2.1 Soil collection and preparation
In 2018, non-contaminated background and PHC-contaminated soil was collected from the site of a former petroleum distribution terminal in the boreal region of northern Ontario, Canada. The petroleum distribution terminal was active from 1920 to 1998, where accidental gasoline and diesel releases to soil occurred over the period of operation. Contaminated soil was collected from a single hotspot area where weathering of PHCs occurred for a minimum of 25 years. Background soil was collected from a grassland area located approximately 300 m to the northeast of the contaminated area, where no known industrial activities or contamination had occurred. Background and contaminated areas were sampled to a depth of 0.01–0.6 metres below ground surface. The uppermost topsoil (1 cm) was removed from each area during excavation of the soils to eliminate herbaceous biomass.
At the start of invertebrate toxicity testing, the soils collected from the background and contaminated areas contained, on average, < 100 and 12,800 mg/kg total petroleum hydrocarbons (TPH). The particle size of the dried and sieved (2-mm) site soils were analyzed by ALS Environmental using ASTM D619 and ASTM D422-63 reference methods. The background soil was comprised of 6.2% clay loam (< 0.005 mm), 23% silt (0.005 mm – 0.75 mm), 53% fine sand (0.075 mm – 0.425 mm), 17% medium sand (0.425 mm – 2.0 mm), 1.1% coarse sand (2.0 mm – 4.75 mm), while the contaminated soil was comprised of 4.6% clay, 25% silt, 29% fine sand, 24% medium sand, and 18% coarse sand. As such, both site soils are considered predominantly fine-grained (< 0.005–0.425 mm).
2.2 Test species preparation
Cultures of two springtail species (Folsomia candida and Proisotoma minuta) were received from Environment Canada’s Biological Assessment and Standardization Section. Every two months, new colonies were cultured in polyethylene containers lined with 8:1 Plaster-of-Paris and activated charcoal substrate. Colonies were incubated in a Conviron ATC60 growth chamber at 20 ± 2 oC and a photoperiod regime of 16:8 hours (light:dark) as recommended by Environment Canada (2014). These conditions are similar to the average five-year summertime temperature (17 oC) and daily photoperiod (16 hrs) at the site in northern Ontario (Environment Canada 2023). The colonies were aired bi-weekly to avoid fungal growth and fed activated dry yeast (Saccharomyces cerevisiae) as the sole food source. Colony vessels were gently sprayed with deionized water bi-weekly to moisten the substrate (Environment Canada 2014; OECD 2016).
F. candida and P. minuta were age-synchronized prior to conducting reference toxicant and definitive survival and reproduction tests. To produce same-age springtails, unhatched eggs were collected from colonies using a sterile paintbrush moistened with deionized water. Eggs were placed onto 1-cm diameter squares of moistened filter paper which were spread across fresh substrate-lined 100 mm x 15 mm Petri dishes. The filter paper squares were removed after hatching: 24 hrs for P. minuta and 72 hrs for F. candida (Environment Canada 2014). Juveniles remained in each Petri dish where they were monitored and fed deionized water and 2–3 grains of activated dry yeast daily to ensure health until testing. During both reference toxicant and definitive tests, F. candida were age-synchronized to 10–12 days and P. minuta to 14 days old as per Environment Canada (2014) recommendations. Ten age-synchronized springtails (5 male: 5 female P. minuta or 10 female F. candida) were added to each test vessel to reduce test variability.
3.3 Reference toxicant tests
Fourteen-day survival and reproduction tests with the reference toxicant, boric acid, were conducted to ensure test validity and high culture performance at the time of definitive tests. Artificial soil (70% rinsed and dried silica-based sand, 20% Edgar’s Plastic Kaolin clay, and 10% Sphagnum peat moss) was created and neutralized to pH 6.0-7.5 using reagent-grade CaCO₃ to the standards of Environment Canada (2014) and the OECD Test no. 232 (2016). Juveniles were gently transported to prepared test vessels through use of a low-suction water-based vacuum aspiration system (Pang et al. 2023a).
The soil was then moistened with deionized water to 70% of its soil water holding capacity (SWHC). Boric acid was selected as the reference toxicant due to its prevalence, bioavailability in soil, and reproducible toxicity results (Amorim et al. 2012; Princz et al. 2017; Pang et al. 2023b). F. candida test vessels were individually spiked with 0, 190, 270, 370, 520, 1120, 1430, and 2000 mg/kg boric acid while P. minuta vessels were spiked with the same concentrations until a maximum of 1120 mg/kg boric acid as complete mortality was consistently observed at this concentration during preliminary tests.
Four replicates per concentration were employed for experimental robustness. Each 125-mL (7 cm diameter x 6.4 cm height) test vessel was filled with 30-g of artificial soil. Parafilm was used to seal the P. minuta test vessels in place of lids to prevent escape while allowing sufficient gas exchange. Test vessels for F. candida were sealed with metal lids as this species is less sensitive to elevated soil carbon dioxide levels than other springtails (Fountain and Hopkin 2001, 2005). All test vessels were briefly opened for air exchange, fed activated dry yeast, and gently sprayed with deionized water as needed to maintain soil moisture on a biweekly basis. The temperature (20 ± 2 oC) and light regime (16:8 light:dark) used for culturing colonies was maintained.
At the end of the experiments, test vessels were slowly filled with deionized water and placed over ice baths for 10 minutes to stop movement of the springtails and enable counting under a Leica Wild M3Z Stereozoom Microscope with 10x/21 mm adjustable lenses and digital imaging software (Pang et al. 2023a). The number of adults informed on survival rates of the species while the number of progenies indicated the rates of reproduction. Reference tests were considered valid as there was 80% (F. candida) and 70% (P. minuta) adult survival observed in the non-spiked artificial soil (Environment Canada 2014).
2.4 Definitive survival and reproduction tests
The background and PHC-contaminated site soils were air-dried for ≥ 72 hrs and then crushed using a metal shovel in preparation for definitive survival and reproduction tests. Large rocks were removed by hand. Subsequently, soils were sieved through a 2-mm mesh and then the homogenized through 35 cycles of riffle splitting to create single mixtures of background and contaminated site soils. The soils were then homogenized to single multi-concentration mixtures of 25%, 50%, and 75% relative contamination compared to the background (0% contaminated) and contaminated (100% contaminated) site soils. Test soils are referred to throughout by their relative content of contaminated soil; for example, the 25% contaminated soil is a mix of 25%:75% contaminated soil to background soil. Soils were then mixed with deionized water to 70% of their SWHC.
Age-synchronized F. candida and P. minuta were used to carry out definitive survival and reproduction tests. Definitive tests for F. candida and P. minuta lasted 28 and 21 days, respectively (Environment Canada 2014). Six replicates, each containing 30-g of soil, were used per soil concentration as recommended for definitive testing using site soils (Environment Canada 2014). The multi-concentration soils were subject to the same moisture regime as the reference toxicant tests. Test conditions, including the ambient temperature, daily photoperiod, and procedure used for measuring survival and reproduction, were also consistent with reference tests. Adult survival and levels of live progeny during the definitive and reference toxicant tests were compared to Environment Canada’s (2014) standard test validity criteria. The tests were valid, with > 100 live progenies and adult survival rates of > 70% (F. candida) and > 60% (P. minuta) observed in the 0% contaminated soil (Environment Canada 2014).
2.5 Soil chemical analyses
Plant nutrient levels in the artificial and site soils were reported by Roy et al. (2023). The background and PHC-contaminated site soils share similar nutrient properties except for seven times higher potassium in the contaminated site soil. Concentrations of metals were quantified in the background and contaminated site soils by Maxxam Analytics (now Bureau Veritas) using US EPA 6020B reference method. Neither site soils were contaminated with metals (OME 2011; US EPA 2014a).
Soil boron concentrations were verified at the start of each reference toxicant test for all soil concentrations at the CALA-accredited Queen’s University Analytical Services Unit (ASU). Boron quantification followed the United States Environmental Protection Agency reference method 200.7 (US EPA 1994). One duplicate sample was analyzed per sample run. Boron concentrations were subsequently converted into units of boric acid (Environment Canada 2014).
Total Petroleum Hydrocarbon (TPH) values presented in this study are the sum of PHC concentrations in the CCME F2, F3, and F4 fractions. The concentration of these fractions in the definitive test soils were verified at the ASU at the beginning of the definitive survival and reproduction tests (CCME 2008). PHC F1 and BTEX (benzene, toluene, ethylbenzene, xylene) concentrations were not measured in the test soils as they were below detection limits (< 10 mg/kg) in the 100% contaminated soil when measured during preliminary analyses (Roy et al. 2023). The PHC F2 to F4 extraction process involved using 1:1 hexane/acetone extraction with three cycles of 20-minutes of ultrasonication. Samples were concentrated on a Büchi rotovapor R-114, and then mixed with activated silica gel and 1:1 hexane: dichloromethane to remove polar and biogenic compounds. Analytes were subsequently quantified via gas-chromatography with flame ionization detection (Agilent GC 7890A). A minimum of one method blank, one duplicate and a 5,000 mg/kg diesel fuel quality control spike were extracted and analyzed during each sample run.
Sixteen unbranched polycyclic aromatic hydrocarbon (PAH) concentrations were measured in the site soils using US EPA Method 8270d (2014b), as reported by Roy et al. (2023). The background site soil contained < 40 ng/g of any single PAH compound while the contaminated site soil was concentrated in six PAHs (ng/g): naphthalene (1,280), acenaphthene (339), fluorene (1,580), phenanthrene (2,090), anthracene (157), and fluoranthene (775). Excluding naphthalene (CCME PHC F1), five of these PAHs constitute CCME PHC F2s (CCME 2001).
2.6 Data and statistical analysis
Mean species survival and reproduction endpoints were measured during reference toxicant and definitive tests. Using R 4.2.0, normality of residuals was assessed through Shapiro-Wilk Tests. Bartlett and Levene’s Tests were carried out to calculate homogeneity of variances. Asymptotic Fisher-Pitman Permutation Tests (α = 0.05) were conducted using the “coin” package for significance testing between springtail species. Rosenthal’s Coefficients (r) were calculated as a non-parametric measures of effect size (Rosenthal 1991).
For visualizing the data, soil constituent (i.e., boric acid or TPH) concentration versus number of individuals was graphed using the “ggpubr” package in R 4.2.0. Kruskal-Wallis Tests and Wilcoxon tests with Bonferroni adjusted p-values to assess significant differences in survival and reproduction. Lethal concentrations (LCp), the percent (p) reduction in survival of the springtails, were used to model four parameter (minimum, maximum, LC50, slope) dose response curves (DRCs) for mortality as recommended by Environment Canada (2005), using “tidyverse”, “broom”, “drc”, “modelr”, “egg”, “grid”, “sandwich”, and “lmtest” packages. Similarly, inhibitory concentrations (ICp), the percent (p) reduction in reproduction of the springtails, were used to model DRCs for progenies formation. Logarithmic (log10) concentrations were converted into arithmetic values of expected concentrations. Model selection for log-logistic and Weibull curves were based on calculation lowest Aikake’s Information Criterion (AIC). Hill slopes (nH) were calculated for each DRC through “Non-linear regression (curve fit)” in Graphpad Prism 8.4.3. The TPH concentrations associated with LC25/50 and IC25/50 were interpolated from DRCs modelled in the software, as described by Roy et al. (2023).