This article was downloaded by: [New York University] On: 26 May 2015, At: 05:44 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Environmental Technology Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/tent20
Influence of biofilms on heavy metal immobilization in Sustainable urban Drainage Systems (SuDS) a
a
ab
a
c
Marnie Feder , Vernon Phoenix , Sarah Haig , William Sloan , Caetano Dorea & Heather Haynes
d
a
University of Glasgow, College of Science and Engineering, Glasgow, Scotland G12 8QQ, UK b
University of Michigan, Department of Civil & Environmental Engineering, Michigan, 48109–2125, USA c
Département génie civil et génie des eaux, Université Laval, Québec (QC), Canada G1V 0A6
Click for updates
d
Heriot-Watt University, School of the Built Environment, Edinburgh, Scotland EH14 4AS, UK Accepted author version posted online: 18 May 2015.
To cite this article: Marnie Feder, Vernon Phoenix, Sarah Haig, William Sloan, Caetano Dorea & Heather Haynes (2015): Influence of biofilms on heavy metal immobilization in Sustainable urban Drainage Systems (SuDS), Environmental Technology, DOI: 10.1080/09593330.2015.1049214 To link to this article: http://dx.doi.org/10.1080/09593330.2015.1049214
Disclaimer: This is a version of an unedited manuscript that has been accepted for publication. As a service to authors and researchers we are providing this version of the accepted manuscript (AM). Copyediting, typesetting, and review of the resulting proof will be undertaken on this manuscript before final publication of the Version of Record (VoR). During production and pre-press, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal relate to this version also.
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions
Publisher: Taylor & Francis Journal: Environmental Technology DOI: 10.1080/09593330.2015.1049214
Influence of biofilms on heavy metal immobilization in Sustainable urban Drainage Systems (SuDS)
us c
University of Glasgow, College of Science and Engineering, Glasgow, Scotland G12 8QQ, UK
2
University of Michigan, Department of Civil & Environmental Engineering, Michigan, 481092125, USA Département génie civil et génie des eaux, Université Laval, Québec (QC), Canada G1V 0A6
4
Heriot-Watt University, School of the Built Environment, Edinburgh, Scotland EH14 4AS, UK
M an
3
ABSTRACT
ce
pt e
d
This paper physically and numerically models the influence of biofilms on heavy metal removal in a gravel filter. Experimental flow columns were constructed to determine the removal of Cu, Pb and Zn by gabbro and dolomite gravel lithologies with and without natural biofilm from SuDS. Breakthrough experiments showed that, whilst abiotic gravel filters removed up to 51% of metals, those with biofilms enhanced heavy metal removal by up to a further 29%, with Cu removal illustrating the greatest response to biofilm growth. An advection-diffusion equation successfully modelled metal tracer transport within biofilm columns. This model yielded a permanent loss term (k) for metal tracers of between 0.01– 1.05, correlating well with measured data from breakthrough experiments. Additional 16S rRNA clone library analysis of the biofilm indicated strong sensitivity of bacterial community composition to the lithology of the filter medium, with gabbro filters displaying Proteobacteria dominance (54%) and dolomite columns showing Cyanobacteria dominance (47%).
Ac
Downloaded by [New York University] at 05:44 26 May 2015
1
rip
t
Marnie Feder1, Vernon Phoenix1, Sarah Haig1,2, William Sloan1 Caetano Dorea3 and Heather Haynes4
KEYWORDS: biofilm, metals, SuDS, gravel filtration, advection-diffusion INTRODUCTION Sustainable urban Drainage Systems (SuDS) are designed to reduce the effects of urban development on the environment by temporarily storing or redirecting flood water, improving the quality of runoff and then safely discharging it into watercourses. Crucially, development and retrofitting of SuDS infrastructure is now a statutory requirement for
us c
ce
pt e
d
M an
Whilst filter drains are only designed to be active intermittently during rainfall events, they typically retain moisture between runoff events via surface tension around gravel particles or within pore spaces. Together with a stable substrate and high nutrient input from the runoff pollutants, this provides a conductive environment, which provides all the necessary resources for biofilms. Biofilm are an aggregation of microorganisms irreversibly attached to a solid surface and enclosed by a matrix of extracellular polymeric substance (EPS). EPS is can be the major component of the biofilm matrix and typically comprises 85% [13] of the total organic carbon of the biofilm. Both the EPS and the bacterial cells can enhance metal immobilization by increasing the surface area and enhancing the metal adsorption capacity of the surface on which they grow [14]. In addition to adsorption, biomineralization, bioaccumulation and biotransformation also contribute to bioremediation of toxic metals in the environment [15, 16] and thus may contribute to metal immobilization in SuDS. Although the bacterial populations of the biofilm are known to be influenced by the chemical composition of specific mineral groups [17], SuDS design manuals fail to provide any lithological specification. The material used in SuDS therefore varies geographically, despite recent research (see [18]) clearly showing filter removal efficiencies being dependent on material geochemistry and the natural geochemical changes associated with weathering processes. Thus, it is likely that both geochemical and biological processes contribute important, yet not well understood, processes in SuDS gravel filter removal efficiency.
Ac
Downloaded by [New York University] at 05:44 26 May 2015
rip
t
surface water runoff in many countries in the developed world (e.g. Water Environment (Controlled Activities) (Scotland) Regulations 2011). A widely implemented type of SuDS is a filter drain; this is a gravel filter trench designed to store and treat runoff from an adjacent roadway. The road runoff contains a variety of vehicular pollutants at concentrations commonly above regulatory limits, including suspended solids, polycyclic aromatic hydrocarbons (PAHs), and an array of heavy metals [1-8]. Although the underlying geochemical mechanisms responsible for pollutant treatment have received attention by the scientific research community [3, 5, 6, 9-11], studies tend to be site-specific and yield wildly varying pollutant removal efficiency data. This is hardly surprising, as there is no data on how the biology or lithology of filter drains affects their performance, nor is there a solid benchmarked data-set of metal removal efficiencies from well-controlled laboratory experiments. This means that current engineering design manuals typically report the overall performance of SuDS as only a qualitative or semi-quantitative treatment capacity range (i.e. 60-80%, 80-100%), with little reference made to the underlying removal mechanisms. [12].
Focusing on SuDS, the overall aim of this study is, therefore, to determine if bioremediation is a significant process in the immobilization of heavy metals in engineered filter gravels. Specific objectives are to: i) use breakthrough experiments to quantify the influence of lithology and biofilm on filter efficiency for metal immobilization; ii) characterize the biofilm community structure in natural SuDS via the creation of 16S rRNA clone libraries with an emphasis on determining the lithological influence on community composition.
MATERIALS AND METHODS Column design
rip
t
An up-flow vertical Perspex flow cell column (Figure 1) was constructed with internal diameter of 100mm and 135mm depth; this provided a 1 liter volume of angular gravel media. Four separate inlets were located at the base of the column and four outlets at the top, to permit mixing and dispersal of inflow and effluent in the adjacent mixing chambers. Two mesh diffuser plates separated the gravel filter from flow mixing chambers. A WatsonMarlow 323S/D peristaltic pump circulated water in the up-flow system with flow rates checked and recalibrated periodically to maintain accuracy.
pt e
Experimental setup
d
M an
us c
Natural SuDS gravel and receiving pond water was sourced from a working filter drain system located adjacent to a major road in Fenwick, Scotland (55°39'50.0"N 4°26'48.0"W). This material was compliant with Volume 1 design Specifications for Highway Works, Series 500 [19] in that “Type B graded filter drain material with a 40mm single size crushed rock be used in highway filter drains”. The collected gravel was crushed igneous gabbro that exhibited noticeable surficial biofilm growth. Material was placed into the sterilized and rinsed up-flow column and pond water was recirculated for a period of 10 months at room temperature (~20°C) with a diurnal cycle of natural sunlight (the experimental set-up was placed against a window); no nutrients were added to the column. This yielded a dense biofilm with dark green colouration suggesting it contained an abundance of phototrophic microbes. This biofilm was used as a stock biofilm to inoculate the following experiments.
ce
From the biofilm “stock” a sub-sample of biomass was used to inoculate the individual SuDS water influents (4 ml/min) of four identical up-flow columns (Experiments 1b-4b). Flow was recirculated for a period of 8 months, after which, the columns clearly exhibited visible biofilm growth. These experiments are described by the prefix Bio- in Table 1 and were prepared specifically to analyse bioremediation potential (Table 1). Thus, data output from these Bio-columns was quantitatively compared to a further four control experiments (Table 1 prefix Control-), of equivalent set-up but without biofilms (Experiments 1a-4a). Bio- and Control- experiments were both undertaken using two lithologies (Table 1); an igneous gabbro (G) and a sedimentary dolomite (D). These two lithologies were chosen as the gabbro is dominated by silicate minerals, while the dolomite is dominated by carbonate minerals. This enabled us to test the impact of lithology on metal immobilization and biofilm formation.
Ac
Downloaded by [New York University] at 05:44 26 May 2015
Biofilm Stock growth
Breakthrough experiments
us c
pt e
d
M an
For each experiment in Table 1, both a conservative tracer and subsequent heavy metal solution were run through the columns to produce a total of 16 breakthrough curves. Each set of breakthrough experiments was run for a total of 4.5 hours; this constituted a 3 hour pulse of tracer or metal solution and a further 1.5 hours without tracer or metal solution (i.e. unamended pond water) so as to determine the rate at which metal concentrations returned to zero. During breakthrough experiments, all columns were run with a flow rate of 15-16.5 ml/min, resulting in a residence time of 1 hour. 10mL samples were taken every 5 minutes from the outflow at the top of each column for concentration analysis. These were analysed for pH and conductivity, with heavy metal concentrations measured using Atomic Absorption Spectrometry (AAS; Perkin-Elmer AAnalyst 400). For breakthrough curve analysis, the AAS included triplicate sample analysis and a detection limit of 1.5 µg/L for Cu and Zn, whilst 15 µg/L for Pb (Perkin Elmer Manual).
ce
Breakthrough modelling
In order to determine the concentration of metals that passed through the column, percentages were obtained by calculating the area under the curve with the trapezoidal rule and comparing each experimental curves area to a theoretical breakthrough of 100%.
Ac
Downloaded by [New York University] at 05:44 26 May 2015
rip
t
The conservative tracer was the naturally high background sodium (Na) concentration in the SuDS pond water which provided electrical conductivity measurement when deionized water (DI) was used as a pulse during the breakthrough experiment. As this type of tracer is not immobilized by any biogeochemical processes, the breakthrough curve reveals transport due to advection and diffusion without permanent loss within the columns. In comparison, the heavy metal breakthrough experiments employed a 5ppm pulse of either Cu solution or a mixture (M+) of Cu, Pb, and Zn (Table 1) dissolved in SuDS pond water to assess permanent loss within the columns. These metals are the most common heavy metal pollutants in the road runoff [21-24] specific to gravel filter drains, with Cu typically the highest in concentration. These multi-metal M+ solutions were prepared by dissolving metal salts of Copper(II) nitrate (Cu(NO3)2 · 3H2O), Lead(II) nitrate (Pb(NO3)2) and Zinc nitrate (Zn(NO3)2 · 6H2O) (Sigma-Aldrich) in deionized water to achieve a stock solution of 1000ppm. This was then diluted to 5 ppm using SuDS water influent, so as to reflect the natural complexity of SuDS water chemistry
Transport of contaminants within porous media can be described by the advection diffusion equation for solutes within a homogeneous medium at steady state flow [20, 25-27]. To account for immobilization of heavy metals within the column, a linear loss term was incorporated into the diffusion-advection equation ∂C ∂ C ∂C =D −v − kc , ∂t ∂x ∂x
.
where Cr(x,t) is the resident concentration, D is the dispersion coefficient, v is the average pore water velocity, x is the distance down the column from the surface, t is time and k is the loss term. Boundary and initial conditions were given by, , = 0 = 0 0 < = 0,
= 0 3 < ≤ 4.5,
= ,
= 0 0 < ≤ 4.5,
rip
0 ≤ ≤ 3,
t
= 0,
=
≤ ,
M an
us c
A Matlab code used finite-difference approximations to solve this 1-D advection diffusion equation at discrete points through the column. The first order loss term effectively assumes that any heavy metal loss was due to permanent immobilization (i.e. permanent precipitation or adsorption with no opportunity to remobilize). Whilst the concentration profile was calculated throughout the column and a high temporal resolution the data comprised a breakthrough curve at x=L, thus the model parameters were selected to best fit the observed breakthrough curve. A unconstrained non-linear optimisation algorithm package in Matlab was used to systematically search for the best parameters. DNA extraction and clone library construction
ce
pt e
d
In order to characterize the community composition of the SuDS biofilm the 16S rRNA gene was analyzed by creating clone libraries [28]. A 16S rRNA clone library was created using the universal prokaryotic primers (27F: [5’- GAGTTTGATCCTGGCTCAG-3’] and 1392R: [5’ACGGGCGGTGTGTRC-3’]) on DNA extracted from 0.5 g (wet weight) of biofilm biomass using a Fast DNA Spin kit for soil (MP Bio-Medical, Cambridge, UK). The resulting PCR product was then used to make the clone library by using the TOPO TA kit (Invitrogen, UK) following the manufacturers instructions. One hundred clones were screened using amplified ribosomal DNA restriction analysis, with the restriction enzyme HAEIII (Promega,UK). Operational taxonomic units (OTUs) were identified, based on restriction cleavage patterns, and clones representing the different OTUs were sequenced (Genepool, Edinburgh). Taxonomy was assigned using RDP Classifier with a confidence interval of >90% [29]. All sequences were analyzed in the context of the complete data set using Bellerophon to identify PCR artifacts (chimeras), of which none were found. The resulting clone libraries were used to determine the bacterial community compositions of the SuDS biofilms with phyla distribution illustrated in Figure 6 and comprehensive classification of taxonomy found in the Appendix.
Ac
Downloaded by [New York University] at 05:44 26 May 2015
where Cin is the concentration at the inlet and L is the length of the column.
All sequences were used for Phylogenetic tree reconstruction (Appendix) with the MrBayes v. 3.2 software program [30]. A general time-reversible gamma-distributed rate variation model was specified. A Markov chain Monte Carlo analyses (MCMC) was performed for 106
generations with sampling every 1000 generations to create a posterior probability distribution of 7000 trees. The average standard deviation of split frequencies as well as convergence diagnostics for the posterior probabilities of bipartitions (Stdev(s)) and branch lengths potential scale reduction factor (PSRF) of Gelman and Rubin (31] were used in all cases to check for convergence. RESULTS AND DISCUSSION Breakthrough experiment analysis
us c
M an
pt e
d
Given that all columns exhibited 100% conservative tracer breakthrough (Figure 2), metal breakthrough analysis (Figures 3a-d) can confidently ascribe incomplete breakthrough to (bio)geochemical reactions within the filter media. This data is summarized in Table 2 in terms of the percentage of metal retention in both Bio and Control columns and analysed specific to lithological influence and biological influence. Geochemical metal removal
ce
The control experiments reveal lithological controls on metal immobilization. From Table 2 it is evident that gabbros removed a greater amount (28-40%) of metal than dolomites. This effect was most notable in Zn suggesting that gabbro exhibits preferential binding of this metal. Previous studies report that, under these experimental conditions, Zn is undersaturated while Pb and Cu are supersaturated with respect to their metal-hydroxide phases [18]. From this it was concluded that while Zn, Pb and Cu removal could be due to adsorption to the gravel media, in the case of Cu and Pb removal could also be due to precipitation of metal hydroxide phases [18]. Furthermore, the enhanced metal uptake of gabbro is believed due to surficial weathering of minerals to smectite clays that offer preferential surfaces for metal complexation due to the high cation exchange capacity and net negative surface charge of the clay [32]. Table 2 also suggests that there is a small amount (7%) of inhibition of metal immobilization in multi-metal solutions, compared to
Ac
Downloaded by [New York University] at 05:44 26 May 2015
rip
t
Figure 2 shows 100% breakthrough of the conservative tracer (DI) in all experiments. The control experiments (Figure 2a) exhibit near identical breakthrough profiles and, therefore, near identical dispersion. There is no lithological distinction, indicating equivalence of gravel geometry and packing. However, an apparent variability was observed in the breakthrough curves for columns affected by biofilm growth (Figure 2b). Reviewing the tail of these curves illustrates that gabbro columns reach saturation (such that when the measured concentration/initial concentration (C/C0) → 1) at higher pore volumes than dolomite columns, which may be an indication of differences in biofilm growth between the two lithologies. The slight variability between runs employing the same lithology (e.g. ~0.2 pore volumes at C/C0=0.5) likely reflects the influence of spatial heterogeneity of biological growth on tracer dispersion (Figure 2b).
single-metal solutions, presumably due to competition between metals for removal and/or adsorption sites. Biological metal removal
us c
M an
d
pt e
ce
Ac
Downloaded by [New York University] at 05:44 26 May 2015
rip
t
The bio-experiments removed up to 65% of metals through a combination of biogeochemical processes. Table 2 clearly shows that columns with biofilm immobilized up to 29% more metals than Control columns without biofilm. A pairwise t test showed that the difference between the Control columns (N = 8, M = 30.30, SD = 70.70) and the Bio columns (N = 8, M = 45.53, SD = 19.65) were statistically significant, t(7) = 4.49, p = 0.003, 95% CI [7.23, 23.270). Whilst bioremediation appears most effective for Cu removal, notable increases in metal binding of Pb (14%) and Zn (up to 8%) in columns with biofilm growth confirms, with the exception Zn in dolomite columns, that microorganisms sequester metals more readily than the gravel surface alone. This enhanced metal immobilization may be due to cellular processes such as bioaccumulation, [33] biomineralization and biotransformation and passive processes such as biosorption onto microbial cells [34]. Bacterial cell surfaces exhibit an abundance of anionic functional groups (e,g, carboxyl and phosphoryl groups) which facilitate adsorption of metals to the cell surface [35]. Furthermore, the EPS within the biofilm exhibits negative functional groups and therefore significant metal adsorption capacity [36-43]. In addition to adsorption processes, there is evidence to support surface mineralization and precipitation mediated by the biofilm community itself. Surface mineralization and precipitation of metals can occur due to pH changes [44], like those experienced during the eight month Bio experiments (pH changes from ≈7.5 to ≈8.5). Increases in pH can increase the saturation index of many metal hydroxide phases, thus facilitating metal hydroxide precipitation. Such an increase in hydroxyl ion concentration is the likely effect of photosynthesis by phototrophic microorganisms such as Cyanobacteria. While biofilms enhance metal removal in most cases, enhanced removal by biofilms is not seen for zinc in the dolomite gravel (Table 2). This is notable as biofilms in the gabbro system do enhance zinc removal. Evidently there is some lithological influence on the ability of biofilms to remove zinc. Just as the lithology might be expected to affect the composition of the biofilm communities, the different communities might likewise be expected to affect removal of specific contaminants. Zn has been shown to be toxic to specific groups of bacteria [45, 46], and this toxicity may have repercussions towards other bacterial groups within the community. With the gabbro biofilm demonstrating a higher concentration of DNA extracted from biomass when compared to dolomite, the greater bioremediation potential of gabbro may be a function of greater biofilm growth (which is also demonstrated from the breakthrough curves in Figure 2). Breakthrough curve modelling Table 3 summarizes the results of the advection diffusion model where the best fitting dispersion coefficient (D, m2/s) and the loss term for the metal tracers (k, (mg/l)/s) are estimated; note that k was fixed at zero for Control experiments since there is no
permanent immobilization of the conservative tracer within the column experiments. The root mean squared error (RMSE) is used to measure of how accurately the model predicts the observed breakthrough curves.
us c
M an
pt e
d
From Table 3 it is shown that gravel columns exhibited dispersion at values of 10-6, indicating approximate equivalence in all columns. Dispersion for the metal tracers in the Bio columns was between D = 2.32x10-6 – 4.34x10-6 (average 3.08x10-6) while Control columns D values were between 4.28x10-6 – 7.65x10-6 (average 5.47x10-6). This supports earlier breakthrough data (Figure 2), indicating that biofilm growth in porous media decreases pore space, velocity and hydraulic conductivity [41, 47-49] while increasing path length and friction.
ce
Effect of geochemistry on metal bioremediation The results of the 16S rRNA clone library and breakdown of phyla can be seen in Figure 6 while classification of taxonomy can be found in the appendix. Although the initial filter drain biofilm growth (Figure 6a) is Cyanobacteria rich (71%), their relative dominance reduces after the growth period (8 months) within the filter columns (Figure 6b and 6c). However, despite this reduction, Cyanobacteria remained an important component of the biofilm in both dolomite and gabbro systems. This enabled photosynthesis to drive up pH, thus likely facilitating metal hydroxide precipitation
Ac
Downloaded by [New York University] at 05:44 26 May 2015
rip
t
Table 3 clearly shows that a majority of RMSE values are low (
Environmental Technology Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/tent20
Influence of biofilms on heavy metal immobilization in Sustainable urban Drainage Systems (SuDS) a
a
ab
a
c
Marnie Feder , Vernon Phoenix , Sarah Haig , William Sloan , Caetano Dorea & Heather Haynes
d
a
University of Glasgow, College of Science and Engineering, Glasgow, Scotland G12 8QQ, UK b
University of Michigan, Department of Civil & Environmental Engineering, Michigan, 48109–2125, USA c
Département génie civil et génie des eaux, Université Laval, Québec (QC), Canada G1V 0A6
Click for updates
d
Heriot-Watt University, School of the Built Environment, Edinburgh, Scotland EH14 4AS, UK Accepted author version posted online: 18 May 2015.
To cite this article: Marnie Feder, Vernon Phoenix, Sarah Haig, William Sloan, Caetano Dorea & Heather Haynes (2015): Influence of biofilms on heavy metal immobilization in Sustainable urban Drainage Systems (SuDS), Environmental Technology, DOI: 10.1080/09593330.2015.1049214 To link to this article: http://dx.doi.org/10.1080/09593330.2015.1049214
Disclaimer: This is a version of an unedited manuscript that has been accepted for publication. As a service to authors and researchers we are providing this version of the accepted manuscript (AM). Copyediting, typesetting, and review of the resulting proof will be undertaken on this manuscript before final publication of the Version of Record (VoR). During production and pre-press, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal relate to this version also.
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions
Publisher: Taylor & Francis Journal: Environmental Technology DOI: 10.1080/09593330.2015.1049214
Influence of biofilms on heavy metal immobilization in Sustainable urban Drainage Systems (SuDS)
us c
University of Glasgow, College of Science and Engineering, Glasgow, Scotland G12 8QQ, UK
2
University of Michigan, Department of Civil & Environmental Engineering, Michigan, 481092125, USA Département génie civil et génie des eaux, Université Laval, Québec (QC), Canada G1V 0A6
4
Heriot-Watt University, School of the Built Environment, Edinburgh, Scotland EH14 4AS, UK
M an
3
ABSTRACT
ce
pt e
d
This paper physically and numerically models the influence of biofilms on heavy metal removal in a gravel filter. Experimental flow columns were constructed to determine the removal of Cu, Pb and Zn by gabbro and dolomite gravel lithologies with and without natural biofilm from SuDS. Breakthrough experiments showed that, whilst abiotic gravel filters removed up to 51% of metals, those with biofilms enhanced heavy metal removal by up to a further 29%, with Cu removal illustrating the greatest response to biofilm growth. An advection-diffusion equation successfully modelled metal tracer transport within biofilm columns. This model yielded a permanent loss term (k) for metal tracers of between 0.01– 1.05, correlating well with measured data from breakthrough experiments. Additional 16S rRNA clone library analysis of the biofilm indicated strong sensitivity of bacterial community composition to the lithology of the filter medium, with gabbro filters displaying Proteobacteria dominance (54%) and dolomite columns showing Cyanobacteria dominance (47%).
Ac
Downloaded by [New York University] at 05:44 26 May 2015
1
rip
t
Marnie Feder1, Vernon Phoenix1, Sarah Haig1,2, William Sloan1 Caetano Dorea3 and Heather Haynes4
KEYWORDS: biofilm, metals, SuDS, gravel filtration, advection-diffusion INTRODUCTION Sustainable urban Drainage Systems (SuDS) are designed to reduce the effects of urban development on the environment by temporarily storing or redirecting flood water, improving the quality of runoff and then safely discharging it into watercourses. Crucially, development and retrofitting of SuDS infrastructure is now a statutory requirement for
us c
ce
pt e
d
M an
Whilst filter drains are only designed to be active intermittently during rainfall events, they typically retain moisture between runoff events via surface tension around gravel particles or within pore spaces. Together with a stable substrate and high nutrient input from the runoff pollutants, this provides a conductive environment, which provides all the necessary resources for biofilms. Biofilm are an aggregation of microorganisms irreversibly attached to a solid surface and enclosed by a matrix of extracellular polymeric substance (EPS). EPS is can be the major component of the biofilm matrix and typically comprises 85% [13] of the total organic carbon of the biofilm. Both the EPS and the bacterial cells can enhance metal immobilization by increasing the surface area and enhancing the metal adsorption capacity of the surface on which they grow [14]. In addition to adsorption, biomineralization, bioaccumulation and biotransformation also contribute to bioremediation of toxic metals in the environment [15, 16] and thus may contribute to metal immobilization in SuDS. Although the bacterial populations of the biofilm are known to be influenced by the chemical composition of specific mineral groups [17], SuDS design manuals fail to provide any lithological specification. The material used in SuDS therefore varies geographically, despite recent research (see [18]) clearly showing filter removal efficiencies being dependent on material geochemistry and the natural geochemical changes associated with weathering processes. Thus, it is likely that both geochemical and biological processes contribute important, yet not well understood, processes in SuDS gravel filter removal efficiency.
Ac
Downloaded by [New York University] at 05:44 26 May 2015
rip
t
surface water runoff in many countries in the developed world (e.g. Water Environment (Controlled Activities) (Scotland) Regulations 2011). A widely implemented type of SuDS is a filter drain; this is a gravel filter trench designed to store and treat runoff from an adjacent roadway. The road runoff contains a variety of vehicular pollutants at concentrations commonly above regulatory limits, including suspended solids, polycyclic aromatic hydrocarbons (PAHs), and an array of heavy metals [1-8]. Although the underlying geochemical mechanisms responsible for pollutant treatment have received attention by the scientific research community [3, 5, 6, 9-11], studies tend to be site-specific and yield wildly varying pollutant removal efficiency data. This is hardly surprising, as there is no data on how the biology or lithology of filter drains affects their performance, nor is there a solid benchmarked data-set of metal removal efficiencies from well-controlled laboratory experiments. This means that current engineering design manuals typically report the overall performance of SuDS as only a qualitative or semi-quantitative treatment capacity range (i.e. 60-80%, 80-100%), with little reference made to the underlying removal mechanisms. [12].
Focusing on SuDS, the overall aim of this study is, therefore, to determine if bioremediation is a significant process in the immobilization of heavy metals in engineered filter gravels. Specific objectives are to: i) use breakthrough experiments to quantify the influence of lithology and biofilm on filter efficiency for metal immobilization; ii) characterize the biofilm community structure in natural SuDS via the creation of 16S rRNA clone libraries with an emphasis on determining the lithological influence on community composition.
MATERIALS AND METHODS Column design
rip
t
An up-flow vertical Perspex flow cell column (Figure 1) was constructed with internal diameter of 100mm and 135mm depth; this provided a 1 liter volume of angular gravel media. Four separate inlets were located at the base of the column and four outlets at the top, to permit mixing and dispersal of inflow and effluent in the adjacent mixing chambers. Two mesh diffuser plates separated the gravel filter from flow mixing chambers. A WatsonMarlow 323S/D peristaltic pump circulated water in the up-flow system with flow rates checked and recalibrated periodically to maintain accuracy.
pt e
Experimental setup
d
M an
us c
Natural SuDS gravel and receiving pond water was sourced from a working filter drain system located adjacent to a major road in Fenwick, Scotland (55°39'50.0"N 4°26'48.0"W). This material was compliant with Volume 1 design Specifications for Highway Works, Series 500 [19] in that “Type B graded filter drain material with a 40mm single size crushed rock be used in highway filter drains”. The collected gravel was crushed igneous gabbro that exhibited noticeable surficial biofilm growth. Material was placed into the sterilized and rinsed up-flow column and pond water was recirculated for a period of 10 months at room temperature (~20°C) with a diurnal cycle of natural sunlight (the experimental set-up was placed against a window); no nutrients were added to the column. This yielded a dense biofilm with dark green colouration suggesting it contained an abundance of phototrophic microbes. This biofilm was used as a stock biofilm to inoculate the following experiments.
ce
From the biofilm “stock” a sub-sample of biomass was used to inoculate the individual SuDS water influents (4 ml/min) of four identical up-flow columns (Experiments 1b-4b). Flow was recirculated for a period of 8 months, after which, the columns clearly exhibited visible biofilm growth. These experiments are described by the prefix Bio- in Table 1 and were prepared specifically to analyse bioremediation potential (Table 1). Thus, data output from these Bio-columns was quantitatively compared to a further four control experiments (Table 1 prefix Control-), of equivalent set-up but without biofilms (Experiments 1a-4a). Bio- and Control- experiments were both undertaken using two lithologies (Table 1); an igneous gabbro (G) and a sedimentary dolomite (D). These two lithologies were chosen as the gabbro is dominated by silicate minerals, while the dolomite is dominated by carbonate minerals. This enabled us to test the impact of lithology on metal immobilization and biofilm formation.
Ac
Downloaded by [New York University] at 05:44 26 May 2015
Biofilm Stock growth
Breakthrough experiments
us c
pt e
d
M an
For each experiment in Table 1, both a conservative tracer and subsequent heavy metal solution were run through the columns to produce a total of 16 breakthrough curves. Each set of breakthrough experiments was run for a total of 4.5 hours; this constituted a 3 hour pulse of tracer or metal solution and a further 1.5 hours without tracer or metal solution (i.e. unamended pond water) so as to determine the rate at which metal concentrations returned to zero. During breakthrough experiments, all columns were run with a flow rate of 15-16.5 ml/min, resulting in a residence time of 1 hour. 10mL samples were taken every 5 minutes from the outflow at the top of each column for concentration analysis. These were analysed for pH and conductivity, with heavy metal concentrations measured using Atomic Absorption Spectrometry (AAS; Perkin-Elmer AAnalyst 400). For breakthrough curve analysis, the AAS included triplicate sample analysis and a detection limit of 1.5 µg/L for Cu and Zn, whilst 15 µg/L for Pb (Perkin Elmer Manual).
ce
Breakthrough modelling
In order to determine the concentration of metals that passed through the column, percentages were obtained by calculating the area under the curve with the trapezoidal rule and comparing each experimental curves area to a theoretical breakthrough of 100%.
Ac
Downloaded by [New York University] at 05:44 26 May 2015
rip
t
The conservative tracer was the naturally high background sodium (Na) concentration in the SuDS pond water which provided electrical conductivity measurement when deionized water (DI) was used as a pulse during the breakthrough experiment. As this type of tracer is not immobilized by any biogeochemical processes, the breakthrough curve reveals transport due to advection and diffusion without permanent loss within the columns. In comparison, the heavy metal breakthrough experiments employed a 5ppm pulse of either Cu solution or a mixture (M+) of Cu, Pb, and Zn (Table 1) dissolved in SuDS pond water to assess permanent loss within the columns. These metals are the most common heavy metal pollutants in the road runoff [21-24] specific to gravel filter drains, with Cu typically the highest in concentration. These multi-metal M+ solutions were prepared by dissolving metal salts of Copper(II) nitrate (Cu(NO3)2 · 3H2O), Lead(II) nitrate (Pb(NO3)2) and Zinc nitrate (Zn(NO3)2 · 6H2O) (Sigma-Aldrich) in deionized water to achieve a stock solution of 1000ppm. This was then diluted to 5 ppm using SuDS water influent, so as to reflect the natural complexity of SuDS water chemistry
Transport of contaminants within porous media can be described by the advection diffusion equation for solutes within a homogeneous medium at steady state flow [20, 25-27]. To account for immobilization of heavy metals within the column, a linear loss term was incorporated into the diffusion-advection equation ∂C ∂ C ∂C =D −v − kc , ∂t ∂x ∂x
.
where Cr(x,t) is the resident concentration, D is the dispersion coefficient, v is the average pore water velocity, x is the distance down the column from the surface, t is time and k is the loss term. Boundary and initial conditions were given by, , = 0 = 0 0 < = 0,
= 0 3 < ≤ 4.5,
= ,
= 0 0 < ≤ 4.5,
rip
0 ≤ ≤ 3,
t
= 0,
=
≤ ,
M an
us c
A Matlab code used finite-difference approximations to solve this 1-D advection diffusion equation at discrete points through the column. The first order loss term effectively assumes that any heavy metal loss was due to permanent immobilization (i.e. permanent precipitation or adsorption with no opportunity to remobilize). Whilst the concentration profile was calculated throughout the column and a high temporal resolution the data comprised a breakthrough curve at x=L, thus the model parameters were selected to best fit the observed breakthrough curve. A unconstrained non-linear optimisation algorithm package in Matlab was used to systematically search for the best parameters. DNA extraction and clone library construction
ce
pt e
d
In order to characterize the community composition of the SuDS biofilm the 16S rRNA gene was analyzed by creating clone libraries [28]. A 16S rRNA clone library was created using the universal prokaryotic primers (27F: [5’- GAGTTTGATCCTGGCTCAG-3’] and 1392R: [5’ACGGGCGGTGTGTRC-3’]) on DNA extracted from 0.5 g (wet weight) of biofilm biomass using a Fast DNA Spin kit for soil (MP Bio-Medical, Cambridge, UK). The resulting PCR product was then used to make the clone library by using the TOPO TA kit (Invitrogen, UK) following the manufacturers instructions. One hundred clones were screened using amplified ribosomal DNA restriction analysis, with the restriction enzyme HAEIII (Promega,UK). Operational taxonomic units (OTUs) were identified, based on restriction cleavage patterns, and clones representing the different OTUs were sequenced (Genepool, Edinburgh). Taxonomy was assigned using RDP Classifier with a confidence interval of >90% [29]. All sequences were analyzed in the context of the complete data set using Bellerophon to identify PCR artifacts (chimeras), of which none were found. The resulting clone libraries were used to determine the bacterial community compositions of the SuDS biofilms with phyla distribution illustrated in Figure 6 and comprehensive classification of taxonomy found in the Appendix.
Ac
Downloaded by [New York University] at 05:44 26 May 2015
where Cin is the concentration at the inlet and L is the length of the column.
All sequences were used for Phylogenetic tree reconstruction (Appendix) with the MrBayes v. 3.2 software program [30]. A general time-reversible gamma-distributed rate variation model was specified. A Markov chain Monte Carlo analyses (MCMC) was performed for 106
generations with sampling every 1000 generations to create a posterior probability distribution of 7000 trees. The average standard deviation of split frequencies as well as convergence diagnostics for the posterior probabilities of bipartitions (Stdev(s)) and branch lengths potential scale reduction factor (PSRF) of Gelman and Rubin (31] were used in all cases to check for convergence. RESULTS AND DISCUSSION Breakthrough experiment analysis
us c
M an
pt e
d
Given that all columns exhibited 100% conservative tracer breakthrough (Figure 2), metal breakthrough analysis (Figures 3a-d) can confidently ascribe incomplete breakthrough to (bio)geochemical reactions within the filter media. This data is summarized in Table 2 in terms of the percentage of metal retention in both Bio and Control columns and analysed specific to lithological influence and biological influence. Geochemical metal removal
ce
The control experiments reveal lithological controls on metal immobilization. From Table 2 it is evident that gabbros removed a greater amount (28-40%) of metal than dolomites. This effect was most notable in Zn suggesting that gabbro exhibits preferential binding of this metal. Previous studies report that, under these experimental conditions, Zn is undersaturated while Pb and Cu are supersaturated with respect to their metal-hydroxide phases [18]. From this it was concluded that while Zn, Pb and Cu removal could be due to adsorption to the gravel media, in the case of Cu and Pb removal could also be due to precipitation of metal hydroxide phases [18]. Furthermore, the enhanced metal uptake of gabbro is believed due to surficial weathering of minerals to smectite clays that offer preferential surfaces for metal complexation due to the high cation exchange capacity and net negative surface charge of the clay [32]. Table 2 also suggests that there is a small amount (7%) of inhibition of metal immobilization in multi-metal solutions, compared to
Ac
Downloaded by [New York University] at 05:44 26 May 2015
rip
t
Figure 2 shows 100% breakthrough of the conservative tracer (DI) in all experiments. The control experiments (Figure 2a) exhibit near identical breakthrough profiles and, therefore, near identical dispersion. There is no lithological distinction, indicating equivalence of gravel geometry and packing. However, an apparent variability was observed in the breakthrough curves for columns affected by biofilm growth (Figure 2b). Reviewing the tail of these curves illustrates that gabbro columns reach saturation (such that when the measured concentration/initial concentration (C/C0) → 1) at higher pore volumes than dolomite columns, which may be an indication of differences in biofilm growth between the two lithologies. The slight variability between runs employing the same lithology (e.g. ~0.2 pore volumes at C/C0=0.5) likely reflects the influence of spatial heterogeneity of biological growth on tracer dispersion (Figure 2b).
single-metal solutions, presumably due to competition between metals for removal and/or adsorption sites. Biological metal removal
us c
M an
d
pt e
ce
Ac
Downloaded by [New York University] at 05:44 26 May 2015
rip
t
The bio-experiments removed up to 65% of metals through a combination of biogeochemical processes. Table 2 clearly shows that columns with biofilm immobilized up to 29% more metals than Control columns without biofilm. A pairwise t test showed that the difference between the Control columns (N = 8, M = 30.30, SD = 70.70) and the Bio columns (N = 8, M = 45.53, SD = 19.65) were statistically significant, t(7) = 4.49, p = 0.003, 95% CI [7.23, 23.270). Whilst bioremediation appears most effective for Cu removal, notable increases in metal binding of Pb (14%) and Zn (up to 8%) in columns with biofilm growth confirms, with the exception Zn in dolomite columns, that microorganisms sequester metals more readily than the gravel surface alone. This enhanced metal immobilization may be due to cellular processes such as bioaccumulation, [33] biomineralization and biotransformation and passive processes such as biosorption onto microbial cells [34]. Bacterial cell surfaces exhibit an abundance of anionic functional groups (e,g, carboxyl and phosphoryl groups) which facilitate adsorption of metals to the cell surface [35]. Furthermore, the EPS within the biofilm exhibits negative functional groups and therefore significant metal adsorption capacity [36-43]. In addition to adsorption processes, there is evidence to support surface mineralization and precipitation mediated by the biofilm community itself. Surface mineralization and precipitation of metals can occur due to pH changes [44], like those experienced during the eight month Bio experiments (pH changes from ≈7.5 to ≈8.5). Increases in pH can increase the saturation index of many metal hydroxide phases, thus facilitating metal hydroxide precipitation. Such an increase in hydroxyl ion concentration is the likely effect of photosynthesis by phototrophic microorganisms such as Cyanobacteria. While biofilms enhance metal removal in most cases, enhanced removal by biofilms is not seen for zinc in the dolomite gravel (Table 2). This is notable as biofilms in the gabbro system do enhance zinc removal. Evidently there is some lithological influence on the ability of biofilms to remove zinc. Just as the lithology might be expected to affect the composition of the biofilm communities, the different communities might likewise be expected to affect removal of specific contaminants. Zn has been shown to be toxic to specific groups of bacteria [45, 46], and this toxicity may have repercussions towards other bacterial groups within the community. With the gabbro biofilm demonstrating a higher concentration of DNA extracted from biomass when compared to dolomite, the greater bioremediation potential of gabbro may be a function of greater biofilm growth (which is also demonstrated from the breakthrough curves in Figure 2). Breakthrough curve modelling Table 3 summarizes the results of the advection diffusion model where the best fitting dispersion coefficient (D, m2/s) and the loss term for the metal tracers (k, (mg/l)/s) are estimated; note that k was fixed at zero for Control experiments since there is no
permanent immobilization of the conservative tracer within the column experiments. The root mean squared error (RMSE) is used to measure of how accurately the model predicts the observed breakthrough curves.
us c
M an
pt e
d
From Table 3 it is shown that gravel columns exhibited dispersion at values of 10-6, indicating approximate equivalence in all columns. Dispersion for the metal tracers in the Bio columns was between D = 2.32x10-6 – 4.34x10-6 (average 3.08x10-6) while Control columns D values were between 4.28x10-6 – 7.65x10-6 (average 5.47x10-6). This supports earlier breakthrough data (Figure 2), indicating that biofilm growth in porous media decreases pore space, velocity and hydraulic conductivity [41, 47-49] while increasing path length and friction.
ce
Effect of geochemistry on metal bioremediation The results of the 16S rRNA clone library and breakdown of phyla can be seen in Figure 6 while classification of taxonomy can be found in the appendix. Although the initial filter drain biofilm growth (Figure 6a) is Cyanobacteria rich (71%), their relative dominance reduces after the growth period (8 months) within the filter columns (Figure 6b and 6c). However, despite this reduction, Cyanobacteria remained an important component of the biofilm in both dolomite and gabbro systems. This enabled photosynthesis to drive up pH, thus likely facilitating metal hydroxide precipitation
Ac
Downloaded by [New York University] at 05:44 26 May 2015
rip
t
Table 3 clearly shows that a majority of RMSE values are low (
