Statistical Analysis of Left-Censored Geochemical Data
MS Tomlinson Defense
1. Viability of Using DGT Passive
Samplers to Measure Dissolved Trace
Elements in Subtropical Freshwater
and Estuarine Environments
Master of Science
Thesis Defense by
Michael S. Tomlinson
4. ➟To quantify dissolved trace element inputs
to aquatic habitats:
➟Methods time consuming and expensive
➟Ambiguous, definitions of dissolved vary
➟Discrete water samples are snapshots in time
➟Relation between sediment/tissue
concentrations difficult to relate to water
column concentrations
➟Often trace element concentrations <MDL
Motivation for the study
5. Nonpoint source pollution (after NPDES)
➟“Nonpoint source [NPS] pollution . . .
a significant factor in coastal water
degradation” (U.S. Congress, 1990)
➟“Stormwater linked to major coastal problems”
(EPA, 1993)
➟“May be greatest threat to marine ecosystems”
(Clark, 1995)
➟bioavailability can ultimately threaten human
health through consumption of aquatic
organisms
6. Why dissolved trace elements?
➟Dissolved phases considered bioavailable
➟“Bioavailability–the fraction of total contaminant
in surrounding medium which is correlated with a
quantitative biological response such as
biomagnification” (EPA, 1992)
➟Definition of “dissolved” is operational & varies
with filter pore size (typically 0.2 to 1 µm)
9. Water column sampling:
➟Concentrations may be <MDL
➟Snapshot in time
➟Sampling, containment, &
preservation can alter
chemistry
➟Filtering can alter chemistry
➟Ambiguity between dissolved
& particulate phases
10. Sediment sampling:
➟Sediments tend to be patchy
requiring numerous replicates
➟Bioturbation & other disturbances
can confound results
➟Difficult to obtain undisturbed
sediment sample
➟Sampling, containment, &
preservation can alter chemistry
➟Concentration relation
[sediment] ≠ [water column]
11. Organism bioaccumulation:
➟Difficult to locate suitable type/quantity of animals
➟May accumulate dissolved & particulate pollutants
➟Inter- & intra-specific comparisons difficult
➟Animals can metabolize or depurate pollutants
➟Non-sessile organisms can move in & out of area
➟Concentration relation [organism] ≠ [water column]
NOAA NS&T Mussel Watch Program
Ostrea sandvicensis (Hawai‘i)
Mytilus edulis (Maine to Delaware Bay &
US West Coast)
14. DGT (Diffusive Gradients in Thin-films)
➟Developed by Davison and Zhang (1994) of
Lancaster University
➟Measures dissolved Cd, Cr, Cu, Pb, Zn, Co, Ni,
Ag, Mn, Fe, Al
➟Work in saltwater, freshwater, sediments & soils
➟Consists of membrane filter, diffusive hydrogel,
resin gel, and housing (see diagram)
➟Effective pore size generally 0.002-0.005 µm
& no >0.020 µm (“standard” DGT)
➟Inexpensive (£10 or ~$17, March 2002)
15. Components of a DGT sampler
➟outer sleeve & piston
➟0.45-µm, polysulfone membrane filter
➟polyacrylamide hydrogel (~95% water)
➟layer of Chelex-100® resin in hydrogel
16. Cb = bulk solution concentration
DBL = diffusive boundary layer
δ = DBL thickness
∆g = diffusive gel thickness (ideally ≥10 × δ)
How
the
DGT
works
17. DGT facts
➟Generally only labile trace elements measured
➟Temperature-related effects are predictable
➟Diffusion coefficient independent of ionic
strength of receiving water (must be >1 mM)
➟Operating pH range of 5-10 for most elements
➟Not affected by hydrodynamic conditions
➟MDL for DGT after 1 day deployment is <4 pM
(concentration factor of ~300 times)
➟Analysis involves batch leaching (typically with
80 % recovery) followed by AAS or ICP-MS
18. Flow effects on Cd accumulation
C—concentration
DGT—diffusive gradient in thin-films
ASV—anodic stripping voltammetry
19. Cd accumulation with time &
different gel thicknesses
Time (hours) 1/∆g (1/mm)
∆g = diffusive gel thickness
MassCd(ng)
(Zhang & Davison, 1995)
20. Effects of ionic strength & pH on
Cd accumulation in DGTs
(Zhang & Davison, 1995)
30. Discrete sampling program
➟Manual quarterly sampling, typically near base
flow conditions (4 years)
➟Automated storm sampling (4 years)
➟Streamflow & water quality (T, C, pH, DO &
turbidity) at 5-minute intervals (4 years)
➟Estuarine grab samples collected & water quality
measured in situ concurrently with DGT (8 months)
38. DGT study design
➟Compare stream DGT results
with data from discrete base-
& storm-flow samples
collected over 4 years
➟Compare estuarine DGT
results with discrete samples
collected concurrently with
DGT retrievals over 8 months
43. DGT processingStep 1 - DGT disassembly
Step 2 - Removal of resin gel
Step 3 - Resin gel leaching
Step 4 -
ICP-MS
analysis
of DGT
leachate
44. Calculating mean concentration
where:
Cw = mean metal concentration in water
M = mass diffused into DGT
)g = diffusive hydrogel thickness +
membrane filter thickness
DT = diffusion coefficient at any temperature
t = deployment (exposure) time
A = area of DGT window
55. Stream results and why
➟Results of various methods for determining means
from discrete samples differed considerably
➟Rating curves were appropriate for upper but not
lower watershed (except for Pb)
➟DGT results generally comparable to, but less
than, grab sample means
➟DGTs measure the aquo ion, inorganic complexes,
and possibly small organic complexes & colloids
➟Grabs include larger colloids & organic complexes
➟No clear relation between flow & dissolved trace
element concentration
62. Estuary results & why
➟DGT deployments >2 weeks not recommended
➟Grab samples collected at different stages of tide &
under different streamflow & weather conditions
➟DGT results were significantly different (α = 0.05)
from concurrent grab results except for Co
➟DGT results were not consistently higher or lower
than results from concurrent grab samples
➟CuDGT > Cugrab & > chronic & acute HAR 11-54
standard (2.9 µg/L)
➟Need many more grab samples to accurately
characterize estuary
63. Conclusions:
➟To date this study is the longest deployment
of DGTs in diverse aquatic environments
➟DGTs preconcentrate dissolved trace
elements & remove matrix interference for
ICP-MS
➟DGTs are a simpler, faster, economical way
to measure dissolved trace elements
➟DGTs provide mean concentrations but they
also can show long-term variability
64. Conclusions (continued):
➟Watershed DGT & sample mean trace
element concentrations were similar
➟DGT means, however, often were less than
means from discrete samples
➟DGTs measure aquo ions, inorganic
complexes, small organic complexes, &
very small colloids
➟DGTs do not measure trace elements in
larger colloids or organic complexes &
small particulates
67. Conclusions (continued):
➟Except for Co, DGT & concurrent estuary
samples were significantly different
➟Estuarine DGT results were not consistently
less or greater than discrete sample results
➟Dynamics & complexity of estuary requires
far more samples to characterize chemistry
➟DGTs can be deployed for up to 3 months
in relatively clean, freshwater systems
68. Conclusions (continued):
➟Biofouling limits DGT deployments
1-4 weeks in subtropical estuaries
➟Operational pH range for DGTs (5-10) is
normally not a problem
➟Ionic strength rarely < 1 mM (~0.2 % of the
time in the upper watershed during storms)
➟DGTs are viable method for measuring
dissolved trace elements in subtropical
freshwater & estuarine environments
69. Eric, a man who loves his work . . .
. . . maybe a little too much?
70. Let it never be said that Eric . . .
. . . hides from his students!
90. Revised Sampling Scheme
➟Multiple (>5) blank checks before deployment
➟Three replicate DGTs deployed at each site
➟Dilute leachate by no more than 4 times
➟Continue temperature recording with TidbiTs
➟Locate inexpensive conductivity recorder
➟Deploy mid-depth in deeper stream waters
➟Collect or locate OC & speciation data
➟Multiple depths & locations in estuary
➟Deploy short- and long-term DGTs in freshwater