Discussion
Data Quality
Samples contained much higher concentrations of major elements than expected based on the maximum known concentrations of analytes present in arthropod reference materials (KRIK-1, BFLY-1, and VORM-1). Because the calibration standards were diluted based on the concentrations of analytes in those reference materials, for many samples, major analyte concentrations fell outside of the calibration range, even after in-run automatic 400x dilution. However, these data were still included due to high confidence in the calibration curves, almost all of which had a correlation coefficient of 1.0000 (Table 2). For some trace elements, the concentrations obtained for some samples fell below the limit of detection. This is also likely due to the high dilution factors used to create calibration standards based on known concentrations of analytes in arthropod reference materials.
A major source of error in this study is likely the limited precision of the balance used to measure the dry masses of each sample before digestion. This error is especially evident for the taxa Araneae and Tanytarsus, which had the smallest initial dry masses (Table 5). For most analytes, these taxa also had the longest interquartile range. It is unlikely that this spread in the data reflects true variation in the elemental concentrations across samples, as all triplicates of both taxa contained 10-40 individual arthropods. It is more likely that the imprecision of the balance resulted in lots of variation around the true masses of the small samples, resulting in strong variation in analyte concentrations despite the high accuracy and precision of the ICP-MS. Future analyses should use a microbalance to ensure as precise masses and accurate concentrations as possible.
Modifications for Future Methods
Knowing that samples of similar dry masses can be diluted 100+ times, I will run a semi-quantitative analysis on a representative subset of my samples (1 sample per taxon) to understand what order of magnitude to expect the concentrations of each analyte to fall within for each type of arthropod. This will allow me to better adjust the concentrations of stocks in my calibration standards so that major and trace elements should reliably fall within the calibration curve. It will also minimize the risk of accidentally running samples with too high of a TDS through the instrument.
Though quality control standard recoveries were close to 1 for all analytes in their optimal modes, future analyses should include at least one quality control standard with a biological matrix that undergoes the same preparation as the unknown samples. KRIK-1, BFLY-1, and VORM-1 from the National Research Council Canada would likely have more similar interferences to the sample matrices compared with other standards that do not contain arthropod organic material. Such interferences would make quality control recoveries more relevant to the arthropod samples. Further, following unknown samples with spiked matrix tests of samples at different dilutions would help with evaluating the presence of any significant matrix effects.
Timeframe and Budget
The proposed analysis run time and budget were 1 hour and $1,240. My actual run time on the ICP-MS was under my estimation (about 40 minutes), while my budget was above expectations. If every row of data, including calibration standards, quality control standards, and blanks costs $20, then ICP-MS analysis alone cost my project $1,140. Many samples were run twice, because the instrument automatically diluted and re-ran samples that were over the calibration range undiluted. Additionally, acid digestion at $60 per sample for my 21 samples and a procedural blank cost $1,320. That brings the total budget to $2,460. Future analyses would be more costly, adding the cost of purchasing one or more arthropod reference materials from the National Research Council Canada ($400+) and additional costs for semi-quantitative analyses and added rows of data for matrix spikes.