Environ Sci Pollut Res

DOI 10.1007/s11356-017-0177-x

RESEARCH ARTICLE

Framework for assessment and phytoremediation

of asbestos-contaminated sites
Cédric Gonneau 1 & Kinsey Miller 1 & Sanjay K. Mohanty 2,3 & Rengyi Xu 4 &
Wei-Ting Hwang 4 & Jane K. Willenbring 2,5 & Brenda B. Casper 1

Received: 15 June 2017 / Accepted: 11 September 2017

# Springer-Verlag GmbH Germany 2017

Abstract We examine the feasibility of phytoremediation as of concern. Crop cultivar survived on asbestos-contaminated
an alternative strategy to limit the exposure of asbestos in site soil. Grasses from serpentine ecotype did not show higher
with asbestos-containing materials. We collected soils from biomass than the cultivar. Overall, these results showed that
four locations from two sites—one with naturally occurring soil conditions play a critical role in screening different crop
asbestos, and another, a superfund site, where asbestos- species for phytoremediation and that asbestos concentration
containing materials were disposed over decades—and per- has limited to no effect on plant growth. Our study provided a
formed ecotoxicology tests. We also performed two experi- framework for phytoremediation of asbestos-contaminated
ments with crop cultivar and two grasses from serpentine eco- sites to limit long-term asbestos exposure.
type and c ultivar to determined best choice for
phytoremediation. Asbestos concentrations in different size Keywords Asbestos . ASTM method . Crop cultivar . Heavy
fractions of soils varied by orders of magnitude. However, metal . Native grass . Phytostabilization . Size fraction
different asbestos concentrations had little effect on germina-
tion and root growth. Presence of co-contaminants such as
heavy metals and lack of nutrients affected plant growth to Introduction
different extents, indicating that several of these limiting fac-
tors should be considered instead of the primary contaminant Asbestos is a group of six naturally occurring fibrous min-
erals, whose properties include resistance to heat or fire and
Responsible editor: Elena Maestri high tensile strength (Schreier 1989; Dodson and Hammar
Electronic supplementary material The online version of this article
2011; Kumar et al. 2016). Because of these useful properties,
(https://doi.org/10.1007/s11356-017-0177-x) contains supplementary asbestos has been mined and once used extensively in many
material, which is available to authorized users. commercial applications, including the construction and auto-
mobile industries (Morrison and Murphy 2010). Despite re-
* Cédric Gonneau strictions on asbestos use in recent decades in the USA, nearly
[email protected] 2.5 million metric tons of asbestos are produced annually
worldwide, and asbestos products are still widely used in
1
Department of Biology, University of Pennsylvania, India and China—home to more than one third of the world’s
Philadelphia, PA 19104, USA population (Virta 2006). Mining of asbestos minerals and the
2
Department of Earth and Environmental Science, University of production of asbestos-containing materials generate waste
Pennsylvania, Philadelphia, PA 19104, USA that itself can pose serious health hazards (Van Gosen
3
Department of Civil and Environmental Engineering, University of 2007a; Boulanger et al. 2014). Exposure to asbestos fibers
California, Los Angeles, Los Angeles, CA 90095, USA can cause asbestosis, mesothelioma, and lung and stomach
4
Department of Biostatistics and Epidemiology, University of cancer (Cunningham and Pontefract 1971; Institute of
Pennsylvania, Philadelphia, PA 19104, USA Medicine 2006; Fortunato and Rushton 2015). Therefore, it
5
Geosciences Research Division, Scripps Institution of is critical to assess the extent of asbestos contamination and
Oceanography, University of California, San Diego, CA 92093, USA improve remediation design to minimize asbestos exposure.
 Environ Sci Pollut Res

Asbestos fibers are found at two main types of sites, where Brownfield sites or chromite mines (Kumar and Maiti 2015),
it naturally occurs (referred to as naturally occurring asbestos its utility to treat asbestos-contaminated sites has not been
or NOA) and in asbestos-containing material (ACM) from examined to date. Plants are known to stabilize the topsoil
waste disposal (Schreier 1989). NOA refers to asbestiform and minimize erosion (Alkorta et al. 2010; Brown and
minerals occurring within rocks or soils that can be released Chaney 2016), which could limit asbestos exposure via air.
by human activities or weathering processes. NOA is mostly Furthermore, recent laboratory studies show that organic acids
found in ultramafic rock and especially serpentinite (Lee et al. typically released from plant roots or soil microbe can leach
2008). These sites include many serpentine sites with specific out elements, alter surface charge of fibers, and induce a lower
vegetation, serpentinophytes, adapted to particular soil prop- toxicity (Daghino et al. 2006; Favero-Longo et al. 2013;
erties. Other NOA localities include mines or quarries for Holmes and Lavkulich 2014; Mohanty et al. 2017). Thus,
heavy metals or other materials, such as several chromite phytostabilization and phytoremediation, more broadly, could
mines in the Eastern USA (Pearre and Heyl 1960; Van be viable technology for asbestos-contaminated sites.
Gosen 2007a) and a vermiculite mine contaminated with am- The success of this Bgreen^ strategy depends on the surviv-
phibole asbestos in Libby, MT (Bandli and Gunter 2006). al and performance of plants in the presence of asbestos—a
Vegetation may be absent or reduced in some mining areas basic premise that has not been tested systematically. It seems
(Meyer 1980), in part due to the presence of other pollutants important to adjust soil properties if necessary and/or choose
such as Ni and Cr (Levitan et al. 2015). Indeed, asbestos the best plants as phytoremediation agents. Among different
pollution is found in many sites with asbestos and heavy metal phytoremediation strategies (Ali et al. 2013), assisted
gradients. The disposal sites where ACM is found, on the phytostabilization with soil amendments and planting of crop
other hand, do not necessarily share common geological cultivars seems an alternative choice (Trivedi and Ahmad
properties. 2011). In chromite asbestos mines in India, several studies
Determining the concentration and form of asbestos in con- showed that using suitable plants with metal low shoot
taminated soil or waste material is the necessary first step to metal uptake, such as grasses or legumes, could act as a
designing the remediation plan as it helps in risk assessment potential barrier for metal transport in food chain (Trivedi
and risk based decisions for brownfields and Superfund sites and Ahmad 2011; Kumar and Maiti 2015; Kumar et al.
(Wroble et al. 2017). Based on the Framework for 2017). In a Vermont asbestos site, biomass production of
Investigating Asbestos-Contaminated Superfund Sites (US grass and clover with two different compost mixtures
EPA), asbestos concentration in contaminated soil, storage showed higher compared to the control (Chaney et al.
piles, and waste materials is estimated by the California Air 2011). Serpentinophytes species from ultramafic substrates
Resource Board (CARB) Method 435. This method involves may represent another alternative due to their tolerance for
grinding a representative bulk soil sample and quantifying the high concentrations of heavy metals, such as Ni and Cr, and
amount of asbestos fiber in the ground sample using polarized low levels of essential nutrients, namely nitrogen, phospho-
light microscopy (PLM). Consequently, this method does not rus, potassium, and calcium (Brady et al. 2005).
account for the distribution of asbestos based on the grain size In the USA, 1312 sites are contaminated with either NOA
of contaminated soil or waste materials. Because the size of or ACM (Van Gosen 2005, 2007b, 2008). In some of these
asbestos fibers controls their exposure pathways (Boulanger sites, including the Ambler, PA Superfund site disposal waste
et al. 2014), it is critical to develop a screening method that site studied here, asbestos is the primary contaminant of con-
accounts for the fractionation of asbestos based on the particle cern. For these sites, ex situ treatment is typically not preferred
size of contaminated soil or waste materials. To overcome this because asbestos fibers may be emitted when large quantities
limitation, the ASTM D7521-13 method has been proposed, of contaminated material must be relocated (Paik et al. 1983;
which requires measuring asbestos concentrations in different Brown 1987). For in situ treatment, EPA recommends capping
size fractions of contaminated soil or waste (D22 Committee with uncontaminated soil at least 2 m thick, which could be
2013). Although EPA is evaluating this sampling method for expensive (Lee and Jones-Lee 1997; Millano 1998).
its applicability to Superfund site characterization (US EPA), a Brownfield sites or other uncategorized asbestos-
comparison of the two methods for assessing asbestos con- contaminated sites are typically left untreated because alterna-
tamination is lacking. tive, cost-effective remediation methods are lacking, despite
Most asbestos mines or sites contaminated with asbestos the possibility of significant health risk associated with long-
materials have low vegetation cover and/or diversity, indicat- term asbestos exposure.
ing low colonization of plants from surrounding vegetated Here, we compared two methods to test asbestos con-
areas. In this case, assisted phytostabilization with soil amend- tamination in soils and examined the feasibility of
ments could be a cost-effective solution (Brown and Chaney phytoremediation to treat asbestos-contaminated sites. We hy-
2016). Although phytostabilization has been used to treat soils pothesized that asbestos itself would not pose any risk to
containing heavy metals, such as cadmium or lead, in many plants that are typically utilized for phytoremediation. To test
Environ Sci Pollut Res

this hypothesis, we collected soils from two locations at each the EPA framework to assess asbestos contamination in rocks
of two known asbestos-contaminated sites in Pennsylvania, and soils including serpentine aggregate storage piles (US
the Superfund site with ACM and a site with NOA. We com- EPA). For this analysis, the bulk sample was crushed in a mill
pared the asbestos concentration as determined using the cur- (Mill 8000M, SPEX, USA) and sieved to retain only the par-
rent standard method (CARB 435) and an alternative method ticle size < 75 μm. The ground samples were analyzed for
under consideration by EPA (ASTM D7521-13). To examine asbestos under polarized light microscopy based on a 400-
toxicity of asbestos for plants, we conducted an ecotoxicology point count technique with a detection limit of 0.25% (or 1
test using three species commonly employed for count). In contrast to the CARB 435 method that uses only the
phytoremediation. We examined germination and root growth finely crushed part of the bulk sample, the ASTM D7521-13
for each of the species. Finally, we performed a large screen- method requires analysis of asbestos in different grain size
ing of plants and soil microbes suitable for phytoremediation fractions. Briefly, for the latter, soil samples were dried and
of the Superfund site in two complementary experiments, with sieved to separate factions according to the following size
(i) commercial crop species classically used for heavy metal ranges: > 19 mm, 2–19 mm (coarse), 0.106–2 mm (medium),
remediation and (ii) two grasses, including one serpentine and < 0.106 mm (fine). Masses for each fraction are recorded,
ecotype of each species. We also examined whether soil mi- and the presence of asbestos was measured first by
crobial inoculum collected at the serpentine site facilitated stereomicroscopy, followed by PLM, to determine whether
plant growth in soils from the Superfund site. In addition to fibers were observed in matrices or as isolated material. If
assessing plant growth, we also measured elemental concen- asbestos was not detected by the PLM results in any fraction,
trations in aboveground tissues to detect any major nutrient then only the fine fraction of the sample was re-analyzed for
deficiency or heavy metal toxicity. detection of asbestos fibers using transmission electron mi-
croscopy (TEM), a more precise technique.

Materials and methods

Soil characterization
Soil sampling
For soil characterization, air-dried soil samples were sieved to
In spring 2015, we collected soil samples from two sites. One obtain soil with particle size < 2.0 mm. Soil pH was measured
is the BoRit Superfund site (Ambler, PA; 40° 09′ 18.7″ N 75° by making a slurry consisting of a 1:5 ratio of soil to ultrapure
13′ 49.6″ W), where ACM wasdiscarded from a nearby water. Phosphorus was estimated by the Bray-1 method (Bray
manufacturing plant over several decades. The other, located and Kurtz 1945). Briefly, 1.0 g of soil was suspended in
in Nottingham Park (Chester County, PA, 39° 44′ 8.53″ N 76° 10.0 mL solution containing 0.025 M HCl and 0.03 M
2′ 9.94″ W), is a serpentine site, containing NOA in the NH4F and shaken vigorously for 1 min before centrifuging
serpentinite bedrock (Smith and Barnes 1998). The site has a at 6000 rpm for 5 min. The supernatant was transferred to a
long history of mining for the extraction of sodium feldspar, spectrophotometer cuvette and analyzed at 880 nm (Hach, DR
chromium, and building stone (Lookingbill et al. 2007). The 6000, USA). Pseudo-total concentrations of major elements
area was once the world’s leading source of chromite (Pearre (Al, Ca, Fe, K, and Mg) and trace elements (Cd, Co, Cr, Cu,
and Heyl 1960). We collected soil with asbestos-containing Mn, Ni, Pb, and Zn) were determined by digesting 0.5 g of
materials from two locations at BoRit: from what had been a ground soil (< 200 μm) in an aqua regia solution: mixture of
reservoir before the removal of water and along the banks of a 10 mL HNO3 65%, 5 mL H2O2 30%, and 5 mL HCl 37% in a
stream. We collected bulk soil and rock also from two loca- DigiPrep system (EPA Method 5030).
tions at Nottingham: a grassland area with native C4 grasses as To determine fractionation of elements based on their asso-
the main constituents of the plant community (Gonneau et al. ciation with different types of minerals or sorption sites, we used
2017) and at the waste pile of a former chromite mine. Soil a three-step sequential extraction procedure developed by the
samples were collected in zip-lock plastic bags, and a well- Commission of the European Communities Bureau of
mixed portion of soil was stored at 4 °C for mineralogical Reference (BCR; Clevenger 1990). Elemental concentrations
studies whereas another portion was air-dried for 48 h. The in the extracted solutions at different steps were determined
latter was used to characterize soil elemental content and for using Inductive Coupled Plasma Atomic Emission
performing an ecotoxicology test. Spectroscopy (ICP-AES, Spectro Genesis, Spectro Analytical
Instruments, Germany).
Asbestos quantification In order to evaluate fertility and potential limiting factors
for phytoremediation, critical values were assigned to each of
We measured asbestos concentrations using two methods: 16 physical or chemical soil parameters: ten parameters were
CARB 435 and ASTM D7521-13. CARB 435 is included in based on the method proposed by the Fertility Capability
 Environ Sci Pollut Res

Classification (Sanchez et al. 2003), and six parameters were greenhouse and watered daily. After 12 weeks, the above-
the main phytotoxic heavy metals (Nagajyoti et al. 2010). ground portions were harvested, cleaned with tap water, and
dried for 48 h at 60 °C before weighing. To determine ele-
Ecotoxicology test mental concentrations, a portion of the dried shoot was
ground to small pieces and digested in 1 N HNO3 for 3 h
To measure potential toxicity induced by different soils on at 95 °C in a DigiPREP system (Gonneau et al. 2014).
plants (Baran and Tarnawski 2013), we employed the Digested samples were diluted with ultrapure water, filtered
Phytotoxkit (MicroBioTest, Belgium). The kit includes three with 0.45 μ membrane, and analyzed for elemental concen-
plant species—Sorghum saccharatum (S. saccharatum, trations using ICP-AES.
Poaceae), Lepidium sativum (L. sativum), and Sinapis alba
(S. alba, Brassicaceae)—and based on plant performance al- Serpentine native soil inoculum experiment with C4 grasses
lows assessment of toxicity of soil compared to an internation-
al standard soil (control). We considered two main parameters A separate experiment was conducted to determine if soil
of performance that are relevant early in the plant life cycle, microbes naturally occurring at the Nottingham Serpentine
germination and root growth. The ecotoxicology test was con- site could facilitate growth of two C4 grasses in the BoRit
ducted in accordance with the procedure recommended by the asbestos-contaminated soils. Both serpentine ecotypes and
manufacturer. The percent inhibition of seed germination (IG) commercial cultivars of S. nutans and A. gerardii, Holt and
and inhibition of root growth (IR) for each soil were calculated Niagara, respectively, were grown. Seeds and soil to be used
with the formula: IG or IR = [(A − B)/A] × 100, where A and B as inoculum were collected in Fall 2015 at Nottingham,
are the mean seed germination or root length in the control and Chester County, PA. Inoculum soils were collected under five
tested soil, respectively. individuals of the two grasses, brought back to lab in a cooler
and stored at 4 °C. Soils were separated into two parts. One
Germination tests of crop cultivars on BoRit soil part was maintained as fresh inoculum (live) and other part
was autoclaved 1 h at 121 °C (sterile) and used as a control for
Seed germination in the two localities from BoRit was also the abiotic soil fraction in the inoculum. Contaminated soils
tested for cultivars of eight crops from belonging either to the from each of the two BoRit locations, reservoir and stream
family Poaceae—Andropogon gerardii (A. gerardii), Lolium bank, were placed in Conetainers™ to approximately 20 cm
perenne (L. perenne), Panicum virgatum (P. virgatum), and depth and covered with 5 cm of live or sterile inoculum as it
Sorghastrum nutans (S. nutans)—or Brassicaceae: Brassica might be in the process of remediating a contaminated site.
juncea (B. juncea), Brassica oleracea (B. oleracea), Seeds were germinated and grown in sterilized sand and seed-
L. sativum, and S. alba. Seeds were obtained from Sheffield’s lings transplanted at 2 weeks of age. There were six replicates
Seed Co., Inc. or from Johnny’s Selected Seeds (Table S1). of each soil × seed source × inoculum type for each plant
Seeds were soaked in 70% ethanol for 15 min and washed twice species. Aboveground portions were harvested after 15 weeks,
with deionized water. Thirty seeds per species were placed on dried, and weighed, and the shoots analyzed for elemental
soil in a square petri dish (10 cm) containing either 60 g of test concentrations as described above.
soil or control, a mixture of sand and compost (v/v 1:1). Petri
dishes were watered every day and placed in an incubator at Statistical analysis
25 °C. The number of germinated seeds was recorded daily for
10 days after sowing. Each species × soil type combination was All statistical analyses were performed using R v3.2.5. For
replicated three times. soil properties, differences between the four soil locations in
soil properties were analyzed by the Kruskal-Wallis test (a
Experimental design for plant growth non-parametric test). For the first seed germination with
S. saccharatum, S. alba, and L. sativum, the difference in
Screening crop cultivars germination and root growth inhibition relative to the control
was analyzed by Kruskal-Wallis test. For the second seed
Six seedlings of five crop species and the three species from germination test, we analyzed the proportion of germinated
the ecotoxicological test, above (Table S1), were grown in the seeds (out of 30 seeds planted) on day 9 using a series of
two soils from the BoRit site (Reservoir and Stream Bank) and logistic regression models. The three levels of substrate,
their growth compared to that in a control soil, a mixture of namely BoRit stream bank, BoRit reservoir, and control, were
compost and sand. Tube-shaped Conetainer™ were filled to treated as fixed effects and the eight species as a nested factor
approximately 16 cm with soil and capped with 2 cm of com- within the soil substrate factor. Eight species were further
post to minimize entrainment of asbestos fibers in air and combined into two species categories for comparison in a
allow better plant growth. Plants were maintained in the separate analysis: the Poaceae group with A. gerardii,
Environ Sci Pollut Res

S. nutans, L. perenne, and P. virgatum (group 1) and the BoRit site contained a higher concentration of asbestos than
Brassicaceae group with B. juncea, B. oleracea, L. sativum, the Nottingham site. All soil fractions at the grassland at
and S. alba (group 2). Log odds of germination (i.e., log of Nottingham were similar in asbestos concentration (2.0%).
probability of a seed is germinated over the probability not The asbestos concentration in the waste pile at the chromite
germinated) for each substrate and species combination were mine location at Nottingham was below the detection limit
estimated. The significance of the substrate and species factors (0.25%). In the chromite mine location, some minerals, such
was tested sequentially by comparing alternative models using as chromite, lizardite, and dolomite (Smith and Barnes 1998),
a deviance test. Finally, differences in growth and elemental were present that are rare or absent in most soils. Lizardite is a
concentrations were analyzed in the cultivar experiment by a non-fibrous mineral, which belongs to the serpentine-group
one-way ANOVA and in the inoculum experiments by three- with a composition similar to chrysotile.
way ANOVA including seed source, inoculum and soil. Our results indicate that the current framework that evalu-
ates the level of soil asbestos using only the CARB method
likely underestimates the contribution of medium and coarse
soil fractions to overall asbestos contamination. Because as-
Results and discussion bestos fibers are more likely to be airborne when they are
present as fine particles, the presence of asbestos in medium
Size-dependent asbestos concentration or coarse fractions could limit their mobility or exposure route
in the environment. In these sites, exposure to asbestos can be
Our results showed that the ASTM method provides better minimized by implementing a remediation plan that lowers
insight than the CARB 435 method regarding asbestos con- soil erosion or abrasion of asbestos-containing material. This
centration and distribution among different soil particle size result also gives credence to the recent EPA initiative to eval-
fractions. The results from the ASTM method showed that the uate the utility of the ASTM method in Superfund site.
coarse and medium fractions of soil from BoRit contained
orders of magnitude higher concentrations of asbestos (10–
12 and 6–8%, respectively) than the fine fraction of soil Other soil properties and fertility
(< 1%), while the concentration in the fine fraction alone
matched the asbestos concentration measured by the CARB Soil properties and fertility that affect plant growth are impor-
method (Table 1). In contrast to the ASTM method, the CARB tant factors that can determine the types of vegetation and
method resulted in an asbestos percentage consistently lower possibly soil amendments needed to implement a
than 1% in both BoRit locations and in both locations in the phytoremediation strategy at a contaminated site. Soil physi-
grassland in Nottingham (Table 1). Among these three loca- cal and chemical properties varied mostly not only between
tions, soil from the stream bank (0.75%) contained slightly sites but also between locations within a site (Table 2). Soil pH
more asbestos than the soil from the reservoir (0.5%). was near neutral for the Grassland (6.72) at Nottingham,
Microscopic analysis indicates that the asbestos form dif- whereas pH was alkaline for the Stream Bank (8.13) at
fered between the BoRit site sampling locations and the grass- BoRit. Elemental concentrations of Co, Cr, Mn, and Ni were
land at Nottingham. Both locations at BoRit contained chrys- higher in Nottingham than in BoRit whereas P and Ca/Mg
otile whereas the grassland soil at Nottingham contained an- ratio were higher in BoRit. The Ca/Mg ratio, a major indicator
thophyllite (amphibole). Based on the ASTM method, the of soil fertility in serpentine soils (Brady et al. 2005), was

Table 1 Asbestos form and size distribution in two locations at the BoRit-contaminated site and Nottingham serpentine soils

Method CARB 435 ASTM method D7521-13

PLM PLM non-asbestos PLM asbestos PLM asbestos concentration PLM TEM
by sieve fraction weighed

Site Locations Type % fibrous % non-fibrous % Coarse Medium Fine Bulk (%) Fine

BoRit Reservoir Chrysotile 0 99.5 0.5 12 8 0.75 8.2

Stream Bank Chrysotile 0 99.25 0.75 10 6 0.25 7
Nottingham Grassland Anthophyllite < 1% 98.5 0.5 2 2 0.5 1.9 Chrysotile
Chromite Mine nd 0 100 BDL nd nd nd nd

Coarse fraction (> 2 mm), medium [100 μm–2 mm], and fine (< 106 μm)
BDL below detection limit (0.25), nd non-detected, PLM polarized light microscopy, TEM transmission electron microscopy
 Environ Sci Pollut Res

Table 2 Physical and chemical

characteristics of soils collected in BoRit Nottingham
two locations at the BoRit-
contaminated site and Parameters Unit Stream bank Reservoir Grassland Chromite mine
Nottingham serpentine soils Sand % 62.5 66.9 48.1 67.7
Clay % 16.6 6.27 19.1 13.2
Silt % 20.9 26.7 32.8 19.1
Texture USDA Sandy loam Sandy loam Loam Sandy loam
pH 8.13 a 7.63 b 6.72 c 7.56 b
2+ + −1
Ca cmol kg 5.90 a 5.14 a 2.73 a 0.912 b
Mg2+ cmol+ kg−1 1.9 a 1.8 a 6.7 b 4.1 ab
Ca/Mg 8.24a 10.2a 0.3b 0.12b
K+ cmol+ kg−1 0.122 a 0.136 a 0.317 b 0.149 a
Na+ cmol+ kg−1 0.284 a 0.534 b 0.204 a 0.112 a
Al3+ cmol+ kg−1 0.004 a 0.004 a 0.004 a 0.012 a
Fe3+ cmol+ kg−1 0.002 a 0.002 a 0.003 a 0.004 a
CEC cmol+ kg−1 10.1 a 10.0 a 25.6 b 5.60 a
Cd mg kg−1 <1 <1 <1 <1
Co mg kg−1 0.848 a 0.120 a 79.4 b 15.5 b
Cr mg kg−1 9.14 a 9.30 a 284 b 69.9 b
Cu mg kg−1 31.5 a 23.4 a 25.3 a 15.5 a
Mn mg kg−1 211 a 117 a 984 b 197 b
Ni mg kg−1 12.3 a 10.6 a 1284 b 266 b
P mg kg−1 17.4 a 33.7 b 3.54 c 2.45 c
Pb mg kg−1 <1 <1 <1 <1
Tl mg kg−1 <1 <1 <1 <1
Zn mg kg−1 40.8 a 40.8 a 57.2 a 57.2 a

Different lowercase letters represent significant difference at p < 0.05

above 1.0 in BoRit and below 1.0 at Nottingham. These dif- soil pH (Fig. S1). The presence of asbestos in soil induces
ferences in Ca/Mg ratio were attributed to a difference in ex- high pH values, which represent the main limitation for plant
changeable Ca and Mg between BoRit and Nottingham sites. growth. At asbestos mine tailing sites, for example, the pH is
The phosphorus concentration in soil from the reservoir at mostly alkaline (pH ~ 10). High pH is a limiting factor be-
BoRit was higher (33.4 mg kg−1) than at the two locations at cause it makes elements less mobile and thus lowers nutrient
Nottingham and in the stream bank at BoRit (all below availability for plants (Meyer 1980). In BoRit, the pH is within
20 mg kg−1). Concentrations of heavy metals in BoRit are in or close to the range of values recommended for crop growth
the range of values typically found in soils in the contermi- (pH 6.5–8; USDA), indicating that plants should grow well in
nous USA (Smith et al. 2013). Indeed, the BoRit site showed this soil. In contrast to BoRit, the Nottingham soil has a greater
lower concentrations than the threshold value for a metallifer- number of limiting factors (Table S2).
ous site (Burt et al. 2003).
Among 16 parameters considered, limiting soil factors, as Ecotoxicology tests
determined from Fertility Capability Classification, varied be-
tween the four locations. Limiting factors include the follow- By monitoring seed germination and early-stage root growth
ing: (1) percentage of gravel in the stream bank sample at in three common species used for phytoremediation, we
BoRit, (2) pH for both sites at BoRit and the chromite mine were able to evaluate limiting factors and fertility described
at Nottingham, (3) nutrient reserves (K) at the chromite mine above (Fig. 1). In all the soils from four localities, seed ger-
at Nottingham, and (4) lower than threshold P (P-Bray) con- mination for both Brassicaceae species was inhibited by 0 to
centrations and higher than threshold Ni and Cr total concen- 15% compared to control soil (Fig. 1a) while germination for
trations for both locations at Nottingham (Table S2). the S. saccharatum species (Poaceae) was inhibited to a
Nevertheless, the Ni and Cr exchangeable fractions, an indi- greater extent: 15 to 35%. Compared to seed germination
cator for their bioavailability, were low, mainly due to high inhibition, root growth inhibition varied to a greater extent
Environ Sci Pollut Res

Germination test

Germination of eight species, collectively, was not affected by

the BoRit soil compared to the control (Fig. 2). The logistic
regression model with only the substrate as a factor sug-
gested no difference across the three substrates (p val-
ue = 0.66). The model with both the species and substrate
factors indicated that the proportion of germination at day
9 differed across the species within each substrate (p val-
ue = 0.01). When eight species were combined into two
plant families, seeds in the Brassicaceae had a significantly
higher odds (probability) of germination than seeds in the
Poaceae (p < 0.01; Fig. S2). Seed germination is an impor-
tant stage since new seedlings represent first stage of a
remediation strategy (Kranner and Colville 2011), and seed
dormancy can interfere (Finch-Savage and Leubner-
Metzger 2006). Here, any effect of soil properties affects
seed germination. Many species in the Brassicaceae are
known to tolerate high levels of heavy metals (Baker
1987; Krämer 2010), although it would not be a factor in
our study because heavy metal concentrations in BoRit are
low (Table 2). For Poaceae, the germination level was low-
er than 50% after 9 days. Overall, these results suggest that
Fig. 1 Percentage of a germination and b root growth inhibition relative all four of the tested species Brassicaceae can be used for
to the control for three species in different substrates: black, stream bank phytoremediation of soils contaminated with high asbestos
at Ambler; gray, reservoir at Ambler; olive, grassland at Nottingham; and low metal content. In Poaceae, however, it may be
white, chromite mine at Nottingham. Zero indicated no inhibition for necessary to determine the best conditions for germination
Lepidium sativum in reservoir and for Sinapis alba in chromite mine soil
and sowing. That is, seeds may need stratification or chem-
ical scarification (Clarke and French 2005; Finch-Savage
and Leubner-Metzger 2006).
between soil types and plant species, and the variability was
more pronounced for S. saccharatum. For the two Growth and elemental concentrations of cultivar
Brassicaceae species, root growth inhibition was less appar- in inoculum experiment
ent in BoRit compared to Nottingham. Especially, chromite
mine soil inhibited root growth more than any other soil for Among the eight species tested (Fig. 3), three species,
L. sativum and S. alba. Root growth inhibition of namely B. oleracea, A. gerardii, and Sorghum bicolor,
S. saccharatum was similar in all four soils compared to showed significantly different biomass among the three
controls (Fig. 1b). soils (control and both BoRit locations). Biomass of all
The pattern of germination is not related to the concentra- species was higher in BoRit soil than in the control (com-
tion of asbestos, which suggests that asbestos does not directly post and sand mixture), although the biomass differed be-
pose any threat to plants. Interestingly, BoRit soils only weak- tween the two localities within BoRit (Fig. 3). For instance,
ly affected both germination and root growth of L. sativum biomass of S. bicolor was greater on reservoir soil than on
and S. alba. Indeed, the BoRit site presented a lower number soil from the stream bank. Elemental concentrations varied
of limiting factors and better soil for plant growth compared to mostly with species but also slightly with substrate
Nottingham. This could be partly explained by the slightly (Table S3). For major elements (Ca, K, Mg, P), concentra-
alkaline pH at BoRit, which lowers the solubility of heavy tions were mostly in the sufficient range of concentrations
metals. Between all plants tested here, S. saccharatum appears expected in plant species (Munson 1997). The lowest con-
to be the most sensitive. S. saccharatum was also found to be centrations of Ca were found in some species of Poaceae,
less tolerant than two other species when tested on three high- with Ca concentrations below 2000 mg kg−1 compared to
ly contaminated sediments with heavy metal (Baran and four species of Brassicaceae. The low concentrations of Ca
Tarnawski 2013). Similarly, the BoRit soil did not inhibit root in Poaceae are typical due to their particular cell wall type 2
growth as much as did soil from Nottingham with its higher (Vogel 2008). The concentration of K (threshold = 10,000)
levels of heavy metals. was found to be deficient in five species when grown on
 Environ Sci Pollut Res

Fig. 2 Cumulative germination of eight species on a control soil gerardii; Lp, Lolium perenne; Pv, Panicum virgatum; Sn, Sorghastrum
(compost and sand), b stream bank, and c reservoir soil for 10 days nutans; Bj, Brassica juncea; Bo, Brassica oleracea; Ls, Lepidium
after sowing. Red and black color represent species within Poaceae and sativum; Sa, Sinapis alba
Brassicaceae, respectively. Codes are given in Table S1. Ag, Andropogon

stream bank soil, B. oleracea, L. perenne, S. alba, Plants grown in reservoir soil had higher Mg and P concen-
S. bicolor, and S. saccharatum, and also in reservoir for trations than the plants grown in stream bank soil, and con-
S. bicolor and S. saccharatum. Concentrations of Mg were centrations of these elements were typically higher when
always > 2000 mg kg−1. P deficiency was observed in plants were cultivated with live inoculum (Table S4). Live
A. gerardii and S. saccharatum in all soils, with concentra- inoculum potentially includes not only pathogens but also
tions < 2000 mg kg−1. For micronutrients (Fe, Ni, and Zn), beneficial microorganisms, including arbuscular mycorrhi-
the concentrations of Fe for all species and in three soils zal fungi, which can be indispensable for plant nutrition
were < 100 mg kg−1, which is considered deficient. For all (Smith and Read 2010). Both species are known to be
soils, Zn concentrations were in the range of normal, be- mycotrophic (Casper et al. 2008; Ji et al. 2012).
tween 20 and 300 mg kg − 1 (Kabata-Pendias 2000; Interestingly, for both species, the commercial cultivar
Marschner 2012). The same results were found for Ni with and the native ecotype responded similarly in biomass to
concentrations <15 mg kg−1. Concentrations of other toxic the live inoculum (Table S4). Other elemental concentra-
heavy metals, such as Cd, Co, and Pb, were below their tions showed differences based on the interaction of inocu-
detection limit. Overall, with soil amendments, plant lum with species, soil origin or seed source, i.e., commer-
growth and plant nutrition are favorable without any impor- cial or native. K and Zn differed significantly by interaction
tant ecophysiological impacts of the presence of 10% of species and inoculum with higher concentrations for
asbestos in soil. S. nutans with presence of live inoculum. The K tissue
In the inoculum experiment with C4 grasses, biomass concentration was also higher for S. nutans from
varied between species and there was a significant species Nottingham (species × seeds effect) with higher concentra-
× seed source interaction. Biomass was greater for S. nutans tions in reservoir soil. Finally, concentrations of K and Zn
cv. Holt than the three other species × seed source combi- could be slightly deficient with concentrations lower than
nations (Fig. 4). There was no effect of live inoculum from 10,000 and 20 mg kg−1 respectively (Marschner 2012).
Nottingham on biomass for either species in these soils, but Concentrations of Ca and Ni in plant tissues did not differ
the inoculum did affect elemental uptake by the plants. with inoculum or seed source.

Fig. 3 Shoot biomass of eight

crop species cultivated on control
soil (compost and sand), (white),
stream bank (black), and reservoir
(gray). Letters indicate significant
differences between soils within a
species at p < 0.05 and ns
p > 0.05. Vertical bars represent
standard error
Environ Sci Pollut Res

permeable barrier below the 2 m of soil. Indeed,

phytostabilization could be a viable remediation strategy to
minimize asbestos exposure from contaminated sites (e.g.,
Brownfield sites) that are currently left untreated due to pro-
hibitive cost of other available technologies such as capping.
Phytoremediation as a restoration strategy has several bene-
fits: (1) greening of abandoned sites, (2) stabilization of top-
soil, thereby limiting dispersion of asbestos fibers via air, and
(3) increase in soil fertility when amendments are added with
high metal binding capacity.

Fig. 4 Shoot biomass of A. gerardii and S. nutans on BoRit soil

according to significant Species × seed interaction. Letters indicate Acknowledgements The content is solely the responsibility of the au-
significant difference at p < 0.05 for effect of species × seed interaction. thors and does not necessarily represent the official views of the National
Vertical bars represent standard error. Niagara and Holt are commercial Institutes of Health.
cultivars

Funding information This research was supported by National

Conclusions Institute Environmental Health Sciences of the National Institutes of
Health under award number P42 ES023720 (Penn Superfund Research
Program Center Grant).
Our results provide a framework for testing asbestos contam-
ination and designing phytoremediation plans to limit asbestos
exposure in the affected areas. We showed that, compared to
the conventional method included in EPA framework, the References
ASTM method provides a better estimation of asbestos con-
centration in a range of soil size fractions. More importantly, it Ali H, Khan E, Sajad MA (2013) Phytoremediation of heavy metals—
allowed detection of non-friable asbestos-contaminated mate- concepts and applications. Chemosphere 91:869–881. https://doi.
org/10.1016/j.chemosphere.2013.01.075
rial potentially not detected with a whole soil milling process.
Alkorta I, Becerril JM, Garbisu C (2010) Phytostabilization of metal
We refer to the class of asbestos-containing material that can- contaminated soils. Rev Environ Health 25:135–146
not be crumbled, pulverized, or reduced to powder by pressure Baker AJM (1987) Metal tolerance. New Phytol 106:93–111. https://doi.
(Perkins et al. 2007). Currently, EPA is evaluating this sam- org/10.1111/j.1469-8137.1987.tb04685.x
pling method for its applicability to Superfund site Bandli BR, Gunter ME (2006) A review of scientific literature examining
the mining history, geology, mineralogy, and amphibole asbestos
characterization. health effects of the Rainy Creek igneous complex, Libby,
We showed also that asbestos itself did not affect plant Montana, USA. Inhal Toxicol 18:949–962. https://doi.org/10.
health, although insufficient nutrients or high levels of other 1080/08958370600834982
phytotoxic contaminants could. Thus, it is important to con- Baran A, Tarnawski M (2013) Phytotoxkit/Phytotestkit and Microtox®
sider all limiting parameters in soils, not just the primary pol- as tools for toxicity assessment of sediments. Ecotoxicol Environ
Saf 98:19–27. https://doi.org/10.1016/j.ecoenv.2013.10.010
lutant of contaminated sites, such as asbestos or heavy metals. Boulanger G, Andujar P, Pairon J-C et al (2014) Quantification of short
Based on biomass production from different plant species and long asbestos fibers to assess asbestos exposure: a review of
tested here, crops may be adequate for phytoremediation in fiber size toxicity. Environ Health 13:59. https://doi.org/10.1186/
most sites with low and moderate concentrations of heavy 1476-069X-13-59
Brady KU, Kruckeberg AR, Bradshaw HD (2005) Evolutionary ecology
metals in soils. Alternatively, plants making up one particular
of plant adaptation to serpentine soils. Annu Rev Ecol Evol Syst 36:
vegetation type, called serpentinophytes, may be more appro- 243–266
priate with higher concentrations of heavy metal in soils be- Bray RH, Kurtz LT (1945) Determination of total, organic, and available
cause of their tolerance to particular edaphic properties of forms of phosphorus in soils. Soil Sci 59:39–46. https://doi.org/10.
serpentine soil (Brady et al. 2005). They did not prove advan- 1097/00010694-194501000-00006
Brown SK (1987) Asbestos exposure during renovation and demolition
tageous in the BoRit soil.
of asbestos-cement clad buildings. Am Ind Hyg Assoc J 48:478–
Similarly to solutions already used in other Superfund sites 486. https://doi.org/10.1080/15298668791385075
contaminated with Pb or Zn, we propose to place soil amend- Brown SL, Chaney RL (2016) Use of amendments to restore ecosys-
ments on top of the soil up to 5 cm thick and use cultivated tem function to metal mining-impacted sites: tools to evaluate
plant crop cultivars (Dietterich and Casper 2016; Brown and efficacy. Curr Pollut Rep 2:91–102. https://doi.org/10.1007/
s40726-016-0029-1
Chaney 2016) or serpentinophytes. This solution seems eco- Burt R, Wilson MA, Mays MD, Lee CW (2003) Major and trace elements
nomically more realizable than capping up to two meters of of selected Pedons in the USA. J Environ Qual 32:2109. https://doi.
non-contaminated soils and with placement of a water org/10.2134/jeq2003.2109
 Environ Sci Pollut Res

Casper BB, Bentivenga SP, Ji B et al (2008) Plant-soil feedback: testing abandoned chromite–asbestos mine. J Soils Sediments 17:1358–
the generality with the same grasses in serpentine and prairie soils. 1368. https://doi.org/10.1007/s11368-015-1323-z
Ecology 89:2154–2164 Lee GF, Jones-Lee A (1997) Hazardous chemical site remediation
Chaney RL, Newhart G, Mahoney, M, Schmeltzer J (2011) Ecosystem through capping: problems with long-term protection. Remediat J
restoration of the long-barrenserpentine mine waste at the Vermont 7:51–57. https://doi.org/10.1002/rem.3440070406
Asbestos Group Mine Site. Int. Phytotechnol. Soc. Conference, Lee RJ, Strohmeier BR, Bunker KL, Van Orden DR (2008) Naturally
Sept.12–15. Portland, OR occurring asbestos—a recurring public policy challenge. J Hazard
Clarke S, French K (2005) Germination response to heat and smoke of 22 Mater 153:1–21. https://doi.org/10.1016/j.jhazmat.2007.11.079
Poaceae species from grassy woodlands. Aust J Bot 53:445–454. Levitan DM, Hammarstrom JM, Gunter ME et al (2015) Mineralogy of
https://doi.org/10.1071/BT04017 mine waste at the Vermont Asbestos Group mine, Belvidere
Clevenger TE (1990) Use of sequential extraction to evaluate the heavy Mountain, Vermont. Am Mineral 94:1063–1066. https://doi.org/
metals in mining wastes. Water Air Soil Pollut 50:241–254. https:// 10.2138/am.2009.3258
doi.org/10.1007/BF00280626 Lookingbill TR, Engelhardt KA, Florkowski LN, et al (2007) Evaluation
Cunningham HM, Pontefract RD (1971) Asbestos fibres in beverages and of the Nottingham Park serpentine barrens. University of Maryland
drinking water. Nature 232:332–333 1-53
ASTM. Standard D7521-16-Standard Test Method for Determination of Marschner P (2012) Marschner’s mineral nutrition of higher plants.
Asbestos in Soil [Internet]. 2013. https://www.astm.org Academic Press
Daghino S, Turci F, Tomatis M et al (2006) Soil fungi reduce the iron Meyer DR (1980) Nutritional problems associated with the establishment
content and the DNA damaging effects of asbestos fibers. Environ of vegetation on tailings from an asbestos mine. Environ Pollut Ser
Sci Technol 40:5793–5798. https://doi.org/10.1021/es060881v Ecol Biol 23:287–298. https://doi.org/10.1016/0143-1471(80)
Dietterich LH, Casper BB (2016) Initial soil amendments still affect plant 90071-9
community composition after nine years in succession on a heavy Millano EF (1998) Hazardous waste: storage, disposal, remediation, and
metal contaminated mountainside closure. Water Environ Res 70:721–745
Dodson RF, Hammar SP (2011) Asbestos: risk assessment, epidemiolo- Mohanty SK, Gonneau C, Salamatipour A et al (2017) Siderophore-
gy, and health effects, Second Edition. CRC Press mediated iron removal from chrysotile: implications for asbestos
Favero-Longo SE, Turci F, Fubini B et al (2013) Lichen deterioration of toxicity reduction and bioremediation. J Hazard Mater. https://doi.
asbestos and asbestiform minerals of serpentinite rocks in Western org/10.1016/j.jhazmat.2017.07.033
Alps. Int Biodeterior Biodegrad 84:342–350 Morrison RD, Murphy BL (2010) Environmental forensics: contaminant
Finch-Savage WE, Leubner-Metzger G (2006) Seed dormancy and the specific guide. Academic Press
control of germination. New Phytol 171:501–523. https://doi.org/
Munson R (1997) Principles of plant analysis. In: Kalra Y (ed) Handbook
10.1111/j.1469-8137.2006.01787.x
of reference methods for plant analysis. CRC Press
Fortunato L, Rushton L (2015) Stomach cancer and occupational expo-
Nagajyoti PC, Lee KD, Sreekanth TVM (2010) Heavy metals, occur-
sure to asbestos: a meta-analysis of occupational cohort studies. Br J
rence and toxicity for plants: a review. Environ Chem Lett 8:199–
Cancer 112:1805–1815
216. https://doi.org/10.1007/s10311-010-0297-8
Gonneau C, Genevois N, Frérot H et al (2014) Variation of trace metal
Paik NW, Walcott RJ, Brogan PA (1983) Worker exposure to asbestos
accumulation, major nutrient uptake and growth parameters and
during removal of sprayed material and renovation activity in build-
their correlations in 22 populations of Noccaea caerulescens. Plant
ings containing sprayed material. Am Ind Hyg Assoc J 44:428–432.
Soil 384:271–287. https://doi.org/10.1007/s11104-014-2208-4
https://doi.org/10.1080/15298668391405085
Gonneau C, Mohanty SK, Dietterich LH et al (2017) Differential elemen-
tal uptake in three pseudo-metallophyte C4 grasses in situ in the Pearre NC, Heyl Jr AV (1960) Chromite and other mineral deposits in
eastern USA. Plant Soil 416:149–163 serpentine rocks of the piedmont upland, Maryland, Pennsylvania,
Holmes EP, Lavkulich LM (2014) The effects of naturally occurring acids and Delaware
on the surface properties of chrysotile asbestos. J Environ Sci Health Perkins RA, Hargesheimer J, Fourie W (2007) Asbestos release from
Part A 49:1445–1452 whole-building demolition of buildings with asbestos-containing
Institute of Medicine (2006) Asbestos: Selected Cancers. Washington, material. J Occup Environ Hyg 4:889–894. https://doi.org/10.
DC: The National Academies Press. https://doi.org/10.17226/11665 1080/15459620701691023
Ji B, Bentivenga SP, Casper BB (2012) Comparisons of AM fungal spore Sanchez PA, Palm CA, Buol SW (2003) Fertility capability soil classifi-
communities with the same hosts but different soil chemistries over cation: a tool to help assess soil quality in the tropics. Geoderma 114:
local and geographic scales. Oecologia 168:187–197 157–185. https://doi.org/10.1016/S0016-7061(03)00040-5
Kabata-Pendias A (2000) Trace elements in soils and plants. CRC Press Schreier H (1989) Asbestos in the natural environment. Elsevier
Krämer U (2010) Metal hyperaccumulation in plants. Annu Rev Plant Smith RC, Barnes JH (1998) Geology of Nottingham County Park.
Biol 61:517–534 Commonwealth of Pennsylvania, Department of Conservation and
Kranner I, Colville L (2011) Metals and seeds: biochemical and mo- Natural Resources, Bureau of Topographic and Geologic Survey
lecular implications and their significance for seed germination. Smith SE, Read DJ (2010) Mycorrhizal symbiosis. Access Online via
Environ Exp Bot 72:93–105. https://doi.org/10.1016/j. Elsevier
envexpbot.2010.05.005 Smith DB, Cannon WF, Woodruff LG et al (2013) Geochemical and
Kumar A, Maiti SK (2015) Effect of organic manures on the growth of mineralogical data for soils of the conterminous United States. US
Cymbopogon citratus and Chrysopogon zizanioides for the Geol Surv Data Ser 801:19
phytoremediation of chromite-asbestos mine waste: a pot scale ex- Trivedi AK, Ahmad I (2011) Effects of chrysotile asbestos contaminated
periment. Int J Phytorem 17:437–447. https://doi.org/10.1080/ soil on crop plants. Soil Sediment Contam Int J 20:767–776. https://
15226514.2014.910174 doi.org/10.1080/15320383.2011.609197
Kumar A, Prasad MNV, Maiti SK, Tripti (2016) Chapter 13—asbestos: Van Gosen BS (2005) Reported historic asbestos mines, historic asbestos
resource recovery and its waste management. In: Environmental prospects, and natural asbestos occurrences in the Eastern United
Materials and Waste. Academic Press, pp 285–305 States
Kumar A, Maiti SK, Tripti et al (2017) Grasses and legumes facilitate Van Gosen BS (2007a) The geology of asbestos in the United States and
phytoremediation of metalliferous soils in the vicinity of an its practical applications. Environ Eng Geosci 13:55–68
Environ Sci Pollut Res

Van Gosen BS (2007b) Reported historic asbestos mines, historic asbes- Vogel J (2008) Unique aspects of the grass cell wall. Curr Opin Plant Biol
tos prospects, and natural asbestos occurrences in the Rocky 11:301–307
Mountain States of the United States (Colorado, Idaho, Montana, Wroble J, Frederick T, Frame A, Vallero D (2017) Comparison of soil
New Mexico, and Wyoming). Geological Survey (US) sampling and analytical methods for asbestos at the Sumas
Van Gosen BS (2008) Reported historic asbestos mines, historic asbestos Mountain Asbestos Site—working towards a toolbox for better as-
prospects, and natural asbestos occurrences in the southwestern sessment. PLoS One 12:e0180210. https://doi.org/10.1371/journal.
United States (Arizona, Nevada, and Utah). Geological Survey (US) pone.0180210
Virta RL (2006) Worldwide asbestos supply and consumption trends
from 1900 through 2003. US Geological Survey Reston, VA