Arsenic accumulation by edible aquatic macrophytes

Ecotoxicology and Environmental Safety 99 (2014) 74–81

Contents lists available at ScienceDirect

Ecotoxicology and Environmental Safety journal homepage: www.elsevier.com/locate/ecoenv

Arsenic accumulation by edible aquatic macrophytes K.A. Falinski a,n, R.S. Yost a, E. Sampaga b,1, J. Peard c a b c

Department of Tropical Plant and Soil Sciences, University of Hawaii at Manoa, 3190 Maile Way, Honolulu, HI 96822, United States Hilo High School, Hilo, HI 96811, United States Hazard and Emergency Environmental Response Office, Hawaii Department of Health, Hilo, HI, United States

art ic l e i nf o

a b s t r a c t

Article history: Received 13 June 2013 Received in revised form 26 September 2013 Accepted 2 October 2013 Available online 5 November 2013

Edible aquatic macrophytes grown in arsenic (As)-contaminated soil and sediment were investigated to determine the extent of As accumulation and potential risk to humans when consumed. Nasturtium officinale (watercress) and Diplazium esculentum (warabi) are two aquatic macrophytes grown and consumed in Hawaii. Neither has been assessed for potential to accumulate As when grown in As-contaminated soil. Some former sugarcane plantation soils in eastern Hawaii have been shown to have concentrations of total As over 500 mg kg
1. It was hypothesized that both species will accumulate more As in contaminated soils than in non-contaminated soils. N. officinale and D. esculentum were collected in areas with and without As-contaminated soil and sediment. High soil As concentrations averaged 356 mg kg
1, while low soil As concentrations were 0.75 mg kg
1. Average N. officinale and D. esculentum total As concentrations were 0.572 mg kg
1 and 0.075 mg kg
1, respectively, corresponding to hazard indices of 0.12 and 0.03 for adults. Unlike previous studies where watercress was grown in As-contaminated water, N. officinale did not show properties of a hyperaccumulator, yet plant concentrations in high As areas were more than double those in low As areas. There was a slight correlation between high total As in sediment and soil and total As concentrations in watercress leaves and stems, resulting in a plant uptake factor of 0.010, an order of magnitude higher than previous studies. D. esculentum did not show signs of accumulating As in the edible fiddleheads. Hawaii is unique in having volcanic ash soils with extremely high sorption characteristics of As and P that limit release into groundwater. This study presents a case where soils and sediments were significantly enriched in total As concentration, but the water As concentration was below detection limits. & 2013 Elsevier Inc. All rights reserved.

Keywords: Arsenic Aquatic macrophytes Plant uptake Soil contamination

1. Introduction The accumulation of arsenic (As) from contaminated soils, sediments and water into edible plants presents a possible route for exposure to humans. Arsenic is a well-known human carcinogen affecting numerous organs, and As toxicity is associated with multisystem disease (Ratnaike, 2003). Exposure is usually from absorption through the small intestine, through ingestion of As in water, food or contaminated soil particles. Potential human health risks due to ingestion of plants containing As have recently received attention, especially for rice (Zhao et al., 2010). Additional studies documenting As uptake in vegetables, including vegetables irrigated with contaminated well water (Baig and Kazi, 2012) or grown on contaminated soil (McBride et al., 2013), and taro grown in contaminated soil and water (Kurosawa et al., 2008), further outline the possible risk of As through plant consumption. n

Corresponding author. E-mail addresses: [email protected], [email protected] (K.A. Falinski). 1 Currently at: University of Washington, Department of Earth and Space Sciences, Seattle, WA 98195, United States. 0147-6513/$ - see front matter & 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.ecoenv.2013.10.008

The accumulation of As is also important in wetland species, where there is a potential for reducing conditions to release As into the soil solution (Ha et al., 2009; Rahman and Hasegawa, 2011), or for co-deposition of As with Fe hydroxides adsorbed to the plant's surface (Zhao et al., 2002). A wide range of aquatic macrophytes have commonly been used for their role in phytoremediation of organic and inorganic pollutants, including As, because of their capacity to accumulate metals in their tissues (Dhir et al., 2009; Rahman and Hasegawa, 2011). There is a potential risk to human health due to consumption of contaminants accumulated by edible aquatic macrophytes, such as watercress and edible ferns. Robinson et al. (1995) found As levels of up to 500 mg kg
1 dry weight (DW) in the leaves of Ceratophyllum demersum (commonly called watercress), an aquatic, flowering plant found in a river system in New Zealand. Researchers were unable to conclude, however, whether the uptake was due to elevated concentrations of As in dissolved form in river water or in the sediments in which the watercress was grown. In a greenhouse experiment, they reported that watercress preferentially stores As in its leaves rather than in the stems (Robinson et al., 2003). Two reports have considered using the commonly

K.A. Falinski et al. / Ecotoxicology and Environmental Safety 99 (2014) 74–81

eaten watercress species Nasturtium officinale for phytoremediation (Duman et al., 2009; Kara, 2002). The species accumulated Zn, Cu, Ni, Cd, Co, and Cr from contaminated wastewater. A wide variety of ferns have been shown to accumulate As at high concentrations. The most widely studied accumulator is the Chinese brake fern (Pterisvittata L.), which has been shown to take up to 23,000 mg kg
1 of total As in its fronds (Ma et al., 2001). Other species of ferns have also shown high As uptake potential (Visoottiviseth et al., 2002; Zhao et al., 2002). Hawaii is unique in having a fern that is eaten as a delicacy. Diplazium esculentum (family Dryopteridaceae), the vegetable fern, known as warabi (from the Japanese), or ho’i’o or pohole (Hawaiian), is a woody fern that grows in wetland areas and along the banks of streams.

1.1. History of As contamination in Hawaii In Hawaii, As-based herbicides, especially sodium arsenite, were used extensively at high application rates on sugarcane (Saccharum spp.) plantations for weed control in the Hilo, Hawai’i, area from 1913 to the 1950s (HDOH, 2007). Sodium arsenite was replaced by pentachlorophenol formulations followed by phenoxy and triazinine herbicides. Most sugarcane plantations in the islands had closed by the early 1990s. In recent years, new developments, including housing, schools and government buildings, seek to re-appropriate former sugar cane land for residential and commercial use, though some former sugar cane lands are also used for different types of agricultural practices (HDOH, 2009; Niemeyer, 2011). Arsenic contaminated soil continues to be found at levels up to two orders of magnitude above typical background levels of 1.0–20 mg kg
1 (Cutler et al., 2013). The Hawaii Department of Health set an action level of 24 mg kg
1 for total As concentration in soil (HDOH, 2011). A similar limit does not currently exist for sediments. De Carlo et al. (2005) analyzed 24 sites on the island of Oahu, and found average sediment total As concentrations of 2.3, 8.1 and 22 mg kg
1 for forested, urban and agricultural sites, respectively. Although As has been found in extremely elevated levels in soils, As contamination of groundwater in Hawaii (as defined by the EPA maximum contaminant level of 10 μg L
1) has not been documented by the Safe Water Drinking branch in many years of testing of public drinking water supplies (Cutler, 2011). Previous studies of As in biota are sparse on Hawaii, and have focused on Waiakea Pond in Hilo, the site of a former Canec plant which manufactured bagasse-based insulation board treated with arsenic-containing compounds from 1932 to 1963 (GlendonBaclig, 2007; Hallacher et al., 1985). The watercress species grown for consumption in Hawaii, N. officinale R. Br. (strain Sylvasprings) or the fern-species D. esculentum have not been tested for As. In early 2011, exploratory soil samples were collected for As analysis at a site in Kaumana Springs Wilderness Area in Hilo, Hawaii. Extremely high As concentrations (up to 1000 mg kg
1, measured by EDXRF) in surface soils were anecdotally reported in the adjacent field. The Kaumana Springs results were yet to be validated and extended, and the extent and severity of As risk in the Wilderness Area was unclear. The initial hypothesis was that As-laden sediments were being transported downslope from higher-elevation former sugarcane fields through erosion processes. Moreover, N. officinale is grown at the lower, easternmost, boundary of the Wilderness Area in the freshwater springs for community consumption. Given the low slope of lower KSW, redox potentials in wetland conditions may chemically release As from its bonds with Fe and Al-oxides. It is hypothesized that aquatic macrophytes will accumulate As when in contact with high As soils. It remains unclear whether aquatic macrophytes grown for human consumption that come in

75

contact with highly As-contaminated bank or field soils present a potential health risk to humans. The objectives of this study were to (1) investigate reports of high As soil and determine the spatial extent and concentration of As in soil and sediments in Kaumana Springs Wilderness, (2) determine if aquatic macrophytes grown in a freshwater spring in contact with high-As soils and sediments accumulates As; and (3) assess risk to the community of exposure to As from consumption of D. esculentum and N. officinale.

2. Materials and methods 2.1. Site description Kaumana Springs Wilderness Area (KSW) is 16 ha, located between the Wailuku and Waialua Rivers in eastern Hawaii at an elevation of 120–220 m (Fig. 1). The area receives 4250 mm of rain annually (Giambelluca et al., 2013). The soil is classified as Panaewa series, a ferrihydritic Lithic Hydrudand formed from volcanic ash over pahoehoe lava. The site has a complicated, not very well-documented hydrology that is categorized by lava tubes and dikes (Paquay et al., 2007). Two freshwater springs (“North” and “South”) flow overland throughout the year. A third spring only surfaces at the lowest end of KSW, and is reported by local residents to connect to the “South” spring underground. The area has a slope of less than 2.51, and the soil is often water saturated. Most of the area is used for cattle forage and is cordoned off by fences. At the lowest (easternmost) end of Kaumana Springs, there was evidence of depositional areas. Throughout the KSW, bank erosion was observed along the freshwater springs. KSW supports small-scale N. officinale production at the lower (easternmost) end on the South branch of the springs. D. esculentum, which is often found near stream banks, was present in the same area. The local families reported that, to their knowledge, sugarcane had never been grown in the area and As-based herbicides had never been used (Sampaga, personal communication). The first set of soil and sediments were collected in September 2011 at discrete points throughout KSW (sites A, B, C, and E), and resampled intensively at site D at the lower extent of Kaumana Springs (Fig. 1). The area referred to as lower Kaumana Springs, approximately 100 m by 100 m, was analyzed in four decision units (D-1, D-2, D-3, and D-4) that are progressively more downstream. A barbed-wire fence separated D-2 and D-3. A follow-up sampling took place in August 2012. N. officinale samples were collected at discrete sites D-1, D-2, D-3 and E, while D. esculentum was collected at site D-4. Freshwater samples were collected in August 2012 at site D-2. Duplicate water samples (1 L, in acid washed polyethelene containers, not filtered) of both the North and South stream were taken in August 2012. 2.2. Location of total As in KSW In order to assess the spatial variability of site D, three experiments were designed. (1) The first study considered how total As varied with distance from the “North” branch of the freshwater spring. Samples were collected every 1 m from

Fig. 1. Map of Kaumana Springs Wilderness and the surrounding area in Hilo, Hawaii. The elevation difference between the higher site A and lower site F is approximately 100 m. Freshwater springs flow on the north and south side of the Wilderness area, as indicated in the lower Kaumana Springs inset by the dashed line.

76

K.A. Falinski et al. / Ecotoxicology and Environmental Safety 99 (2014) 74–81

the bank of the river at three depth intervals: 0–10 cm, 10–20 cm and 20–30 cm. (2) A second experiment investigated the depth distribution of As in the 100 m by 100 m area of site D. Samples were taken at 0–10 cm and 10–20 cm on a grid, every 10 m. (3) A third experiment compared As concentrations in soils with sediments along the bank where watercress is grown at the “South” freshwater spring in lower KSW. 2.3. Sample collection N. officinale and D. esculentum samples were collected in and around the lower portion of KSW. N. officinale samples for As analysis were collected at sites D-1, D-2 and D-3. N. officinale samples included both stems and leaves. For D. esculentum, only the fiddleheads were collected, as this is the part that people eat. It has been shown that As is likely accumulated predominantly in above-ground fern fronds (Tu and Ma, 2002). At least 15 D. esculentum fronds were composited for each site to comprise about 500 g of fresh weight. Samples were stored in pre-weighed paper bags after collection and while they were being dried. For N. officinale samples, approximately 30 subsamples of leaves and stems (combined) were picked by hand and composited. Plant samples were dried within 24 h at 60 1C in a dedicated plant oven. At a given site, three soil cores were collected 30 cm apart using a 20-cm diameter auger and composited according to depth interval. The auger was cleaned with a towel between sites. A subsample of the composite, weighing approximately 100 g, was made and placed in a zip type bag. Soils were air dried for three weeks, and then sieved onto a clean piece of weighing paper to less than 2 mm. For duplicate samples taken during preliminary sampling efforts, the duplicate was taken 1 m from the first sample. Additional surface soil samples were collected on the bank of the freshwater streams in lower KSW. The soil samples were collected by compositing a minimum of five surface samples collected with a hand shovel along a 10 m section of the bank. Stream sediment samples, composited from a minimum of ten samples, were collected from the top 10 cm of the streambed using either a plastic scoop or a small plastic core barrel. The collection area was as close to midstream as possible, within a 5 m stretch. Water samples at the lower KSW site were collected in six 100 mL acid-washed polyethylene bottles, and refrigerated prior to analysis. Samples were combined prior to analysis. The water was not filtered and represented the total As present in the water sampled. It may be considered an overestimate compared of the dissolved total As. 2.4. Chemical analysis Plant samples were prepared first by crushing dried material with mortar and pestle, sieving to 2 mm, and then rinsing with Milli-Q water. Five grams of sample were microwave digested in concentrated nitric acid at 175 1C for 10 min in accordance with EPA Method 3051a. The digested solution was analyzed using ICP-MS with a detection limit of 0.001 mg kg
1. In order to calculate the wet weight concentrations, the percent water content was determined by weighing the fresh samples of N. officinale and D. esculentum, drying for 24 h at 60 1C, and re-weighing. For soil and sediment samples, total As was extracted from 0.5 g of dried soil using the EPA Method 3051a described above. Soils were air dried prior to analysis, and sediments were dried for 24 h in an oven at 40 1C. Soil and sediments were sieved to 2 mm. In some cases, dried sediment samples needed to be crushed with a mortar and pestle prior to sieving. Total As in the extractant was determined using ICP-AES, with a limit of detection of 1 mg kg
1. Water blanks, solution standards, matrix spikes and soil standards were included with each run to calibrate results. Additional samples were re-analyzed using ICP-MS (also extracted with EPA Method 3051a) for confirmation. In addition, oxalate-extractable Fe, Al and P (0.2 M (NH4)2C2O4  H2O at pH 3.0, 1:60 soil:solution, 4 h equilibration time) was determined according to a modified method of acid ammonium oxalate extraction in the dark by McKeague and Day (1966). Sieved, dried soil samples were subsampled for 0.2 g of sample to be analyzed. Extractants were immediately centrifuged and filtered, and Fe, Al and P concentrations were determined by ICP-OES. 2.5. Data analysis The GIS for all layers was completed using NAD83, Zone 5 N coordinate system in a Universal Transverse Mercator projection. The tax map key information was obtained from the Hawaii GIS repository (www.state.hi.us/dbedt/gis), and was last updated for Hawaii County in 2011. Imagery obtained from the NRCS was used to compile the soil series layers. World Views satellite imagery from 2011 was used to explore land use characteristics of the KSW area. The mean plant uptake factor (PUF) was calculated for both N. officinale and D. esculentum. The PUF was calculated according to the EPA method (2009) as the ratio of the concentration of As in the edible parts of the plant to the concentration in the soil or sediment. In this study the concentration of total As in sediment was preferentially used, although adjacent soil samples were used for D. esculentum PUF calculations.

2.6. Risk assessment The exposure rate (mg kg-body weight
1 day
1) was calculated as the average As concentration in the plant in dry weight multiplied by the average daily dose from ingestion divided by the average body weight of an individual and the wet weight to dry weight ratio. The Joint FAO and WHO recommendation for provisional maximum tolerable daily intake (PTDI) of As in food is 15 mg kg-body weight
1day
1(Ng, 2011). The U.S. EPA chronic oral reference dose (RfD) is 0.3 μg As kg-body weight
1 day
1 for inorganic As. It is based on a no-observed adverse effect level of 0.8 mg As mg kg-body weight
1day
1 for dermal effects and possible vascular complications in a Taiwanese farming population exposed to As in well water (Tseng, 1977; U.S. Environmental Protection Agency, 1993). For this paper, the RfD was used to assess risk. The exposure limits presented above are specifically for inorganic As. Plants, including aquatic plants, store As in organic as well as inorganic forms. Information that describes the partitioning of As as inorganic or organic As is not available for N. officinale or D. esculentum. However for rice, the percentage of total As present as inorganic As varies widely (from 10% to 90%). It is assumed that the total As represents an overestimate of As. The hazard index (HI) was defined according to the U.S. EPA as the allowed exposure rate divided by the EPA acceptable non-carcinogenic daily consumption limit. If the HI is greater than one, there is a potential for adverse health risks. The average body weight of an adult is assumed to be 70 kg, and a child is 15 kg. Consumption of both N. officinale and D. esculentum is assumed to be a single serving size, or 100 g wet weight per day per adult. This is a conservative estimate, given that it is unlikely an adult will consume either species daily. For comparison, according to the EPA Exposure Factors Handbook, the daily ingestion rate of leafy vegetables by adult consumers is 0.59 g kg-body weight
1 day
1, or 41 g wet weight per day (U.S. Environmental Protection Agency, 2011).

3. Results 3.1. Spatial extent of As in Kaumana Springs Above background total As concentrations in soils in the Hilo areas sampled were limited to lower KSW and Keaau. Initial sampling of soil and sediments in lower KSW reported very high concentrations of total As in the area surrounding the middle freshwater spring and along the North freshwater spring (Fig. 2). However, less than 50 m upstream at sites D-1 and D-2, mean total As concentrations were less than 5 mg kg
1 (Table 1). Arsenic was not detected at upstream sites A, B, and C. Analytical results of soil samples from Keaau confirmed the high soil total As concentrations previously reported there, with average values of 346 mg kg
1 and 391 mg kg
1 for the two sites chosen. The average relative percent difference (RPD) between different analytical techniques (ICP-MS and ICP-AES) was 13% (n ¼10) for the samples measured with both methods. Between true replicates that were analyzed using ICP-AES, the RPD was 21% (n ¼9). The 10–20 cm depth interval had consistently higher values than the 0–10 cm depth interval in lower KSW, indicating that either sediments were deposited on top of soils where As-based herbicides were applied, or that As-laden sediments were transported to the site at a previous time (Fig. 2a). In addition, the data collected along a transect for the North branch of the stream suggested that fluvial deposition processes occurred. Total As concentrations were highest approximately 2 m from the bank of the river in the 20–30 cm depth interval (Fig. 2b). In all sites, sediment samples had significantly less total As than the soils collected on banks of the stream near the same location (Fig. 3). Similar to the adjoining soils, sediments at D-3 were much higher in total As concentration than at D-1 and D-2, having average concentrations of 45 mg kg
1 versus 2.0 mg kg
1. Elevated total As concentrations were also found in sediments downstream at sites E and F. The average total As concentration in the high As sediments was 23.3 mg kg
1, with a maximum of 112 mg kg
1. This is similar to the average total As concentrations in sediments found previously in streams on Oahu island by De Carlo et al. (2005), although the soil in this study was only sieved to o2 mm, rather than 62 μm as in the De Carlo study.

K.A. Falinski et al. / Ecotoxicology and Environmental Safety 99 (2014) 74–81

77

Fig. 2. (a) Soil total As concentrations in lower Kaumana Springs Wilderness (KSW) by depth interval. The box plot shows the median (bar) and spans from the 25th to 75th percentiles. The black circles represent values outside that range. (b) Total As concentrations along a transect perpendicular to the North stream are shown for three depth intervals: 0–10 cm, 10–20 cm and 20–30 cm. (c) Maximum total As concentrations in lower KSW, shown with the outline of the soil and sediment decision units. Table 1 Total As concentrations in soils and sediments collected within designated decision units in Hilo, HI. For plant type, (x) indicates N. officinale was collected, while (o) indicates D. esculentum was collected at the site. Site

Sample type

Site category

Total As (mg kg
1): all replicates 0–10 cm

A B C D-1 D-2 D-2 D-3

Soil Soil Soil Soil Soil Sediment Soil

Low Low Low Low Low Low High

0.046 0.165 0.027 1.5 2.01 3.61 282 241 40.5 14.0 (0–10 cm) 29.8 13 386, 294 (0–10 cm) 390, 410 (0–10 cm)

D-3 D-4

Sediment Soil

High Low

E F Keaau-1

Sediment Sediment Soil

High – High

Keaau-2

Soil

High

0.347 0.209 0.293 1.73 1.5 0.52 690 270 27.6 3.9 (10–20 cm)

387, 279 (10–20 cm) 342.2, 479 (10–20 cm)

1.57 1.73 0.79 161 287 17.6

410, 381 (20–30 cm) 253, 472 (20–30 cm)

1.44 1.5 0.96 258 112

Average

1.53 1.45 4.24 371 29.2

Standard deviation

Relative standard deviation

0.21 0.03 0.19 0.11 0.23 1.76

1.53 0.23 1.66 0.07 0.14 0.87

320 45.4 9.0

160 38.1 7.15

0.50 0.84 0.80

29.8 13 356

– – 55.0

– – 0.155

o

391

85.0

0.217

o

0.20 0.19 0.16 1.55 1.64 2.02

Plant type

x x x

o x

78

K.A. Falinski et al. / Ecotoxicology and Environmental Safety 99 (2014) 74–81

Mean oxalate-extractable Fe and Al concentrations for the 0– 10 cm soil interval in lower KSW were 4.0% (1.5% s.d., ranging from 0.7% to 6.7%) and 3.6% (1.9% s.d., ranging from 1.9% to 6.6%), respectively, for 29 samples. Oxalate-extractable P in the same lower KSW samples ranged from 234 to 2041 mg kg
1, with an average of 923 mg kg
1. The samples from Keaau were similar, with a mean oxalate-extractable Fe, Al and P of 3.4%, 3.7% and 1230 mg kg
1, respectively. Phosphorus has been shown to co-vary with As (Cutler, 2011). However, in this study it was notable that in DU1, where total As was less than 5 mg kg
1, oxalateextractable P remained at 1607 mg kg
1. At DU3, with high total As concentrations, mean P was similar to DU1 (1588 mg kg
1).

Fig. 3. N. officinale total As concentrations compared with soil and sediment total As concentrations for select sites.

Water samples did not have detectable concentrations of As (o0.001 mg kg
1). 3.2. Aquatic macrophyte concentrations Because sites D-1 and D-2 contained minimal total As concentrations ( o5 mg kg
1) in the embankment soils, and because they were upstream of the other lower KSW sites, they can be considered a reasonable control or “Low” As site for comparing N. officinale As concentrations. In contrast, site D-3 had maximum individual sample concentrations of 690 mg kg
1 in the embankment soil and 112 mg kg
1 in the sediments and is considered “High”. Site E was directly downstream from site D, and therefore downstream of the As soil contamination. The average concentration of total As in N. officinale at D-1 and D-2 was 0.208 mg kg
1, while at D-3 and E the average was more than twice as high, at 0.572 mg kg
1. A N. officinale sample collected at site E had an As concentration of 0.77 mg kg
1 DW, higher than all previous samples. A linear regression analysis showed a slight positive correlation (R2 of 0.35) for ln-transformed plant and soil concentrations (Fig. 4). N. officinale had an average of 22.5 kg of wet weight mass for every 1 kg of dry weight mass, although this value had a wide range for the samples measured (between 12.9 and 35.0 wet:dry). The average plant uptake factor (PUF) for N. officinale at the High As sites was 0.010, which is an order of magnitude higher than the mean PUF in Bacigalupo and Hale's study (2011). Initial results showed that the soil sample at site D where D. esculentum was collected did not contain As above background levels, so two additional sites in the Hilo area were located for D. esculentum collection. The sites (not shown on the map) were located one kilometer apart in Keaau, where the previous vegetable survey had been conducted (HDOH, 2007) and high As soil samples were documented (Cutler et al., 2013). Soil samples (Olaa series, a Typic Hydrudand) were collected concurrently at the Keaau sites (Keaau-1 and Keaau-2). Both sites had returned to dense grass and brush cover. For the D. esculentum samples, five samples did not contain detectable As; the other two had concentrations of 0.144 (D-4) and 0.15 mg kg
1(Keaau). An average of 13 kg of wet weight mass for 1 kg of dry weight mass was found (s.d. ¼1). The ln-transformed data, excluding samples where As was not detected, had an R2 of 7.9e
5, and is not displayed in Fig. 4. The average PUF was 1.8e
4 for D. esculentum. 3.3. Risk assessment

Fig. 4. Regression analysis of ln-transformed plant and soil total As concentrations for N. officinale. The R2 of the linear regression line (dotted line) is 0.3496.

The plant samples were analyzed for child and adult risk at both high and low category total As sites (Table 2). The high total As category site (D-3), a single serving of N. officinale would contain 2.55 mg total As (Table 2). Using the EPA RfD recommendation for chronic daily As intake levels in food of 0.3 mg kg-body weight
1 d
1, the Hazard Index (HI) is 0.6 for a 15 kg child and

Table 2 Select risk analysis parameters for the two plant species collected in this study. Calculations represent risk for a 70-kg adult and 15-kg child.

N. N. D. D.

officinale offiicinale esculentum esculentum

n

Site As category

Mean plant total As, dry weight (mg kg
1)

Amount As per 100-g serving size (μg)

Amount of plant consumed to reach daily limit, adult (kg)

Hazard index, (HI), adult

Amount of plant consumed to reach daily limit, child (kg)

Hazard index (HI), child

Plant uptake factor (PUF)n

High Low High Low

0.572 0.208 0.075 0.029

2.55 0.92 0.54 0.21

0.82 2.3 3.85 10.0

0.121 0.044 0.026 0.010

0.18 0.49 0.82 2.13

0.566 0.206 0.121 0.047

0.010 0.130 0.00018 0.0064

The PUF was calculated using soil total As concentration for site D-3, and sediment total As concentration for site E.

K.A. Falinski et al. / Ecotoxicology and Environmental Safety 99 (2014) 74–81

0.10 for a 70 kg adult. Based on a target Hazard Quotient of 1.0, a mean daily intake (MDI) of 4.5 μg d
1 would be recommended for children and 21 μg d
1 for adults. To reach the MDI, a child would have to eat 176 g of N. officinale (just under two servings) of N. officinale daily for a period of six years. An adult would have to eat the unlikely amount of 980 g per day (over nine servings) of N. officinale daily. If the maximum As value found (at site E) is used for the calculation, and the minimum wet to dry ratio, a potential adult consumer would reach the limit by eating 610 g of N. officinale daily. Similar calculations are presented in Table 2 for D. esculentum. Assuming that a single, 100 g serving of D. esculentum contains 0.54 mg total As, Hazard Indices of 0.12 and 0.03 are calculated or children and adults, respectively indicating that is unlikely that consumption would present a risk. To reach the MDI, a 15 kg child would need to eat 825 g of D. esculentum daily. A 70 kg adult would need to eat over 3 kg daily before a potentially significant, noncancer risk would arise.

4. Discussion 4.1. Spatial extent of As in Kaumana Springs Wilderness Although the preliminary sampling indicated that there was a possible risk in KSW, the follow-up results suggest that contamination is limited to a small part of KSW. In light of the findings that As concentrations remained less than 5 mg kg
1 50 m upstream, and the lack of elevated concentrations of As at sites A, B or C in KSW, it seems most likely that As-based herbicides were applied in situ to the high As area documented in this study. In addition, in lower KSW, a cattle fence separates the areas documented to be generally high and low in total As. Depth analysis showed that the 0–10 cm soil layer did not have the highest concentration for lower KSW, in contrast to previous Department of Health findings for the greater Hilo area (HDOH, 2007). This is consistent with fluvial deposition processes. The samples 4 m from the river show significant (250%) increase in the 20–30 cm depth interval compared to the 10–20 cm, which was again 250% greater than the 0–10 cm interval. The As is unlikely to travel down the soil profile without some physical process. Sediment deposition events during floods may have moved nearby non-As soils from upstream above As-contaminated sediments. This study demonstrated the importance of considering the spatial variability of a site. Arsenic can have a range of concentrations on o10 m scales. Samples in lower KSW ranged from not detectable to 630 mg kg
1, with an average of 122 mg kg
1 and a standard deviation of 157 mg kg
1. In addition, the 22% RPD between duplicate samples taken at the same site indicates that microscale variability may also be important, and that subsampling protocols and larger sample sizes might be necessary. The current method for assigning risk to a site like Kaumana Springs is to sample according to the multi-increment sampling (MIS) method. The MIS method was first proposed by Ramsey and Hewitt (2005). The method involves selecting decision units, and sampling systematically and randomly within the decision unit. More than 30 samples are typically collected and combined to form a single sample. In this case, a surface sampling would have missed the higher concentrations found in the 10–20 cm layer at lower KSW, and compositing over a larger area would have missed detecting the highest As concentrations. The degree of contaminant heterogeneity (both vertically and horizontally) documented at this study site highlights the potential need for multiple horizontal and vertical decision units necessary to provide a detailed description of contaminant spatial distribution using MIS methods.

79

Further investigation would be merited to better understand the relationship between soils and sediments and sorbed As. In this study, the concentration of As in sediments in the river (taken from the river bed and bank) was much less than in the soils next to the river. Sutherland and Tolosa (2000) saw a similar effect for Cu, Pb and Zn for Oahu streams, and hypothesized that contaminated sediments were diluted with fluvial sediments from upstream that had not been contaminated. Arsenic attached to particles was likely transported from finer-grained soils and sediments through erosion processes including surface runoff, fluvial deposits and bedflow movement. Additionally, larger grain size in the sediments provides less surface area on which the As could sorb. Although the pH of freshwater is higher than the pH in the soils, it is not high enough that the stability diagrams of Fe and Al would predict dissolution under aerobic conditions. 4.2. Aquatic macrophyte concentrations The maximum concentration of total As in N. officinale in this study was 0.77 mg kg
1 DW. There was a significant difference (p o0.01) between the upstream As levels and the downstream levels in the N. officinale, corresponding to elevated soil concentrations downstream. Although the sediments in the Robinson et al. (1995) study in New Zealand had similar levels of total As in the sediments (30 mg kg
1), the authors reported that species N. officinale, Ceratophyllum demersum, and Egeria densa had total plant As concentrations up to 1000 mg kg
1 dry weight. A main difference between this study and the Robinson et al. study, was that the Robinson et al. study reported total As concentrations in the river water (4 0.01 mg L
1). The data of this study suggest that N. officinale does not access As from the sediments in quantities that would define it as a hyperaccumulator. The significant As sorption properties of Hawaiian volcanic soils and sediments likely inhibits As release to water. It is possible that As is likely to be released into the water on a different time scale than the time scale of this study. Due to the large percentages of Fe and Al oxides (4% and 3.6%) at the lower KSW site, As is likely minimally mobile in soil, and may remain in the soil for decades or longer (Ortiz-Escobar et al., 2006). Volcanic ash derived soils contain amorphous oxides, which have been shown for P to sorb as much as 3500 mg kg
1 and to contain up to 8% Fe and 4% Al (Fox and Searle, 1978; Hue and Fox, 2010). Arsenic sorption isotherms were recently conducted on the Typic Hydrudands by Cutler (2013); results indicated that for As spike amounts of 1000 mg L
1, the soil retained 97% of available As. Cutler also studied the bioaccessibility of As in Hydrudand soils, and reported that 8.03% of total As was bioaccessible in the soils for the o250 mm size fraction. These soil properties presumably lead to As being less accessible by N. officinale grown in areas with volcanic ash soils such as the Hydrudands. Soil characteristics have previously been shown to cause variation in the PUF (Bacigalupo and Hale, 2011). The soils were collected from a small area in the same soil series and the samples had similar soil characteristics. Studies have shown that soil amendments with phosphate increase As solubility, and this may be a possible concern for future crops grown in the lower KSW area (McBride et al., 2013). In contrast to a greenhouse experiment, this study was able to investigate plant uptake from soil and sediment where As was applied decades ago, possibly affecting the bonding strength of As in the soil. Bacigalupo and Hale (2011) showed that for leafy vegetables, natural log-transformed total As in the tissue was linearly related to natural log-transformed soil total As, with a correlation coefficient of 0.64. In this study, the correlation coefficient was 0.35 for the same transformation, indicating that the N. officinale uptake of

80

K.A. Falinski et al. / Ecotoxicology and Environmental Safety 99 (2014) 74–81

As did not strongly correlate to soil As concentration. However, N. officinale grown next to high As soils did accumulate 275% more As than watercress grown next to low As soils. The As concentration in the edible fern D. esculentum was minimal, with a maximum concentration of 0.15 mg kg
1. It was not found to be a hyperaccumulator of As. 4.3. Risk assessment The N. officinale samples analyzed from sites with elevated soil/ sediment As levels had a significantly higher concentration of total As than other edibles tested in East Hawaii. The HI child risk factor of 0.6 for N. officinale is less than one, which indicates that it is unlikely there is a non-cancer health hazard from chronic consumption. It is noteworthy that the N. officinale samples that were grown on soil and sediment without the elevated As concentrations still had As concentrations in their tissues of 0.208 mg kg
1 dry weight, higher than the previously reported average for leafy vegetables in As-contaminated areas on Hawaii (0.03 mg kg
1 wet weight, approximately 0.003 mg kg
1 dry weight) (HDOH, 2007). Although the highest levels of As uptake in plants have been seen in ferns, the fiddleheads of the most commonly eaten fern in Hawaii were found with only minimal total As. Total As of D. esculentum fell within the typical range for market produce in general and does not appear to reflect a significant, elevated human health risk above dietary background (U.S. Food and Drug Administration, 2007; Yost et al., 2004; Arsenic, 2012). The maximum value of total As found in the watercress species (0.77 mg kg
1 dry weight) was comparable to the maximum values reported in a 2012 survey of rice products commonly consumed in the United States (2012). Watercress, however, has a much higher wet weight to dry weight ratio, and the daily consumption rate for watercress is lower than rice. The proportion of the total As levels found in watercress as inorganic As compounds or organic As compounds is not known, but in other aquatic macrophytes such as rice the percentage of inorganic As has been found to vary from 10–90%. Consequently, the risk findings based on total As levels reported in this study are believed to be conservative. By serving size, watercress grown in the high soil total As areas is within the same range as studies reporting inorganic As for rice, seafood and juice (Borak and Hosgood, 2007; Wilson et al., 2012). The consumption of watercress combined with other mechanisms for As ingestion (food and beverages or soil ingestion) makes the presence of As a possible concern for regular consumers. Given the high soil concentrations within which the watercress is grown, it is critical that watercress consumed should be washed thoroughly before used in cooking. Testing of garden vegetables and fruits revealed levels of As similar to U.S. FDA market basket survey data. A risk assessment concluded that As was not considered a concern in consumed vegetables (HDOH, 2007). In Hawaii, the risk of As exposure through drinking water is negligible. Inadvertent ingestion of contaminated soil, including not washing vegetables properly before consuming them, is the most likely risk of As exposure. The Hawaii Department of Health has clear action limits for total As in soil (HDOH, 2011), but there are no similarly clear limits for ingestion of As in foods derived from plants.

5. Conclusions Although there was some evidence of transport from upstream, elevated As concentrations were limited to a relatively small area sampled in lower KSW. As a result, the As in Kaumana Springs was likely applied on site, rather than transported from upstream. The presence of As at a lower depth layer than the rest of the Hilo area

suggests that application may have taken place sometime previously, perhaps when sugarcane was still commercially grown. It remains possible that the contamination extends further downstream, where sampling was not possible due to access limitations. Elevated soil and sediment As concentrations increased the content of As in plants, even in soils such as the Hydrudands with extremely high sorption capabilities. The sediments in the freshwater streams contained considerably less As than did nearby soils, likely due to the preferential downstream transport of fine grain particles that sorb As, and because of dilution effects from non-contaminated fluvial sediments upstream. Hawaii appears to be unique in having a low risk due to As even with high levels of total As in soils. Neither the N. officinale (watercress) nor D. esculentum (warabi) sampled in this study contain total As levels high enough to pose a significant risk for children or adults if consumed in typical, daily amounts.

Acknowledgments This research was supported by the Department of Health, Hazard Evaluation and Emergency Response Division. We would like to thank Pascale Pinner, Patrick Niemeyer, Guy Porter, Rebecca Briggs and others for their help with site selection and sample collection; Priya Subramoney and Robert Huang for their help with analysis; Tony Kimmet, NRCS, for providing World Vision 2 imagery; and Roger Brewer, Department of Health, for his helpful comments on the manuscript. References Arsenic in Your Food, 2012. Consumer Reports. vol. 77, p. 22. Bacigalupo, C., Hale, B., 2011. Soil-plant transfer factors for garden produce from contaminated soils: site specific versus generic estimates for As and Pb. Hum. Ecol. Risk Assess. 17, 394–413. Baig, J.A., Kazi, T.G., 2012. Translocation of arsenic contents in vegetables from growing media of contaminated areas. Ecotoxicol. Environ. Saf. 75, 27–32. Borak, J., Hosgood, H.D., 2007. Seafood arsenic: implications for human risk assessment. Regul. Toxicol. Pharmacol. 47, 204–212. Cutler, W., 2011. Bioaccessible Arsenic in the Soils of the Island of Hawaii. Geology and Geophysics (Ph.D. thesis). University of Hawaii, Honolulu, HI, p. 136. Cutler, W.G., et al., 2013. Bioaccessible arsenic in soils of former sugar cane plantations, Island of Hawaii. Sci. Total Environ. 442, 177–188. De Carlo, E.H., et al., 2005. Trace elements in streambed sediments of small subtropical streams on O’ahu, Hawai’i: results from the USGS NAWQA program. Appl. Geochem. 20, 2157–2188. Dhir, B., et al., 2009. Potential of aquatic macrophytes for removing contaminants from the environment. Crit. Rev. Environ. Sci. Technol. 39, 754–781. Duman, F., et al., 2009. Growth and bioaccumulation characteristics of watercress (Nasturtium officinale R. BR.) exposed to cadmium, cobalt and chromium. Chem. Speciat. Bioavailab. 21, 257–265. Fox, R.L., Searle, P.G.E., 1978. Phosphate adsorption by soils of the tropics. In: American Society of Agronomy (Ed.), The Diversity of Soils of the Tropics. ASA Special Publication No. 34, Madison, WI. Glendon-Baclig, C., 2007. Distribution and Inorganic Speciation of Arsenic in Waiākea Mill Pond and Wailoa River Estuary, Hawai’i Island. Tropical Conservation Biology and Environmental Science (M.S. thesis). University of Hawaii at Hilo, Hilo, HI, p. 113. Giambelluca, T.W., Q. Chen, A.G. Frazier, J.P. Price, Y.-L. Chen, P.-S. Chu, et al. 2013. Online rainfall atlas of Hawai'i. Bulletin of the American Meteorological Society 94: 313-316. Ha, N.T.H., et al., 2009. Phytoremediation of Sb, As, Cu, and Zn from contaminated water by the aquatic macrophyte Eleocharis acicularis. Clean-Soil Air Water 37, 720–725. Hallacher, L.E., et al., 1985. Distribution of arsenic in the sediments and biota of Hilo Bay, Hawaii. Pac Sci 39, 266–273. HDOH, 2007. Soil Arsenic Assessment Study, Kea’au, Hawaii’i. State Department of Health, Hazard Evaluation and Emergency Response, Honolulu, HI, p. 41. HDOH, 2009. Technical Guidance Manual. Hawaii Department of Health, Hazard Evaluation and Emergency Response. Honolulu, HI. HDOH, 2011. Update to Soil Action Levels for Inorganic Arsenic and Recommended Soil Management Practices. Department of Health, Hazard Evaluation and Emergency Response. Honolulu, HI, pp. 1–37. Hue, N.V., Fox, R.L., 2010. Predicting plant phosphorus requirements for Hawaii soils using a combination of phosphorus sorption isotherms and chemical extraction methods. Commun. Soil. Sci. Plant. Anal., 41.

K.A. Falinski et al. / Ecotoxicology and Environmental Safety 99 (2014) 74–81 Kara, Y.O., 2002. Phytoremediation: Using Nasturtium officinale to clean up Zn þ þ contaminated wastewater. Fresenius Environ. Bull. 11, 1084–1086. Kurosawa, K., et al., 2008. Groundwater-soil-crop relationship with respect to arsenic contamination in farming villages of Bangladesh—a preliminary study. Environ. Pollut. 156, 563–565. Ma, L.Q., et al., 2001. A fern that hyperaccumulates arsenic—a hardy, versatile, fastgrowing plant helps to remove arsenic from contaminated soils. Nature 409, 579. McBride, M.B., et al., 2013. Lead and arsenic uptake by leafy vegetables grown on contaminated soils: effects of mineral and organic amendments. Water Air Soil Pollut. 224, 1–10. McKeague, J.A., Day, J.H., 1966. Dithionite- and oxalate-extractable Fe and Al as aids in differentiating various classes of soils. Can. J. Soil Sci. 46, 13–22. Ng, J.C., 2011. Evaluation of Certain Contaminants in Food: Seventy-second report of the Joint FAO/WHO Expert Committe on Food Additives. World Health Organization. Niemeyer, P., 2011. Personal Communication. In: Falinski K. (Ed.), Hilo, HI. Ortiz-Escobar, M.E., N.V. Hue, W.G. Cutler, 2006. Recent developments on arsenic: contamination and remediation. In: Pandalai S.G. (Ed.), Recent Research Developments in Bioenergetics, Transworld Research Network, vol 4, 2006. Paquay, F.S., et al., 2007. Carbon dioxide dynamics in rivers and coastal waters of the “Big Island” of Hawaii, USA, during baseline and heavy rain conditions. Aquatic Geochem. 13, 1–18. Rahman, M.A., Hasegawa, H., 2011. Aquatic arsenic: phytoremediation using floating macrophytes. Chemosphere 83, 633–646. Ramsey, C.A., Hewitt, A.D., 2005. A methodology for assessing sample representativeness. Environ. Forensics 6, 71–75. Ratnaike, R.N., 2003. Acute and chronic arsenic toxicity. Postgrad. Med. J. 79, 391–396.

81

Robinson, B., et al., 2003. Uptake of arsenic by New Zealand watercress (Lepidium sativum). Sci. Total Environ. 301, 67–73. Robinson, B., et al., 1995. The distribution and fate of arsenic in the Waikato River System, North Island, New Zealand. Chem. Speciat. Bioavailab. 7, 89–96. Sutherland, R.A., Tolosa, C.A., 2000. Multi-element analysis of road-deposited sediment in an urban drainage basin, Honolulu, Hawaii. Environ. Pollut. 110, 483–495. Tseng, W.-P., 1977. Effects and dose-response relationships of skin cancer and blackfoot disease with arsenic. Environ. Health. Perspect. 19, 109. Tu, C., Ma, L.Q., 2002. Effects of arsenic concentrations and forms on arsenic uptake by the hyperaccumulator ladder brake. J. Environ. Qual. 31, 641–647. U.S. Environmental Protection Agency, 1993. Arsenic, Inorganic (CASRN 7440-38-2). Integrated Risk Information System Database. U.S. Environmental Protection Agency, 2011. U.S. EPA. Exposure Factors Handbook 2011 Edition (Final). Washington, DC. U.S. Food and Drug Administration, 2007. Total Diet Study Statistics on Element Results: Revision 4.1, Market Baskets 1991-3 through 2005-4. Center for Food Safety and Applied Nutrition, College Park, MD, p. 211. Visoottiviseth, P., et al., 2002. The potential of Thai indigenous plant species for the phytoremediation of arsenic contaminated land. Environ. Pollut. 118, 453–461. Wilson, D., et al., 2012. Arsenic and lead in juice: apple, citrus, and apple-base. J. Environ. Health 75, 14–20. (quiz 44). Yost, L., et al., 2004. Estimation of dietary intake of inorganic arsenic in US children. Hum. Ecol. Risk Assess. 10, 473–483. Zhao, F.J., et al., 2002. Arsenic hyperaccumulation by different fern species. New Phytol. 156, 27–31. Zhao, F.J., et al., 2010. Arsenic as a food chain contaminant: mechanisms of plant uptake and metabolism and mitigation strategies. In: Merchant S., et al. (Eds.), Annual Review of Plant Biology, vol. 61. Annual Reviews, Palo Alto, pp. 535–559.