 1980b).   These same processes are
used for conventional  pollutants and other toxic metals.  It would thus appear
that reductions in cadmium discharges are being brought about intentionally  by
limitations on cadmium loads and coincidently  by limitations on  other  pollutant
loads.
                                    56

-------
                               REFERENCES
Alchowiak, J., and B. Maestri.  1980.  Environmental release of cadmium from
selected industry sources to surface waters and publicly owned treatment works.
U.S. EPA, Office of Water Regulations and Standards.  Washington, DC.  Contract
No. 68-01-3852.

Arthur D. Little, Inc.  1979.  Risk assessments of priority pollutants: cad-
mium.  U.S. EPA, Office of Water Regulations and Standards.  Washington, DC.
Contract No. 68-01-3857.  Draft.

Battelle.  1977.  Multimedia levels cadmium.  U.S. EPA, Office of Toxic Sub-
stances.  PB 273 198.

Booz-Allen 8 Hamilton, Inc. 1982.  Description and comparison of municipal
sewage sludge generation and disposal data bases.  U.S. EPA, Office of Solid
Waste.  Washington, DC.

Brodie, K.G. and J.P. Matousek.  1974.  Determination of cadmium in air by non-
flame atomic absorbtion spectrometry.  Anal. Chim. Acta, 69 (1):  200-202.

Campbell, W.J.  1976.  Metals in the wastes we burn?  Environ. Sci. Techno!.
10(5):  436.

Carey, A.E. 1979.  Soil cadmium monitoring data.  U.S. EPA.  Washington, DC.
Memorandum, July 23, 1979.

Carey, A.E., J.A. Gowen, T.J. Forehand, H. Tai, and G.B. Wiersma.  1980.  Heavy
metal concentrations in soils of five United States cities, 1972 urban soils
monitoring program.  Pesticides Monitoring Journal 13 (4):  150.

Casey, S.E.  1979.  Agricultural use of cadmium.  U.S. EPA, Office of Water
Regulations and Standards.  Washington, DC. Contract No.  68-01-3867.

Center for Environmental Reporting.  1979.  Environmental  statistics 1978.
Council on Environmental Quality.  Washington, DC.  PB 294 141.

Coleman, R., et al.  1978.  Sources of atmospheric cadmium.  U.S. EPA,
Office of Air Quality Planning and Standards.   Research Triangle Park, NC.
EPA-450/5-79-006.  PB 80-12122-1.

Commission of the European Communities.  1978.  Criteria (dose/effect relation-
ships) for cadmium.  Rep. Working Group of Experts, Directorate-General for
Social  Affairs, Health and Safety Directorate.  Pergamon Press.

Council for Agriculture Science and Technology.  1976.  Application of sewage
sludge to cropland:  appraisal  of potential  hazards of the heavy metals to
plants and animals.  U.S. EPA, Office of Water Program Operations.
Washington, DC.  EPA 430/9-76-013.
                                     57

-------
Cherry, W.H.  1981.  Distribution of cadmium in human tissues.  In:  Cadmium
in the Environment, Part II:  Health Effects.  J.O. Nriagu (ed.).  John Wiley
& Sons.  MY.

Delos, C.G. 1981.  Assessment of need for water quality standards for cadmium.
U.S. EPA, Office of Water Regulations and Standards.  Washington, DC.  Internal
report.

Drury, J.S. and A.S. Hammons.  1979.  Cadmium in foods:  a review of the world's
literature.  Oak Ridge National Laboratory.  ORNL/EIS-149.  EPA 560/2-78-007.

Elinder, C.G. and T. Kjellstrom.  1977.  Cadmium concentrations in kidney
cortex samples from the 19th century.  Ambio 6:  270.

Ellis, K.J., D. Vartsky, I. Zanzi, S.H. Cohn, and S. Yasumura.  1979.  Cadmium:
in vivo measurement in smokers and non-smokers.  Science, 205(20):  323.

Emigh, G.D.  1972.  World phosphate reserves - are there really enough?
Engineering and Mining Journal, 173 (4): 90.

Fielding, S., J. Adams, P. Levins, and K. Thrun.  1981.  Investigation of
toxic pollutant discharge from municipal landfills.  U.S. EPA, Office of Water
Regulations and Standards.  Washington, DC.  Contract No. 68-01-3857.

Fuller, W.H., A. Amoozegar-Fard, and G.E. Carter.  1979.  Predicting the
movement of selected metals in soil: application to disposal  problems.  In:
Municipal Solid Waste: Land Disposal.  Proceedings of the Fifth Annual Research
Symposium.  U.S. EPA, Municipal Environmental Research Laboratory.  Cincinnati,
OH.  EPA-600/9-79-023a.

GCA Corporation.  1981.  Survey of cadmium emission sources.   U.S. EPA, Office
of Air Quality Planning and Standards.  Research Triangle Park, NC.
EPA 450/3-81-013.  PB 82-142-050.

Gilbert, D., et al.  1982.  An exposure and risk assessment for benzene.  UJ.S.
EPA, Office of Water Regulations and Standards.  Washington,  DC.  Contract No.
68-01-5949.

Grant, L.D.  1979.  Key points regarding factors of importance in the environ-
mental regulation of cadmium.  EPA briefing notes, Feb. 20, 1979.

Health and Welfare Canada, and Environment Canada.  1983.  Environmental  Contam-
inants Act hazard assessment report on cadmium.  83-EHD-101,  EPS 3-EP-83-9.

Hewlett-Packard.  1984.  HP-15C  Owner's Handbook.  Hewlett-Packard Co.
Corvallis, OR.

Indraprasit, S., G.V. Alexander, and H.C. Gonick.  1974.  Tissue composition of
major and trace elements in uremia and hypertension.  J. Chron. Dis.  27:
135-161.
                                     58

-------
Johnson, D.E., R.J. Prevost, J.8. Tillery, and R.E. Thomas.  1978.  The distri-
bution of cadmium and other metals in human tissue.  U.S. EPA, Health Effects
Research Laboratory.  Research Triangle Park, NC.  EPA 600/1-78-035.

JRB Associates, Inc. 1980.  Level II materials balance cadmium.  U.S. EPA,
Office of Toxic Substances.  Washington, DC.  Contract No. 68-01-5793.

Kayser, R., D. Sterling, D. Viviani. 1982.  Intermedia priority pollutant
guidance documents.  U.S. EPA, Office of Toxics Integration.  Washington, DC.
Kjellstrom, T., B. Lind, L. Linnman, and C.G. Elinder.  1975.  Variation of
cadmium concentration in Swedish wheat and barley.  Arch. Environ. Health,
30: 321.

Kjellstrom, T.  1979.  Exposure and accumulation of cadmium in populations from
Japan, the United States, and Sweden.  Env. Health Perspectives, 28:  169-197.

Kowal, N.E., D.E. Johnson, D.R. Kraemer, and H.R. Pahren.  1979.  Normal levels
of cadmium in diet, urine, blood, and tissues of inhabitants of the United
States.  J. Toxicology and Environmental Health, 5: 995.

LaConde, K.V., C.L. Eaker, R.J. Lofy, and R.P. Stearns.  1978.  A preliminary
assessment of cadmium additions to agricultural lands via commercial phosphate
fertilizers.  U.S. EPA, Office of Solid Waste.  Washington, DC.  SW-718.

Logan, T.J., and R.L. Chaney.  1984.  Utilization of municipal wastewater and
sludge on land - metals.  In:  Utilization of Municipal Wastewater and
Sludge on Land.  A.L. Page et al. (eds.).  University of California Press.

MacGregor, D.J.  1981.  Unpublished background review document for HWC and
EC (1983).  Environment Canada, Environmental Impact Control Directorate.
Ottawa, Ontario.  [See previously referenced HWC and EC (1983).]

Martin, H.W., and W.R. Mills, Jr.  1976.  Water pollution caused by inactive
ore and mineral mines:  a national assessment.  U.S. EPA, Industrial Environ-
mental  Research Laboratory.  Cincinnati, OH.  EPA-600/2-76-298.  PB 264-936.

Martin, J.H., G.A. Krauer, and A.R. Flegal.  1980.  Cadmium in natural waters.
In:  Cadmium in the Environment, Part I:  Ecological Cycling.  J.O. Nriagu
(ed.).  John Wiley and Sons.  New York.

Meaburn, G.M., K.B. Bolton, H.L. Seagran, T.S. Siewicki, S.M. Bingham, and
P.J. Eldridge.  1981.  Application of a computer simulation model  to estimate
dietary intake of cadmium from seafood by U.S. consumers.  NOAA, National
Marine Fisheries Service.  Charleston, SC.  NMFS SEFC - 74

Meranger, J.C., K.S. Subramanian, and C. Chalifoux.  1981.  Survey for cadmium,
cobalt, chromium, copper, nickel, lead, zinc, calcium, and magnesium in
Canadian drinking water supplies.  J. Assoc. Off. Anal. Chem. 64(1): 44.

Meyer, S.L.  1975.  Data Analysis for Scientists and Engineers.  John Wiley &
Sons.  New York.
                                    59

-------
Miller, D.W., J.J. Geraghty, and  R.S.  Collins.  1963.   Water  Atlas  of  the United
States.   Water  Information  Center.   New  York.

National  Research Council.   1981.  Testing the  Effects  of  Chemicals on  Eco-
systems.   National Academy  Press, Washington, DC.

Nriagu, J.O.  1980.  Cadmium in the  atmosphere  and  in  precipitation.   In:
Cadmium in the  Environment,  Part  I:  Ecological  Cycling.   J.O.  Nriagu  (ed.).
John Wiley & Sons.  New York.

O'Connor, D.J.  1974.  Mathematical  modeling of  natural systems.   In:
Mathematical Modeling of Natural  Systems.  Manhattan College.   New York.

Page, A.L., T.L. Gleason, J.E. Smith,  I.K. Askandar, and I.E. Summers  (eds.).
1984.  Proceedings of 1983  Workshop  on Utilization  of  Municipal Wastewater
and Sludge on Land.  University of California Press.

Pahren, H.R., J.B. Lucas, J.A. Ryan, and G.K. Dotson.   1979.  Health risks
associated with land application of  municipal sludge.   Journal  WPCF, 51(11)-
2588.

Perwak, J., et al.  1980.   An exposure and risk  assessment for  copper.   U.S.
EPA, Office of Water Regulations and Standards.  Washington, DC.   Contract No.
68-01-3857.

Pierce, F.J., R.H. Dowdy, and D.F. Grigal.  1982.   Concentrations  of six trace
metals in some major Minnesota soil  series.  J.  Environ. Qual.  11(3): 416.

Purves, D. 1977.  Trace Element Contamination of the Environment.  Elsevier.
Amsterdam.

Reckhow, K.H., and S.C. Chapra.  1983.   Engineering Approaches  for Lake
Management.  Volume 1.   Butterworth  Publishers.  Boston, MA.

Remington, R.D., and M.A. Schork.   1970.   Statistics with Applications to
Biological and Health Sciences.  Prentice-Hall.  Englewood Cliffs, NJ.

Reuss, J., H.L.  Dooley, and W.  Griffis.  1976.    Plant uptake of cadmium from
phosphate fertilizer.   U.S. EPA, Corvallis Environmental Research  Laboratory.
EPA 600/3-76-053.

Reuss, J.O., H.L.  Dooley, and W. Griffis.  1978.  Uptake of cadmium from phos-
phate fertilizers  by peas, radishes,  and  lettuce.  J. Environ. Qual.  7(1): 128.

Rhodes, R.C., et al.   1979.  Air quality  data for metals 1976 from the National
Air Surveillance Networks.  U.S.  EPA,  Environmental Monitoring and Support
Laboratory, Research Triangle Park, NC.  EPA 600/4-79-054,  PB 80-147-432.

Richmond,  T.B.   1981.   The changing cigarette.   Preface to the 1980 report of
the Surgeon General.   Office of the Surgeon General.  Washington,  DC.
                                    60

-------
Ryan, J.A., H.R. Pahren, and J.B. Lucas.  1982.  Controlling cadmium in the
human food chain:  a review and rationale based on health effects.  Environ-
mental Research, 28: 251.

Shottenfeld, D.  1975.  Cancer Epidemiology and Prevention.  Charles C. Thomas
Publisher.  Springfield, IL.

Stowasser, W.F.  1975.  Phosphate rock.  In:  Mineral Facts and Problems.  U.S.
Bureau of Mines.

Sullivan, R.H., et al.  1977.  Nationwide evaluation of combined sewer over-
flows and urban stormwater discharges, volume I, executive summary.  U.S. EPA,
Municipal Environmental Research Laboratory, Cincinnati, OH.  EPA 600/2-77-064a,

U.S. Department of Agriculture.  1981.  Agricultural Statistics 1981.  Govern-
ment Printing Office.  Washington, DC.

U.S. Bureau of Mines.  1980.  Mineral facts and problems.  U.S. Department of
Interior.  Washington, D.C.

U.S. Bureau of Mines.  1981.  Minerals yearbook.  U.S. Department of Interior.
Washington, D.C.

U.S. Bureau of Mines.  1982.  Minerals yearbook.  U.S. Department of Interior.
Washington, D.C.

U.S. Environmental Protection Agency.  1977.  The prevalence of subsurface
migration of hazardous chemical substances at selected industrial land disposal
sites.   Office of Solid Waste.  Washington, DC.  EPA/530/SW-634.

U.S. Environmental Protection Agency.  1979.  Health assessment document for
cadmium.  Environmental Criteria and Assessment Office.  Research Triangle
Park, NC.  EPA 600/8-79-003.

U.S. Environmental Protection Agency.  1979a.  Environmental impact statement:
criteria  for classification of solid waste disposal facilities and practices.
Office of Solid Waste.  Washington, DC. SW-821.

U.S. Environmental Protection Agency.  1979b.  1978 Needs  Survey, conveyance
and treatment of municipal wastewater, summaries of technical data.  Office of
Water Programs Operations.  Washington, DC.  EPA 430/9-79-002.

U.S. Environmental Protection Agency.  1979c.  Fish kills  caused by pollution
in  1976.  Office of  Water  Regulations and Standards.  Washington, DC.
EPA 440/4-79-024.

U.S. Environmental Protection Agency.  1979d.  Background  document: cumulative
cadmium  application  rates.  Office of Solid Waste.  Washington, DC.

U.S. Environmental Protection Agency.  1979e.  Criteria for classification of
solid waste disposal facilities.  44 FR 53438.
                                     61

-------
U.S. Environmental Protection Agency.  1980.  Ambient water quality criteria
for cadmium.  Office of Water Regulations and Standards.  Washington, DC.
EPA 440/5-80-025.

U.S. Environmental Protection Agency.  1980a.  Distribution and marketing of
sewage sludge products, pre-proposal draft regulation.  Office of Solid
Waste.  Washington, DC.  May 6, 1980.

U.S. Environmental Protection Agency.  1980b.  Treatability manual.  Office of
Research and Development.  Washington, DC.

U.S. Environmental Protection Agency.  1982.  Fate of priority pollutants in
publicly owned treatment works.  Volume 1.  U.S. EPA, Office of Water Regula-
tions and Standards, Washington, DC.  EPA 440/1-82/303.

U.S. Environmental Protection Agency.  1983.  NURP priority pollutant monitoring
project, summary of findings.  Office of Water Regulations and Standards,
Washington, DC.

U.S. Environmental Protection Agency.  1984.  Water Quality Criteria; request
for comments.  49 FR 4551.

U.S. Environmental Protection Agency.  1984b.  Proposed guidelines for exposure
assessment; request for comments.  49 FR 46304.

U.S. Environmental Protection Agency.  1984c.  Updated mutagenicity and
carcinogenicity assessment of cadmium.  Office of Health and Environmental
Assessment.  Washington, DC.  EPA 600/8-83-025B.

Versar, Inc.  1979a.  Cadmium and irrigation water.  U.S. EPA, Office of Water
Regulations and Standards.  Washington, DC.  Contract 68-01-3852.

Versar, Inc.  1979b.  Cadmium in the phosphate industry.  U.S. EPA, Office of
Water Regulations and Standards.  Washington, DC.  Contract No. 68-01-3852.

Versar, Inc.  1980.  Production and use of cadmium.  U.S. EPA, Office of Water
Regulations and Standards.  Washington, DC.  Contract No. 68-01-3852.

Versar, Inc.  1984.  Summary of effluent characteristics and guidelines for
selected industrial point source categories:  industrial status sheets.
Interim final report.  Volume I.  U.S. EPA, Office of Water Regulations and
Standards.  Washington D.C.

Wischmeier, W.H.  1976.  Cropland erosion and sedimentation.  In: Control of
Water Pollution from Cropland, Volume II, Overview.  U.S. EPA, Office of
Research and Development, and U.S. Dept. of Agriculture, Agricultural Research
Service.  EPA 600/275-026b.  ARS-H-5-2.

World Health Organization.  1976.  List of maximum levels recommended for
contaminants by the Joint FAO/WHO Codex Alimentarius Commission.   Secretariet
of the Joint FAO/WHO Food Standards Programs.  Rome, Italy.
                                    62

-------
Yost, K.J., L.J. Miles, and R.A. Greenkorn.  1978.  Cadmium flow in the environ-
ment associated with major uses and selected inadvertent sources.  U.S. EPA,
Office of Toxic Substances, Washington, DC.  Contract 68-01-4709.  December 1,
1978, draft interim report.

Yost, K.J. and L.J. Miles.  1979.  Dietary exposure to cadmium from land-
spreading sewage sludge in the states of Wisconsin and New Jersey.  U.S. EPA,
Office of Solid Waste.  Washington, DC.

Yost, K.J., L.J. Miles, and R.A. Greenkorn.  1980.  An evaluation of source
limitations for reducing exposure to environmental cadmium.  U.S. EPA,
Office of Toxic Substances.  Washington, DC.  Contract No. 68-01-4709.

Zook, E., J. Powell, B. Hackley, J. Emerson, J. Brooker, and G. Knoble, Jr.
1976.  National marine fisheries service preliminary survey of selected
seafoods for mercury, lead, cadmium, chromium, and arsenic content.  J. Agric.
Food Chem.  24(1): 47-53.
                                     63

-------
                                   DUE
US.  Environmental  Protection Agency
Region V, Library
230  South Dearborn Street
Chicago,  Illinois   60604

-------