phosphorus particles, etc.). In addition, the vari-
nation. For each study, six mallards were placed
ous components of the sediment matrix may be
within these pens for various lengths of time, and
segregated or not randomly distributed. Pierre
feeding behavior and mortality were monitored.
Gy (Pitard 1993, Gy 1994) calls these two proper-
Over the course of the study, 282 mallards were
ties constitution heterogeneity and distributional
exposed: 32% died of apparent white phosphorus
heterogeneity. Only the second of these may be
poisoning. Waterfowl mortality occurred within
reduced by homogenization. The error associated
each pen (Table 2).
with constitution heterogeneity, called funda-
One of the most confounding factors in this
mental error, may be reduced by taking larger
study was being unable to quantify white phos-
samples (Ramsey et al. 1989). Correct sampling,
phorus exposure of the mallards within each pen.
where all constitutive elements have an equal
Because of the heterogeneous distribution of
probability of being selected, requires the proper
white phosphorus and the wide range of concen-
sampling tool that delineates and extracts the
trations found in discrete samples, average white
sample from the media. (For example, the proper
phosphorus concentration estimates were not
sampling tool for a layer of sediment would be a
useful for estimating exposure. Almost all the
cylindrical corer.) When designing a sampling
surface samples collected from the pens had mea-
plan, we must consider that these multiple sources
surable amounts of white phosphorus (Racine
of error place limits on achievable sampling objec-
and Brouillette 1995) (Fig. 1), but most birds ex-
tives, especially because there are always time
posed to these contaminated sediments were not
and cost constraints.
acutely poisoned.
Our initial objective at ERF was not to deter-
To see if chronic poisoning was happening,
gizzard contents and fat samples were taken from
mine average concentration. Rather, we wanted
those birds that did not die from acute poisoning
to identify the spatial bounds of white phos-
during exposure to contaminated sediments.
phorus contamination within the salt marsh.
White phosphorus was not detected in any of the
Sampling along transects at intervals correspond-
samples, suggesting that there was no chronic
ing to the fallout radii of the kinds of munitions
poisoning. Instead, acute poisoning following the
used was adequate to meet this objective. Now
ingestion of a single lethal dose of white phos-
that this site is undergoing remediation, ques-
phorus appeared to be the more likely scenario.
tions arise, such as, "Has the level of white phos-
Since the LD50 for mallards is estimated to be 3 to
phorus contamination in an area declined?" or,
4 mg/kg of body weight (Sparling et al. 1995),
"Are lethal quantities of white phosphorus still
available to waterfowl feeding in the area?"
and the waterfowl most at risk weigh from 0.25
Owing to high local spatial variability, the number
(teal) to 1 kg (mallards), a lethal dose is provided
of discrete samples needed for adequate coverage
by the ingestion of milligram-size particles of
and resolution to answer these questions with con-
solid white phosphorus, not the nanogram to
fidence is far greater than what is affordable.
microgram quantities associated with most of the
To minimize analytical costs while providing
sediment samples. The "bioavailable" form of
adequate resolution to characterize a contamin-
white phosphorus must be considered when
ated area, composite sampling is typically used
assessing remedial need or success at this site.
(Garner et al. 1988). By pooling and thoroughly
mixing many samples, the concentration esti-
Objective
mates obtained ideally represent the average con-
For this study, the contaminant of interest is a
centrations in the area sampled (Gilbert 1987),
particulate dispersed in a particulate medium.
and are more likely to be normally distributed
Because particulate samples are inherently hetero-
than estimates obtained by discrete samples
geneous, sampling error is unavoidable; but we
(Exner et al. 1985). However, when the contami-
seek to minimize it by understanding the sources
nant is acutely toxic and located in hot spots,
of error. In the case of white phosphorus contami-
compositing is not generally considered desirable
nation on a firing range, there is the spatial
heterogeneity introduced by the mode of contam-
because the compositing dilutes the samples
ination. Then there is the heterogeneity attribut-
taken from the hot spots (Exner et al. 1985). Infor-
able to the complex mixture of components that
mation on hot spot locations, however, can be
make up the sediment matrix (i.e., silt and clay
obtained if the composite samples are formed
particles, various kinds of organic matter, salts,
from samples taken at contiguous locations (Gore
metallic debris, water, various sizes of white
and Patil 1994).
3
Previous Page
