Improving the dosing method and analysis of metal mixture dose response in soil
Date
2019-08-27
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
ORCID
0000-0003-1714-8362
Type
Thesis
Degree Level
Masters
Abstract
Examination of a stressor’s toxic effects on terrestrial organisms is a relatively new field
compared to human health and aquatic toxicity, and the toxicity of multiple stressors on
terrestrial organisms is newer still. Dose-response functions and statistical analyses have crossed
over from aquatic toxicity research, however the effect of soil properties on soil toxicity is still
being researched extensively.
Being home to about 3.5 million contaminated sites (Swartjes, 2011), the European
Union has a large incentive to research the effect of stressors in soil. Smolders et al. (2009) used
this research to develop the basis for the EU REACH PNEC calculator. The calculator uses pH,
eCEC, clay content, and background metal concentrations to estimate the Predicted No Effect
Concentration (PNEC) of a stressor in soil: the upper concentration where there should be no
significant effect on a soil ecosystem. The calculator is limited to estimating the PNEC of single
stressors, and combinations of are assumed to follow Concentration Addition.
Rather than delving into the chemical and biological complexities of how stressors in soil
interact with one another and test organisms, Jonker et al. (2005) developed a statistics based
tool to evaluate mixture toxicity. Their method uses the concepts of Concentration Addition and
the sigmoid dose-response function paired with a deviation parameter. The deviation parameter
changes depending on whether a user wants to test perfect Concentration Addition, synergistic or
antagonistic effects, dose-ratio dependent deviation, or dose-level deviation. The methods in
Jonker et al. (2005) were adapted to analyse binary mixture behaviour using a Microsoft Excel
spreadsheet. While this worksheet may work well for small mixtures, it has not been used for
larger combinations of stressors.
To improve the results found using the Jonker et al. (2005) methods, I created scripts
using the optim() function in R (R Development Core Team, 2008). The scripts will evaluate
toxicity data for mixtures of up to five stressors for any of the four types of deviation from
Concentration Addition. Rather than using an iterative method as developed in the Excel
spreadsheet, the R scripts use the uniroot.all() function in the rootSolve R package (Soetaert,
iii
2009). The uniroot.all() function simultaneously evaluates all data points to make the Jonker et
al. (2005) equations valid. The optim() function then changes the individual stressors’ dose
response curves to minimize the differences between the predicted response and the observed
response. Using synthetic data with varying data point spread and starting parameters, the R
script produced lower sum of square values than the Excel sheet when modified to evaluate
mixtures of five stressors. In addition to finding a lower sum of squares, the annotated R scripts
offer a solution to researchers examining the effects of mixtures larger than binary without
needing to develop the tools themselves and reduces the “black box” of custom Visual Basic for
Applications (VBA) functions if they are not familiar with that language.
Prior to analyzing mixture toxicity data, one must ensure that one’s data is useful. In
some cases, a stressor may be applied directly to a test medium and a researcher can observe its
effect on an organism. In the case of metal effects on soil organisms, however, a problem may
occur. The standard application of metal salts to a soil and subsequent leaching can remove a
different proportion of each stressor from the test medium. Dosing strategies that require
specific combinations of metals, such as fixed-ratio rays, can pose a challenge to a researcher.
To find an alternative to metal salts, five soils were dosed with five different metal
mixtures in three different ways. Metals were applied as aqueous nitrate salts and leached, as dry
powdered commercially available metal oxides, and as spinel-like minerals. The spinel-like
minerals were made by mixing aqueous nitrate salts, adding iron nitrate in a 2:1 molar ratio of
iron to all other metals, precipitated, and annealed in a muffle furnace to remove nitrates. The
minerals were intended to resemble franklinite, which is commonly found in contaminated
smelter sites. Folsomia candida, an extensively studied and easy to culture hard-bodied soil
invertebrate, was used as a response organism.
The spinel-like minerals and oxides were both more effective at retaining the relative
proportions of each metal to one another. In some cases, over 60% of total metal nitrates added
to soils was removed during leaching in addition to losing different amounts of each metal. Even
though the metal concentrations were lower in nitrate dosed soils, the average reproduction was
similar to the oxide dosed soils, where the spinel dosed soils showed no effect on reproduction.
The research here shows that, with some more research, dosing soils using metal oxides is a
promising alternative to aqueous nitrate dose method.
Description
Keywords
metals, mixture toxicity, soil invertebrates, concentration addition
Citation
Degree
Master of Science (M.Sc.)
Department
Soil Science
Program
Soil Science