The effect of polybrominated diphenyl ether flame retardants on soil chemical contamination after a high severity forest fire on a Mediterranean hillslope

(1)

The effect of polybrominated diphenyl ether flame

retardants on soil chemical contamination after a high

severity forest fire on a Mediterranean hillslope

Bachelor thesis by Stijn Vredevoort

University of Amsterdam

Institute for Biodiversity and Ecosystem Dynamics

Supervisor: Dr. Julian Campo

Second supervisor: Dr. John Parsons

(2)

1

Summary

The abundance and distribution of polybrominated diphenyl ether (PBDE) flame retardants on two Mediterranean hillslopes near the town of Azuébar in Spain was investigated during this research using through gas chromatography / mass spectrometry (GC/MS). The use of PBDEs, which are considered toxic and harmful for humans and natural environments, is highly restricted in the European Union. It is however suspected that PBDEs are used as flame retardants in fire extinguishing materials. Fire extinguisher producers are not transparent about the composition of their products and therefore it is necessary to investigate whether the extinguishers contain chemical compounds that can cause soil contamination. Soil samples were collected in September 2014, shortly after a high severity forest fire. One of the slopes was affected by the forest fire and the other slope was not burned for at least twenty years and used as a control slope. Sediment samples were also collected after rainfall events in order to investigate the role of erosion on the distribution of PBDEs on the slope. The samples were prepared for GC/MS through accelerated solvent extraction, two evaporation stages and a cleanup column. Thereafter concentrations of tri- to hepta-BDEs were detected using a selected ion monitoring method for the GC/MS apparatus. Statistical tests were then performed to indicate significant differences in BDE concentrations for the variables site, slope level, soil depth and vegetation cover. The highest concentrations were found for penta congener BDE-85. In soil samples concentrations were as high as 5.6 ng/g an after the first rainfall event the concentration of BDE-85 was 11.4 ng/g. For BDE-85 there were significantly different concentrations on the burned slope compared to the control slope and concentrations were higher at soil depth of 0-2 cm compared to 2-5 cm. Hexa- and hepta-BDEs were not found and for other tri- to penta-BDEs concentrations were very low and not significant. It cannot be concluded that a penta-BDE mixture was used in the fire extinguisher material due to low concentrations found and absence of other BDE congeners besides BDE-85. Concentrations that were found are probably the result of atmospheric deposition. Additional research is required to evaluate the effect of obtained concentrations on soil chemical contamination.

(3)

2

List of abbreviations

ANOVA= Analysis of variance

ASE = Accelerated solvent extraction BDE = Brominated diphenyl ether BFRs = Brominated flame retardants BS = Bare soil

BQ = Below limit of quantification BU = Burned site

CO = Control site

EFFIS = European Forest Fire Information System FRs = Flame retardants

GC = Gas chromatography

IPCC = Intergovernmental Panel on Climate Change LOD = Limit of detection

LOQ = Limit of quantification MP = Mobile phase

MS = Mass spectrometry NaN = Not a number

NOAA = National Oceanic and Atmospheric Administration NF = Not found

PAHs = Polycyclic aromatic hydrocarbons PBB = Parts per billion

PBDEs = Polybrominated diphenyl ether flame retardants

REACH = Registration, Evaluation, Authorization and Restriction of Chemicals RT = Retention time

SIM = Selective ion monitoring UC = Under canopy

(4)

3

1. Introduction

In this research the effect of polybrominated diphenyl ether flame retardants (PBDEs) on soil chemical contamination after a high severity forest fire on a Mediterranean hillslope is investigated. Soil samples were collected at a burned slope near Azuébar and a control slope 700 meters from there. The samples were analyzed for the abundance of lower brominated diphenyl ethers through gas chromatography/mass spectrometry (GC/MS). In this introduction the relevance of the research, a description of the study area and research questions are covered. The remainder of the research consists of a theoretical framework, the methodology, the results, a discussion section, conclusions and an evaluation of the research.

1.1 Relevance

Forest fires are affecting large areas of land in the Mediterranean region and contribute to the increase of greenhouse gas emission, desertification and erosion (Intergovernmental Panel on Climate Change [IPCC], 2014). While the frequency of forest fires is decreasing, the land surface area affected by the fires is increasing (IPCC, 2014). According to Lung et al. (2013) more areas in Europe will be exposed to the risk of forest fires in the future. This is caused by climatic factors as well as non-climatic factors such as fire ignition potential. Figure 1.0.1 (Lung et al., 2013) illustrates the increased risk areas based on the A1B emission scenario (IPCC, 2014). The figure indicates high risk levels might occur as far north as the Netherlands and Scandinavia. In order to extinguish or control these forest fires or wildfires chemicals are used that can be harmful to the environment and human health (Marlair et al., 2004; Segev et al.,2009). These harmful chemicals are known as flame retardants (FRs) and are not only used to extinguish fires but also in the production process of electronics, furniture, clothes and cars (Segev et al., 2009).

* Figure 1.0.1: Maps of Europe illustrating forest fire risk. (a) Indicates the situation for

the period 1961-1990 and (b) the projected situation for the period 2041-2070 (Lung et al., 2013).

(6)

5

Brominated flame retardants (BFRs) account for a large group of FRs that are used for firefighting. Polybrominated diphenyl ethers (PBDEs) are the second highest production group of BFRs and are known to be used for fire extinguishing (Alaee et al., 2003; Pepper et al., 2011). PBDEs increase the temperature at which an object burns making objects more flame resistant (Alaee et al., 2003). The production and use of the most common PBDEs, penta- octa- and deca-BDE, is nowadays highly restricted after regulations that were introduced by the European Union (Kemmlein et al., 2009). The regulations are included in the Registration, Evaluation, Authorization and Restriction of Chemicals (REACH) and indicate that PBDEs are proven to be toxic to humans and may degrade soils due to debromination in the soil (Eljarrat et al., 2008) . Despite the restrictions PBDEs are still encountered in the environment and therefore it is suspected that they are still used in the process of forest fire extinguishing. Producers of fire extinguishers are however not transparent about the composition of their products. With increasing areas affected by forest fires and thus a growing number of people endangered by forest fires it is necessary to investigate whether PBDEs are used in the process of firefighting. In this research soil samples collected in Spain were analyzed to indicate whether PBDEs are used for fire extinguishing. The presence of PBDEs in environmental samples would implicate that rules set by the European Union are not followed properly.

1.2 Study area

Samples collected after a high severity forest fire near Azuébar in Spain were analyzed. The municipality of Azuébar is located 61 kilometers north of Valencia in the Natural Park of Sierra de Espadán. The fire occurred on the 28th_{of august in 2014 and affected}

approximately 10.59 hectares of forested area. Samples were taken at this site on a concave slope with an angle of 25 degrees at an altitude of 370 meters. The area where soil samples were taken covers approximately 0.25 hectares. This burned site will be referred to as BU in the remainder of this report. Soil samples were also taken at a control site (CO) 700 meters north of the BU. The CO is a slope with similar soil and vegetation characteristics and has not been affected by wildfires for at least the past 20 years. The maps in figure 1.2.1 illustrate the location of Azuébar and the BU and CO. The maps were made using the coordinates of the sites with Google Earth pro. At both sites samples were collected at different slope levels, soil depths and under different vegetation covers. The climate of the study area is considered as meso-Mediterranean with mean annual precipitation of 500 mm and mean annual temperature of 15ºC (Campo, 2014). The most occurring vegetation species at the sites are quercus cocciferae, rosmarinus officinalis, pistachia lentiscus, brachypodium retusum and rhamnus. Due to the dry climate in the area and the steepness of the slope the vegetation density is low. The soil samples were collected in September 2014, shortly after the fire. No precipitation events occurred between the collection date and the fire. Sediment fences were constructed at the site to gather sediment samples after rainfall events. The fences were located beneath the slope where soil transported through erosion accumulated. Sediment samples were collected after four rainfall events spread over a period of one year. The collection periods were November 2014, March 2015, June 2015 and November 2015. Pictures taken at the BU and CO are included in appendix 1.

(7)

6

* Figure 1.2.1: Maps indicating the location of Azuébar (a) and the sample collection sites (b).

1.3 Research questions

In order to investigate whether PBDEs are used for extinguishing fires and the contamination effects some research questions were formulated. The main research question is: What is the role of PBDE flame retardants, used in the extinction of fires, on the chemical contamination of a Mediterranean soil at hillslope scale? In order to answer this question several research sub-questions have to be answered:

- To what extent are PBDEs abundant on a burned slope where FRs were spread? - How are the PBDEs distributed over a burned slope treated with FRs?

- How does vegetation influence the distribution of PBDEs? - How does erosion influence the distribution of PBDEs?

An answer to the first sub-question provides an insight on the quantity of PBDEs that can occur in a burned soil treated with FRs. This quantity can be related to the soil chemical contamination. The second sub-question gives an indication of the dynamics of the PBDEs. Higher concentrations might, for example, be found in the sediment samples compared to the concentrations on hillslope levels due to erosion of the top soil containing FRs. With the third sub-question the impact of vegetation on PBDE concentrations is investigated. Vegetation might influence the movement and quantity of PBDEs in the soil at different locations on the hillslope. Distribution might also be affected severely by erosion due to removal of soil that was collected as sediment. Answering the final research question will give information about the role of erosion affecting PBDE concentrations. In order to provide a scientific answer to the research questions, it was

(8)

7

necessary to investigate a CO with similar characteristics to the BU. Comparing PBDE concentrations with statistical tests, at both sites, may prove that these chemicals are still used in the process of extinguishing forest fires. The following hypotheses are formulated:

- H1: PBDE concentrations in the soil will be significantly higher in samples collected at the BU due to appliance of fire extinguishers and therefore chemical contamination occurs at the BU.

- H2: Erosion will cause significantly higher concentrations of PBDEs in sediment samples compared to samples on higher slope levels.

In chapter 3 the methodology used to answer the research question is explained. The following chapter provides the theoretical framework with information about firefighting and PBDEs.

2. Theoretical framework

This chapter contains extended background information regarding fire extinguishing and PBDEs. This information is required to obtain better understanding of the processes occurring in firefighting and the causes of soil chemical contamination.

2.1 Fire extinguishing

The European Commission publishes technical reports every year on forest fires, using the European Forest Fire Information System (EFFIS). The most recent report dates from 2014 and includes fire data for all countries contributing to EFFIS. Figures 2.1.1 and 2.1.2 indicate fire frequency and burnt area trends in Spain (Schmuck et al., 2014). Forest fires seem to have increased over the past decades, while the burnt area is not necessarily increasing. In order to learn more about preventing and controlling wildfires it is important to know the cause of fires. In a global assessment by the Food and Agriculture Organization it was stated that in 2006 over 90% of the forest fires in the Mediterranean were caused by humans, intentional as well as accidental (Martínez Ruiz et al., 2007). Fires also occur naturally through for example lightning strikes, but this does not occur frequently. In the Mediterranean over 50.000 people can be mobilized yearly to suppress fires and they can utilize over 300 aircrafts (Martínez Ruiz et al., 2007).

* Figure 2.1.1: Burnt areas caused by forest fires in Spain 1980-2014 (Schmuck et al., 2014.

(9)

8

* Figure 2.1.2: Fire frequency in Spain 1980-2014 (Schmuck et al., 2014).

Aerial firefighting was used to extinguish the fire in Azuébar, but additional ground extinguishing is always required. It is not possible to stop a fire completely by airplane (Eastaugh & Molina, 2011). Aerial firefighting is very expensive and not efficient. The method however can be very useful when fighting fires on inaccessible terrain. Efficiency of aerial firefighting is reduced due to aerodynamics of water and wind velocity (Ito et al., 2010). Ito et al. (2010) proved through an experimental setup that distribution of fire extinguishers can be severely affected by relatively low wind velocity. So the distribution of flame retardants may be altered through meteorological conditions.

2.2 Polybrominated diphenyl ether flame retardants

PBDEs are chemical compounds added to polymers and do not occur naturally (Alaee et al.,2003). The compounds are composed of two phenyl rings that are bound together by an oxygen atom. Ten hydrogen atoms are attached to the phenyl rings and these atoms can be substituted by bromine atoms. This results in the fact that mono-, di-, tri-, tetra-, penta-, hexa-, hepta-, octa-, nona-, and decabromodiphenyl ethers can be produced (Rahman et al., 2001). Due to isomerism a total of 209 PBDE congeners is possible. Figure 2.2.1 (Rahman et al., 2001) illustrates the chemical structure of PBDEs in which x and y represent the number of bromine atoms. X and y both have a maximum of 5. The asterisks in the figure indicate the carbon atoms to which bromine atoms are most likely to bind (Rahman et al., 2001). During this research the abundance of some tetra-, penta-, hexa-, hepta-BDE congeners is assessed in the soil samples.

(10)

9

PBDEs are chemicals that are difficult to solve in water and therefore not easily transportable (Hale et al., 2003). Due to this low volatility the chemicals are expected to accumulate in soil and sediments. (Hale et al., 2003). The reactive component of PBDEs is bromine, which is a halogen found in the form of an inorganic salt. (Alaee et al., 2003). The halogen is known to be toxic, persistent and bio-accumulating in the environment (Segev et al., 2009). Due to the process of photolytic degradation in soil and sediments PBDEs can break down into organic compounds that can increase toxicity (Segev et al., 2009). Also an increase of pH value in soil, which can be caused by forest fires, might increase the solubility of PBDEs and can therefore stimulate groundwater contamination (Segev et al., 2009). Bioaccumulation is also a source of health risk. Due to bioaccumulation of PBDEs species higher up in the food chain can contain high levels of PBDEs. These species can be processed to food for humans causing health issues (Boon et al., 2002). PBDEs also affect microbial activity in the soil and might decrease fertility (Zhu et al., 2010). The research by Zhu et al. (2010) concluded that the application of deca-BDE to a soil affects the composition of the bacterial community and decreases urease concentrations. Altogether the different characteristics of the PBDEs appear to influence the soil and human health in a negative way and therefore it is necessary to assess whether PBDEs are still used in the process of fire extinguishing and consequently accumulating in the soil.

BDEs are produced industrially in the form of three different mixtures: Penta- octa- and deca-BDE mixture. These mixtures consist of different congeners and are classified according to average bromine atoms (Alaee et al., 2003). The penta-BDE mixture thus contains mostly lower BDEs such as tetra-, penta- and hexa-BDEs. Production of PBDEs increased globally since 1970 and in 1999 global production of these mixtures exceeded 200,000 metric tons (Alaee et al., 2003). Deca-BDE mixtures accounted for over 50% of the production. In natural environments such as soils and sediments debromination is possible (Lee & He, 2010). This means that BDEs with higher bromine content can lose bromine atoms and change to BDEs with a lower number of bromine atoms. Debromination can be caused by microbial activity and can be harmful for the environment because lower BDEs are considered to be more toxic (Lee & He, 2011). It is not known which of the BDE-mixtures is used for firefighting. The flame extinguisher that was most likely used for extinguishing the fire near Azuébar is Fire-Trol 931, which according to the company that produces it contains ammonium polyphosphate, clay, a corrosion inhibitor and coloring agents (ICL BIOGEMA S.A.S, 2013). Specified composition information on the Fire-Trol 931 is however not available. According to Koufopoulou et al. (2015) the Fire-Trol 931 causes soil hardening, increased electrical conductivity and decreased pH-values of leaching water and therefore has a negative environmental impact. Companies producing fire extinguishing materials refuse to be transparent about the precise composition of the materials and thus it is extremely difficult to assess the environmental impact of the extinguishers.

(11)

10

3. Methodology

In this chapter the methodology of the research is described. The most important part of the methodology is the GC/MS but before this could be performed an extensive preparation step was necessary. The preparation included sample selection, sample weighting, accelerated solvent extraction (ASE), evaporation steps and column cleanup. The GC/MS apparatus requires a clear transparent liquid of a few microliters, while the samples initially consisted of milled soil particles. After preparation and GC/MS the obtained data had to be processed through analysis of chromatograms and calculations. The research was combined with a research investigating polycyclic aromatic hydrocarbons (PAHs) concentrations by Esther Brakkee and therefore the used extraction methods had to be suitable for both BDEs and PAHs.

3.1 Sample overview

Due to time limitations it was possible to analyze only a small set of samples. The sample choice was based on location characteristics. Due to the concave character of the slopes it was expected that highest concentrations of BDEs would occur in the middle section of the slope. Figure 3.1.1 contains a schematic overview of the slope (Campo, 2014). The figure indicates that the slope is divided in three slope levels: Erosion, transport and deposition. Furthermore the sediment fences at the bottom are indicated. On the right figure the sampling points and transportation routes are illustrated. Due to the concave character of the slopes it was expected that highest concentrations of BDEs would occur in the middle section of the slope. A total of 32 samples were analyzed during the research (Table 3.1.1). First two blank samples using clean heated sand were prepared. From the BU and CO both twelve samples were chosen. The samples were chosen based on soil depth, vegetation cover and slope level. Sediment samples from the four different rainfall events at the BU were also included. Finally one of the samples of the BU was analyzed in triplicate. From the soil samples two grams was weighted and prepared for ASE. For the sediment samples the weight was adjusted to the organic matter content to prevent sample turbidity for GC/MS.

* Figure 3.1.1: Schematic overview of the slope indicating slope levels, canopy, transport routes and sediment fences (Campo, 2014).

(12)

11

The 32 samples were spiked with 50μl of 13_{C isotope BDE and PAH mixtures, which were}

used as internal standards in the GC/MS. The PAH mixture was added for another project and analysis of PAHs will not be discussed in this report. The BDE mixture contained 13_C

isotope concentrations of BDE-47, BDE-99 and BDE-153. These are tetra-, penta- and hexa-BDE congeners. The concentrations were respectively 83.5, 100 and 101 ng/ml. The internal standard BDEs have higher masses than normal BDEs with 12_{C isotopes and}

do not occur natural, therefore they can be used as a reference tracer for BDE concentrations with GC/MS. In table 3.1.1 an overview of the samples is given. The table includes the variables site, slope level, depth and vegetation cover. Furthermore the sample weight, internal standard addition and comments are included in the table.

Sample Site Slope Level Depth Canopy Msoil(g) IS addition (μl) Comment 1 NA NA NA NA NA 50 Blank sample 2 NA NA NA NA NA 50 Blank sample 3 Burned Erosion 2 - 5 UC 2.01 50 Triplicate 4 Burned Erosion 2 - 5 UC 2.00 50 Triplicate 6 Burned Erosion 0 - 2 UC 1.99 50 - 5 Burned Erosion 2 - 5 UC 2.01 50 - 7 Burned Erosion 0 - 2 BS 2.00 50 - 8 Burned Erosion 2 - 5 BS 2.02 50 - 9 Burned Transport 0 - 2 UC 2.00 50 - 10 Burned Transport 2 - 5 UC 2.00 50 - 11 Burned Transport 0 - 2 BS 2.00 50 - 12 Burned Transport 2 - 5 BS 1.99 50 - 13 Burned Deposition 0 - 2 UC 2.00 50 - 14 Burned Deposition 2 - 5 UC 2.00 50 - 15 Burned Deposition 0 - 2 BS 2.04 50 - 16 Burned Deposition 2 - 5 BS 2.00 50 - 17 Burned Sediment NA NA 1.00 50 Collected nov-14 18 Burned Sediment NA NA 1.50 50 Collected mar-15 27 Burned Sediment NA NA 1.50 50 Collected jun-15 28 Burned Sediment NA NA 2.00 50 Collected nov-15 29 Control Erosion 0 - 2 UC 2.00 50 - 26 Control Erosion 2 - 5 UC 1.99 50 - 25 Control Erosion 0 - 2 BS 2.02 50 - 30 Control Erosion 2 - 5 BS 2.01 50 - 31 Control Transport 0 - 2 UC 2.02 50 - 32 Control Transport 2 - 5 UC 2.01 50 - 23 Control Transport 0 - 2 BS 2.01 50 - 24 Control Transport 2 - 5 BS 2.02 50 - 19 Control Deposition 0 - 2 UC 2.00 50 - 20 Control Deposition 2 - 5 UC 2.01 50 - 21 Control Deposition 0 - 2 BS 2.00 50 - 22 Control Deposition 2 - 5 BS 2.03 50 -

* Table 3.1.1: Overview of the samples chosen in this research including the weight of the samples.

The soil samples were milled into particles smaller than 2 mm and approximately two grams of the samples was put into specified cells for ASE. The cells consisted of a metal housing to withstand high temperatures and were filled with the soil sample and additional clean sand silicate separated by cellulose filters. A schematic overview of the cell is illustrated in figure 3.1.2. Composing the cells was a complex task requiring precision and concentration in order to prevent contamination of the samples. Once the samples were prepared they were suitable for ASE.

(13)

12

* Figure 3.1.2: Schematic illustration of ASE cell with on the right an overview of the components inside the cell.

3.2 Accelerated solvent extraction

The next step of the sample preparation for GC/MS consisted of accelerated solvent extraction (ASE). The aim of such an extraction is a matrix exchange: Convert a solid sample into a liquid sample (Giergielewicz-Możajska et al., 2001). Previously such extractions consumed a lot of time and required large amounts of extraction fluids. Furthermore undesired losses of compounds could occur with older methods such as shake flask extraction and Soxhlet extraction (Giergielewicz-Możajska et al., 2001). Accelerated solvent extraction is an effective method for extraction. With ASE compounds in a solid sample can be extracted using an extraction fluid under high temperature and high pressure (Richter et al., 1996). A cell containing the soil sample is placed in the ASE apparatus and extracted. The extract is then purged into a collection vial using a compressed gas (Richter et al., 1996). The cell is located in an oven where the temperature exceeds the boiling point of the extraction fluid, therefore high pressure is needed to prevent evaporation of the extraction fluid (Giergielewicz-Możajska et al., 2001). The process of ASE is based on desorption of analytes in the solid soil sample. Due to high temperatures strong chemical interactions, such as van der Waals forces, are weakened, allowing analyte desorption from the sample (Giergielewicz-Możajska et al., 2001). Under pressure the liquid with adsorbed analytes is then moving through the sample due to diffusion processes (Giergielewicz-Możajska et al., 2001). A schematic overview of accelerated solvent extraction is included in appendix II. The final product of the ASE is a vial containing an unknown amount of extraction fluid with solved analytes. In this research hexane-acetone (1:1) mixture was used as extraction fluid. The cells were extracted under a temperature of 100ºC and a pressure of 1500psi during 15 minutes. The compressed gas used was nitrogen. The duration of one sample extraction was approximately 30 minutes and the ASE apparatus was rinsed with hexane-acetone (1:1) between samples. The obtained liquid was transferred from the vial into a pear-shaped flask. The vials were rinsed with hexane twice to ensure conservation of analytes in the samples.

(14)

13

3.3 Cleanup

After the ASE the samples had to be prepared for a cleanup column. Therefore an evaporation step was necessary. The pear-shaped flask was installed in a sand bath inside a fume hood. A Vigreux column was attached to the flask in order to reduce evaporation time. The sand bath was heated in order to evaporate the fluid to approximately 0.5ml. Boiling chips were added to the substance to prevent superheating, contributing to a safe method of evaporating. Once the fluid was evaporated to 0.5ml the sample was ready for the cleanup column.

The cleanup columns (figure 3.3.1) are used to remove dirt from the sample and induce the transparency of the liquid. The glass columns were filled with 1cm of folded glass wool at the bottom, 6.75cm of aluminum oxide (Al2O3) and 1cm of

sodium sulfate (Na2SO4). The glass wool is required

to provide a solid base for the column that does not drop when the other chemicals are added. The wool however has to be permeable for the fluid flowing through the column. The Al2O3 mixture consisted for

89% of active neutral Al2O3 and 11% of

demineralized water. This chemical, also known as alumina, is used for purification of the samples. Activated alumina has the ability to adsorb different chemical compounds including heavy metals (Sood et al., 1989). The activated alumina is characterized by its high surface area and pore volume, therefore it can serve as excellent adsorption material (Sood et al., 1989). The substance is also used for purifying drinking water through removal of fluoride, arsenic, phosphate and organic matter (Chen et al., 1989). On top of the alumina there was a thin layer of granulated anhydrous Na2SO4, which is often used as

a drying agent that removes water from the extract (Schenck et al., 2002). The samples were pipetted from the pear-shaped flask into the column and rinsed with another 0.5ml of hexane. The solution with hexane was then also added to the column. As soon as the extract had passed through the Na2SO4

another 25ml of hexane was added to the column to ensure passage of the sample through the column. The cleaned extract was again collected in a pear-shaped flask. All of the samples were clear and transparent after the cleanup procedure described above.

After the cleanup stage another evaporation step

was performed in order to obtain a final extract of 1ml. The evaporation procedure was similar to the one used before the cleanup. Again a sand bath was used to evaporate the extract in the pear-shaped flask. A Vigruex column was also used this time as a tool to reduce the evaporation speed. After obtaining the volume of 1ml the samples were pipetted into small vials and stored in a refrigerator to prevent further evaporation. Hexane was used again to rinse the flasks and avoid losses of analytes. Finally 100μl of * Figure 3.3.1: Clean up columns.

(15)

14

the extracts was pipetted into a vial suitable for GC/MS. The vials were sealed with aluminum foil that could be pierced by the GC/MS injector needle.

3.4 Gas chromatography / Mass spectrometry

The quantitative analysis of the BDE concentrations was performed through GC/MS. With this method the following tri- to hepta-BDEs could be identified: 28, 47, 85, 99, 100, 153, 154 and 183. For this research a TraceGC/MS apparatus was used provided by the University of Amsterdam. GC/MS consists of two parts: Gas chromatography and mass spectrometry. At first a small amount of sample, in this research 2μl, is injected into the gas chromatograph with an automatic sampler. The analyte is injected into a 60 meter long column made of glass. This column is located inside an oven which is heated according to a specified temperature program. Attached to the column is a high pressure gas cylinder that contains the carrier gas Helium. This carrier gas is known as the mobile phase (MP). The MP flows through the column. The different components of the analyte will be volatilized at different temperatures due to different boiling points and once they are volatilized they are carried by the MP (Patton et al., 1955). At the end of the column the different chemical compound elute at different times. The time at which a compound elutes is known as the retention time (RT). The elution order composed by Korytár et al. (2005) was used as a reference database for this research. The mass spectrometry follows the gas chromatography. At the end of the column a detector is located, which identifies the masses of the eluted compounds, giving a mass spectrum. The outputs of the GC/MS, consisting of a chromatogram and a corresponding mass spectrum, can be acquired and processed using special software on a computer. Figure 3.4.1 illustrates a schematic overview of the basic components of a GC/MS apparatus (BBC, 2014).

* Figure 3.4.1: Schematic overview of the GC/MS Apparatus (BBC, 2014).

To prevent degradation or loss of analytes the temperature in the oven should not be raised too quickly. For this research the following temperature program was used: The initial temperature was set at 60ºC and was then raised quickly to 110ºC with a rate of 75ºC/min. Thereafter the temperature increased to 150ºC at a rate of 12ºC/min and finally to 320ºC at a rate of 6ºC/min. At 320ºC the temperature was kept constant for 30 minutes. The complete temperature program had a duration of 64 minutes. A complete overview of the used method, with all settings, is included in appendix III. The method was composed with Xcalibur software provided by the University of Amsterdam. There are two types of methods possible for the GC/MS: A full range scan method and a selective ion monitoring (SIM) method. If samples are contaminated some extremely

(16)

15

high peaks can occur in the chromatograms, when using a full range scan method that identifies all possible masses of chemical compounds. Furthermore peaks of desired chemical compounds may disappear in the noise of the chromatogram when doing a full range scan. This problem occurred while looking for the different BDEs in this research and therefore it was decided to use a SIM method. With this SIM method certain masses that are typical for the BDEs and PAHs can be selected. In the chromatogram peaks will only appear when these masses are found. The masses chosen for the BDEs in the SIM method are listed in table 3.4.1. These masses were selected using mass spectra from earlier researches conducted at the University of Amsterdam. The table indicates the amount of bromine atoms in the BDE, the masses used for the SIM method and the carbon isotope that belongs to the mass.

BDE Mass (m/z) C-isotope BDE Mass (m/z) C-isotope

Tri 246.0 12 Penta 575.6 13 Tri 248.0 12 Hexa 481.7 12 Tri 405.5 12 Hexa 483.7 12 Tetra 326.0 12 Hexa 493.7 13 Tetra 338.0 13 Hexa 495.7 13 Tetra 485.7 12 Hexa 641.5 12 Tetra 498.0 13 Hexa 643.5 12 Penta 403.8 12 Hexa 653.0 13 Penta 405.8 12 Hexa 655.0 13 Penta 415.8 13 Hepta 561.6 12 Penta 417.8 13 Hepta 563.6 12 Penta 563.6 12 - - -

* Table 3.4.1: Masses chosen for selective ion monitoring method.

All 32 samples were injected into the GC/MS apparatus and analyzed with the SIM method. Besides the samples external standard sets were prepared to acquire reference concentrations for the final quantitative analysis. These external standards contained specified concentrations of 12_{C isotope compounds of all BDEs included in this study. Four}

external standards were prepared, containing 2, 5, 10 and 20 ng/ml of the BDEs. The exact concentrations of the different BDEs in the external standards are included in appendix IV. The external standards additionally contained the same concentrations of

13_{C BDE compounds as the samples. These concentrations could eventually be used to}

calculate the recovery of the 13_{C BDE compounds in the samples. The standard sets were}

processed four times by the GC/MS apparatus, creating a total of 16 standard runs. Thus a total of 48 runs were necessary to obtain the quantitative results for the BDE concentrations.

The output of the GC/MS consisted of chromatograms and corresponding mass spectra. A chromatogram and corresponding mass spectrum is included as appendix V. At first the chromatograms of the external standard runs were analyzed. These were used to make a list of the RTs for the different BDEs. The RTs were determined by looking at the mass spectra of the different peaks in the chromatogram and the elution order composed by Korytár et al. (2005). Table 3.4.2 indicates the RTs found for the BDEs. Knowledge concerning the retention times helped reducing the time of the analysis of the different chromatograms and mass spectra.

BDE Bromines Retention time BDE Bromines Retention time

28 Tri 25.21 min 100 Penta 31.23 min

47 Tetra 28.71 min 153 Hexa 35.08 min

85 Penta 33.41 min 154 Hexa 34.00 min

99 Penta 32.04 min 183 Hepta 38.68 min

(17)

16

3.5 Data processing

Processing the chromatograms of the samples and standards was an extensive time consuming process that was performed using the Xcalibur software. The processing consisted of designing layouts for different BDEs, designing processing methods, peak identification and calculations. In this paragraph the different components of the processing will be described.

First of all processing layouts were made for the different BDEs in the Qual browser of the Xcalibur software. The layouts were used to identify BDE peaks in a small section of the chromatograms. These sections were based on the retention times. In figure 3.5.1 the layout of BDE-99 is visualized. The top chromatogram indicates the original chromatogram, the middle chromatogram only shows the abundance of 12_{C BDE masses}

and the chromatogram at the bottom indicates only the 13_{C BDE masses. Due to such a}

layout it can be proven whether a peak represents a BDE. The layouts were used for the quantitative analysis of the chromatograms. In this figure it becomes clearly visible that some peaks in the chromatogram do not represent BDE-99, although the peaks are close to the retention time of BDE-99.

* Figure 3.5.1: Layout of BDE-99 designed with the Xcalibur Qual browser.

After the construction of the layouts a processing method was designed for automatically identifying peaks in the chromatograms. For every BDE such a method was made. Producing the methods consisted of assigning a retention time to a BDE and corresponding masses. A difference was made between 12_{C and}13_{C isotope BDEs and the} 13_{C BDEs were appointed as internal standards with known concentrations. For the}12_C

BDE processing methods the concentrations of the external standards were included and corresponding 13_{C internal standards were assigned. The processing methods then}

automatically identified the peaks and calculated the peak areas in the sample and standard chromatograms. Some of the peaks were not found automatically and had to be assigned manually with help of the layouts mentioned earlier. Manually assigning peaks was done in the Xcalibur Quan browser. The peak areas were eventually used to calculate the concentrations of the BDEs in the different sample. However while analyzing the peaks some things have to be taken into account: The limit of detection (LOD) and limit of quantification (LOQ). Shrivastava & Gupta (2011) wrote an article describing different

(18)

17

methods for determining the LOD and LOQ. The LOD is the minimum concentration of analyte that can be detected in analytical methods and LOQ is the minimum concentration that can be accurately and confidently assigned (Armbruster & Pry, 2008). The LOD and LOQ have to be considered while looking at chromatogram peaks. For this research the signal-to-noise method described by Shrivastava & Gupta (2011) was chosen. The method uses the ratio of the peak to the amplitude of the noise of the chromatogram. For the LOD the peak has to be three times higher than half the amplitude of the noise and for the LOQ the peak has to be ten times higher than half the amplitude of the noise. Figure 3.5.2 illustrates the signal-to-noise method (Shrivastava & Gupta, 2011). When a peak is encountered with a signal-to-noise ratio lower than 3:1 it cannot be determined whether a peak indicates a BDE. Figure 3.5.3 shows an example of a peak encountered in sample 16.

* Figure 3.5.2: LOD (bottom) and LOQ (top), where h is 2 * the amplitude of the noise

and H is the height of the peak (Shrivastava & Gupta, 2011).

(19)

18

Once the peaks are identified a calculation of the concentration in the analyte (C) can be made based on the peak areas of the analyte and the internal standard (Ai/As), a regression coefficient (Q), the sample weight (W) and the mass in the analyte and internal standard (Mi/Ms). The equation 3.5.1 obtained from Serné (2008) is as follows:

C = Mi / W, where Mi = Ms / Q * (Ai/As) [3.5.1]

The unknown values in the equation are Q and the mass of the BDE in the analyte (Mi) (Serné, 2008). The Q value can be obtained from a regression line based on the standard sets. In these standard sets specified amounts of 12_{C and} 13_{C BDEs are abundant}

providing a ratio M12/M13 for every BDE in each standard analysis. Furthermore an area ratio of A12/A13 can be calculated from these standard sets with the obtained peak areas. Figure 3.5.4 illustrates the regression line obtained from the standard sets for BDE-85. Four sets of four standards were used containing approximately 2, 5, 10 and 20 ng/ml 12_{C BDE-85 and 5.1 ng/ml}13_{C BDE-85. The peak areas where exported from the}

Xcalibur software. The graph was made with Microsoft Excel and indicates a regression coefficient (Q) of 0.8548. Similar regression graphs were made for every BDE. Knowing the Q value enables completion of equation 3.5.1. The calculation does not need recovery percentages because they are already included using the A12/A13 and M12/M13 ratios. The concentration calculated are given in ng/g which equals parts per billion (PPB). For example in sample 13 the concentration of BDE-85 was calculated as follows:

C = 5.05 / 0.8548 * (14429/29305) / 2.00, C = 1.45 ng/g

In the next chapter the results of the calculations for the different BDEs in the samples are given.

* Figure 3.5.4: Graph indicating the regression line of BDE-85 based on the standard sets. y = 0,8548x - 0,0139 R² = 0,9583 0,0 0,5 1,0 1,5 2,0 2,5 3,0 3,5 4,0 4,5 5,0 0,00 1,00 2,00 3,00 4,00 5,00 A 12/A13 M12/M13

Regression BDE-85

(20)

19

4. Results

In this chapter an overview of the results of the research is provided. The chapter consist of a paragraph indicating the results of the GC/MS and a paragraph describing the statistical analysis of the obtained results. The results will be discussed in the next chapter.

4.1 GC/MS results

Through GC/MS concentrations of BDEs 28, 47, 85, 99, 100, 153, 154 and 183 in soil samples, collected at two hillslopes near the municipality of Azuébar in Spain, were analyzed. The hepta- and hexa-BDEs 153, 154 and 183 were not discovered in any of the samples. Therefore only the results of the analysis of the tri- to penta-BDEs will be given in this section. The concentrations that were found are presented in table 4.1.1. NF indicates that there was no peak found and BQ means below limit of quantifications. Some peaks were too small to represent a reliable concentration according to the signal-to-noise concentrations.

BDE-28 BDE-47 BDE-85 BDE-99 BDE-100 Sample Site Slope level Soil

depth Canopy [] (ng/g) [] (ng/g) [] (ng/g) [] (ng/g) [] (ng/g) 1 NA NA NA NA NF BQ NF BQ BQ 2 NA NA NA NA NF BQ NF BQ NF 3 Burned Erosion 2 - 5 UC NF 1,5 NF BQ NF 4 Burned Erosion 2 - 5 UC NF 0,8 NF BQ NF 6 Burned Erosion 0 - 2 UC NF 0,7 4,6 BQ BQ 5 Burned Erosion 2 - 5 UC NF 1,4 NF BQ NF 7 Burned Erosion 0 - 2 BS NF 0,9 1,0 BQ 0,6 8 Burned Erosion 2 - 5 BS NF 0,9 NF BQ NF 9 Burned Transport 0 - 2 UC NF 1,1 5,6 0,6 BQ 10 Burned Transport 2 - 5 UC NF 1,0 1,3 BQ BQ 11 Burned Transport 0 - 2 BS NF 1,2 0,7 BQ 0,5 12 Burned Transport 2 - 5 BS NF 1,0 BQ BQ BQ 13 Burned Deposition 0 - 2 UC NF 1,0 1,5 BQ 0,4 14 Burned Deposition 2 - 5 UC 0,3 0,9 BQ BQ 0,6 15 Burned Deposition 0 - 2 BS 2,1 0,8 0,8 BQ NF 16 Burned Deposition 2 - 5 BS 2,1 0,7 BQ BQ NF 17 Burned Sediment 1 NA NA 2,8 2,2 11,4 0,6 0,8 18 Burned Sediment 2 NA NA NF 0,7 2,3 BQ NF 27 Burned Sediment 3 NA NA NF 1,3 1,7 0,7 0,8 28 Burned Sediment 4 NA NA 0,5 1,0 BQ BQ 0,4 29 Control Erosion 0 - 2 UC NF BQ NF NF NF 26 Control Erosion 2 - 5 UC NF 1,2 NF 1,0 0,8 25 Control Erosion 0 - 2 BS 0,6 1,3 NF 1,0 0,7 30 Control Erosion 2 - 5 BS 0,9 BQ NF BQ BQ 31 Control Transport 0 - 2 UC NF BQ NF NF NF 32 Control Transport 2 - 5 UC BQ BQ NF 0,5 BQ 23 Control Transport 0 - 2 BS 0,5 0,9 NF NF BQ 24 Control Transport 2 - 5 BS 1,0 0,7 NF NF BQ 19 Control Deposition 0 - 2 UC 0,3 0,9 NF BQ BQ 20 Control Deposition 2 - 5 UC 1,2 1,1 NF BQ BQ 21 Control Deposition 0 - 2 BS BQ 0,7 NF BQ BQ 22 Control Deposition 2 - 5 BS 1,0 BQ BQ BQ BQ

*Table 4.1.1: Concentrations of BDEs in samples, where NF = not found and BQ = Below

(21)

20

The table indicates that concentrations seem to be low and that BDE concentrations were not found or were too low to quantify in the majority of the samples. In order to make a comparison between the concentrations on the BU and the CO the results are also visualized as graphs in figure 4.1.1. In the graphs the NF and BQ values are treated as zeros. The y-axis indicates the concentration of the BDEs and is kept constant for each BDE. The x-axis consists of three variables: Slope level, vegetation cover and soil depth. For the vegetation cover UC indicates under canopy and BS means bare soil. Samples with common soil depth and vegetation cover have corresponding color bars for each slope level. It should be taken into consideration that concentrations might have been abundant in more samples but due to the LOQ there are no concentrations found.

(22)

21

* Figure 4.1.1: Graphs illustrating BDE-28, 47, 85, 99 and 100 concentrations on burned site and control site.

Besides the concentrations in the soil samples, concentrations were also found in the sediment samples, which are also listed in table 4.1.1. The concentrations are visualized in figure 4.2.2. On the x-axis the different sediment collection events are visible and on the y-axis the concentrations. Different colors represent the different BDEs.

* Figure 4.1.2: BDE concentrations in sediment samples for the different sediment collection events represented by stacked bars.

0,0 2,0 4,0 6,0 8,0 10,0 12,0 14,0 16,0 18,0 20,0

Nov -2014 Mar - 2015 Jun - 2015 Nov -2015

Con ce n tra tio n (n g/g)

Concentrations of BDEs in sediment samples

(23)

22

In order to obtain a more spatial visualization the concentrations were also processed in MATLAB R2015B software. The output consists of graphs indicating the different variables. Each graph represents the middle of the concave slope with on the y-axis the slope level and the soil depth at that level and on the x-axis the vegetation. The dots indicate the concentrations of BDEs on the locations using a color gradients indicated by the color bar. The graphs are presented in figure 4.1.3. In these graphs values below LOQ are treated as zeros. In appendix VI that were below LOQ are included. In order to discover whether there are significant differences between the variables site, slope level, vegetation cover and soil depth for BDE concentrations statistical tests have to be performed. In the next paragraph these tests and the results will be described.

(24)

23

* Figure 4.1.3: Spatial distributions of the BDE concentrations on the BU and CO.

4.2 Statistical analysis

To determine the significance of the results statistical tests were performed in MATLAB R2015B. The script that was composed is included in appendix VII. The script also contains the code that was used to create the graphs in figure 4.1.3. First of all a Lilliefors test was applied to investigate if the samples were normally distributed. This was not the case and therefore an analysis of variance (ANOVA) test was not applicable. A Kruskal-Wallis was instead used to compare the means of sampling groups. The variables site, slope level, vegetation cover and soil depth were compared using the t-test. The output of the test included a P-value for significance. A P-value below 0.05 indicates a significant difference in means with 95% confidence level. The results of the tests are presented in table 4.2.1. The P-values that prove significant differences are highlighted in yellow. NaN stands for not a number. NaN values appear because there were only values of NF or BQ at the control slope for BDE-85.

P-values Kruskal-Wallis

tests BDE-28 BDE-47 BDE-85 BDE-99 BDE-100 BU versus CO 0.209 0.092 0.003 0.266 0.544 Slope level BU 0.078 0.168 0.828 0.243 0.774 Slope level CO 0.675 0.764 NaN 0.241 0.113 Soil depth BU 0.674 1.000 0.013 0.317 0.294 Soil depth CO 0.135 0.740 NaN 0.528 0.902 Vegetation BU 0.401 0.423 0.135 0.317 0.848 Vegetation CO 0.135 0.868 NaN 0.528 0.902

* Table 4.2.1: Results of Kruskal-Wallis tests for significance. Significant P-values are

(25)

24

5. Discussion

In this chapter the previously presented results are discussed. The chapter also contains some recommendations for future research as well as remarks about restrictions and possibilities to improve the quality of this research.

When looking at the results it appears that some of the BDEs are abundant on both the BU and the CO. It was expected that BDEs would only be found on the BU because that is where the forest fire was extinguished, while on the CO there has not been a fire for at least twenty years. One explanation for the detection of BDEs on the CO can be a measuring or method error. Another factor that might have caused occurrence of BDEs on the CO is disturbance of fire extinguisher distribution by wind. As mentioned before relatively low wind velocity can influence the distribution of fire extinguisher when applied by airplane. In order to investigate whether climatic condition at the date of the fire influenced the distribution of fire extinguisher climate data was requested from the National Oceanic and Atmospheric Administration [NOAA] (2016). The NOAA has a database with hourly climate data from weather stations all over the world. For this research data was requested for the 28th_{of August 2014, from the nearest weather}

station Castellon-Almazora, which is located less than thirty kilometers northeast of the BU and CO. Data was received for the 28th_{of august between 12.00 PM and 9.00 PM with}

intervals of three hours (NOAA, 2016). The data included wind speed, wind direction, temperature and precipitation. The complete data table is added as appendix VIII. The fire was reported at 3.48 PM (Lainformacion, 2016) and therefore the 3.00 PM observation data is best suitable to investigate the implications of the wind velocity for the distribution of the fire extinguisher. At 3.00 PM wind velocity was measured at 6 miles per hour which is approximately 9.66 km/h. This velocity might cause disturbance of fire extinguisher distribution, however the wind direction was measured at 120º. This implicates that de wind direction was east-southeast. The CO is located northeast of the BU and therefore wind probably did not cause the fire extinguisher to reach the CO. At 6.00 PM wind velocity had decreased to 3.2 km/h and the direction had shifted to east-northeast reducing the probability of fire extinguisher distribution on the CO. It should be noted that the weather station is located 30 km from the sites, thus the data might not be representative for the BU and CO. Eljarrat et al. (2008) used reference soils from earlier studies for comparison in their research and found that concentrations of BDEs up to 0.84 ng/g can be found in reference soils. These concentrations are according to Eljarrat et al. (2008) attributable to atmospheric deposition. This might explain the abundance of low levels of BDEs on the CO.

When looking at the results of the statistical tests it becomes clear that only for BDE-85 significant results were found. Significantly higher concentrations of BDE-85 were abundant on the BU compared to the CO and on the BU concentrations were higher at a depth of 0-2 cm. BDE-85 levels, which is a penta-BDE, are also the highest of all BDEs with a maximum of 5.6 ng/g in soil samples and 11.4 ng/g in sediment samples. This might implicate that BDE-85 is used in fire extinguisher. According to La Guardia et al. (2006) BDE-85 accounts for only 2-3% of the most widely produced penta-BDE mixtures. The largest contributors to these mixtures are BDE-47, BDE-99, BDE-100, BDE-153 and BDE-154 (La Guardia et al., 2006). Concentrations of these BDEs were lower or not found on the BU. Therefore it is questionable whether a penta-BDE mixture was abundant in the fire extinguisher that was applied to the soil. It should be noted that the sampling group used in this research was relatively low. Concentrations were higher at a depth of 0-2 cm compared to 2-5 centimeter for BDE-85. This can be explained by the absence of rainfall in the period between the fire and sample collection. The fact that no rainfall occurred reduced the possibility of infiltration and therefore vertical distribution of BDE-85. During a fire water repellency of the soil increases so it is also unlikely that

(26)

25

water used for fire extinguishing infiltrated into the soil (DeBano, 2000). This might also explain the high concentration in the sediment sample collected after the first rainfall event. Due to increased water repellency erosion levels increase and therefore the sediments, which contain high levels of organic matter, show high levels of BDEs.

Vegetation did not significantly influence the distribution of BDEs in the soil. However the highest levels of BDE-85 were found under canopy. A research conducted by Huang et al. (2011) proved that vegetation induces the uptake of lower-BDEs in a contaminated soil. This would indicate that concentrations should be lower under canopy, however the vegetation abundant on the BU was burned and therefore not active. On a burned slope vegetation might operate as an obstacle for erosion as shown in figure 3.1.1 and thus vegetated areas can cause accumulation sites for eroded soil. This can be an explanation for the higher BDE levels under canopy. But how high are the observed levels compared to prior researches?

BDE concentrations in soils were mainly investigated near electronic waste recycling sites and on agricultural fields were sewage sludge is applied as a fertilizer, but also in remote areas. Hassanin et al. (2004) investigated BDE concentrations in rural and remote areas in the United Kingdom and compared these to concentrations in remote areas in Norway. In the United Kingdom concentrations of BDE-183 up to 7 ng/g were observed and even in remote areas in Norway 1.4 ng/g of BDE-99 was found (Hassanin et al., 2004). At an electronic waste recycling site in China concentrations of BDE-99 and BDE-100 ranged between 159 and 599 ng/g (Cai & Jiang, 2006) and at an acid leaching site elsewhere in China BDE-209 concentrations even exceeded 1000 ng/g (Leung et al., 2007). In Spain concentrations of BDE-209 in sewage sludge exceeding 1000 ng/g were discovered (Eljarrat et al, 2008). These concentrations are very high compared to the ones found on the BU and CO. The effect of such high concentrations on soil chemical contamination is however not broadly researched.

A method for analyzing the impact of soil contamination was developed by Parolini et al. (2012). Parolini et al. (2012) evaluated BDE concentrations in soil samples, vegetation and cow milk from a grazing site in the Italian Alps. Uptake of BDEs from the soil by vegetation lead to BDE concentrations up to 1.5 ng/g in cow milk (Parolini et al., 2012). Consequently can the abundance of BDEs in cow milk directly be linked to exposure of humans to BDEs. Grazing on a Mediterranean hillslope affected by a forest fire is possible but vegetation recovery is required. In southeast Spain soils suffer from water stress and vegetation recovery rates are typically low (Cerdá et al., 1995). Therefore analysis of contamination severity through concentrations of BDEs in vegetation cannot be performed shortly after the forest fire. Another method to indicate contamination was designed by Sellström et al. (2005). Accumulated concentrations of BDEs in earthworm lipids were discovered. Soils containing concentrations of 1.2-2.1 ng/g of BDEs led to 64 to 110 ng/g lipid in earthworms (Sellström et al., 2009). Earthworms are not largely abundant in the research area (Rutgers et al., 2016) and therefore it might be more difficult for BDEs to reach higher trophic levels. Another method for evaluation of contamination is looking at the effect of BDEs on microbial activity, which has been done in several studies (Lee et al., 2010;Liu et al., 2011;Zhu et al., 2010). These researchers however focused mainly on the application of BDE-209, which was not included in this study. In an experimental setup soils can be replicated, BDEs artificially added and measurements on microbial activity can be performed. Zhu et al. (2010) concluded that microbial behavior was altered by application of BDEs, but in this study the lowest added concentration of BDE-209 was 1000 ng/g.

(27)

26

In general it is difficult to quantify the magnitude of contamination of BDEs in a soil. Additional research on soil composition and microbial behavior is required to assess the effects of PBDEs. Future research could consist of adding the levels of BDEs, that were found in this research, to an artificial soil in a laboratory, where measurements on respiration rates and microbial population can be performed. Furthermore it would be wise to include the investigation of BDE-209 in the research. The deca-BDE congener is the most commonly used and the effects of the compound is broadly researched. It would also be interesting to look at the rate of vegetation recovery on a burned soil treated with BDEs compared to a burned soil that is not polluted. A suggestion regarding regulations on BDEs would be that governmental institution oblige companies to provide complete transparency on the composition of fire extinguishers. With this knowledge it would be much easier to assess the environmental impact of extinguishing chemicals. The following chapter contains conclusions that can be formulated by addressing the research question that were stated.

6. Conclusions

The first research question was: To what extent are PBDEs abundant on a burned slope where FRs were spread? The highest concentration of a single BDE in a sample was 5.6 ng/g of BDE-85. The sum of total BDEs in that sample was 7.3 ng/g, which does not seem to very high. Generally concentrations of BDEs were low on the burned slope and most of the concentrations found can probably be related to atmospheric deposition, because these relatively low concentrations were also found on the control site. The second research question addressed the distribution of BDEs over the slope. A scientifically robust answer cannot be given to this question, but BDEs seemed to be more abundant in the top layer of the soil. There was no clear pattern of BDE concentrations over the different slope levels, highest concentrations were found on the erosion and transport level, which does not seem logical. There was no rainfall event between the time of the event and the sample collection, therefore the distribution was not affected by erosion. The role of vegetation in PBDE distribution was also questioned during this research. Although highest concentrations were found under canopy, there is no significant difference between concentrations under canopy and on bare soil locations. Erosion does seem to be an important factor in the dynamics of PBDE transport on a burned slope. The sum of concentrations found after the first rainfall event was 18.0 ng/g which is much higher than in any soil sample. This implies that BDEs do not infiltrate into the soil easily and are transported downhill through erosion. Levels of contamination could therefore be much higher in places where water accumulates after a high severity rainfall event with surface runoff. This might also indicate that levels of contamination after extinguishing a forest fire are significantly higher on the bottom of a valley compared to a slope.

The main research question of this study was as follows: What is the role of PBDE flame retardants, used in the extinction of fires, on the chemical contamination of a Mediterranean soil at hillslope scale? To this question a comprehensive answer cannot be provided. It is first of all questionable whether a BDE mixture is abundant in the fire extinguisher due to the fact that BDE-85 concentrations were highest. The penta-BDE mixture contains twenty times more BDE-99 compounds and BDE-99 was found in lower concentrations in this research. The severity of chemical contamination is also difficult to address, because BDE studies mostly focus on the BDE-209 congener, which was not included in this research. Generally the concentrations of the different PBDEs seem too low and therefore the contamination might not be severe. However the abundance of PBDEs in a soil often leads to bioaccumulation and potentially hazardous levels of PBDEs in higher trophic levels.

(28)

27

7. Acknowledgements

I would like to thank the University of Amsterdam for providing the possibility to do such an interesting practical research. I especially would like to thank dr. Julian Campo for his guidance and enthusiasm during the complete duration of the project. I wish him the best of luck in completing his Postdoc. Finally I would like to thank dr. John Parsons for his expert supervision and lab technicians Joke Westerveld and Peter Serné for their guidance during the laboratory work. It has been a pleasure to work in the state of the art facilities of the Institute for Biodiversity and Ecosystem Dynamics.

(29)

28

8. References

Alaee, M., Arias, P., Sjödin, A., & Bergman, Å. (2003). An overview of commercially used brominated flame retardants, their applications, their use patterns in different countries/regions and possible modes of release. Environment international, 29(6), 683-689.

Armbruster, D. A., & Pry, T. (2008). Limit of blank, limit of detection and limit of quantitation. Clin Biochem Rev, 29, S49-52.

BBC (2014). Gas chromatography. Retrieved 23-06-2016, from:

http://www.bbc.co.uk/schools/gcsebitesize/science/triple_ocr_21c/further_chemistry/chromatography/revision/ 5/

Boon, J. P., Lewis, W. E., Tjoen-A-Choy, M. R., Allchin, C. R., Law, R. J., de Boer, J., ... & Zegers, B. N. (2002). Levels of polybrominated diphenyl ether (PBDE) flame retardants in animals representing different trophic levels of the North Sea food web. Environmental Science & Technology, 36(19), 4025-4032.

Cai, Z., & Jiang, G. (2006). Determination of polybrominated diphenyl ethers in soil from e-waste recycling site. Talanta, 70(1), 88-90.

Campo, J.M. (2014). Fire in the Mediterranean. Received by email.

Cerdà, A., Imeson, A. C., & Calvo, A. (1995). Fire and aspect induced differences on the erodibility and hydrology of soils at La Costera, Valencia, southeast Spain. Catena, 24(4), 289-304.

Chen, A. S., Snoeyink, V. L., & Fiessinger, F. (1987). Activated alumina adsorption of dissolved organic compounds before and after ozonation. Environmental science & technology, 21(1), 83-90.

DeBano, L. F. (2000). The role of fire and soil heating on water repellency in wildland environments: a review. Journal of Hydrology, 231, 195-206.

Eastaugh, C. S., & Molina, D. (2011). Forest road networks: metrics for coverage, efficiency and convenience.

Australian Forestry, 74(1), 54-61.

Eljarrat, E., Marsh, G., Labandeira, A., & Barceló, D. (2008). Effect of sewage sludges contaminated with polybrominated diphenyl ethers on agricultural soils. Chemosphere, 71(6), 1079-1086.

Fontana, A. R., Lana, N. B., Martinez, L. D., & Altamirano, J. C. (2010). Ultrasound-assisted leaching-dispersive solid-phase extraction followed by liquid–liquid microextraction for the determination of polybrominated diphenyl ethers in sediment samples by gas chromatography–tandem mass spectrometry. Talanta, 82(1), 359-366.

Giergielewicz-Możajska, H., Dąbrowski, Ł., & Namieśnik, J. (2001). Accelerated solvent extraction (ASE) in the analysis of environmental solid samples—some aspects of theory and practice. Critical Reviews in Analytical

Chemistry, 31(3), 149-165.

Hale, R. C., Alaee, M., Manchester-Neesvig, J. B., Stapleton, H. M., & Ikonomou, M. G. (2003). Polybrominated diphenyl ether flame retardants in the North American environment. Environment International, 29(6), 771-779.

Hassanin, A., Breivik, K., Meijer, S. N., Steinnes, E., Thomas, G. O., & Jones, K. C. (2004). PBDEs in European background soils: levels and factors controlling their distribution. Environmental science & technology, 38(3), 738-745.

Huang, H., Zhang, S., & Christie, P. (2011). Plant uptake and dissipation of PBDEs in the soils of electronic waste recycling sites. Environmental Pollution, 159(1), 238-243.

ICL BIOGEMA S.A.S (2013). FIRE-TROL® 931 LONG-TERM RETARDANT - AERIAL APPLICATION. Retrieved

04-18-2016, from: http://www.iclbiogema.com/fire-fighting/fire-retardants-and-fire-fighting-foams_99.html

Intergovernmental Panel on Climate Change. (2014). Climate Change 2014–Impacts, Adaptation and

Vulnerability: Regional Aspects. Retrieved 04-12-2016, from: https://www.ipcc.ch/report/ar5/wg2/

Ito, T., Kato, H., Goda, Y., Tagawa, S., & Negishi E. (2010). Water-Dropping Aerodynamics for Fire-Fighting Amphibian. 27th ICAS, Nice, 2010-5.

Kemmlein, S., Herzke, D., & Law, R. J. (2009). Brominated flame retardants in the European chemicals policy of REACH—regulation and determination in materials. Journal of Chromatography A, 1216(3), 320-333.

The effect of polybrominated diphenyl ether flame retardants on soil chemical contamination after a high severity forest fire on a Mediterranean hillslope