Characterization of Stormwater Runoff in Lisbon Based on Microbial Source Tracking Methods

Gaspar Maria de Castelo Branco Teixeira de Queiroz (Bachelor)

Dissertation to obtain the Master’s Degree in Civil Engineering

Juri President: Professor Doutor Antonio´ Jorge Silva Guerreiro Monteiro Supervisor: Professora Doutora Filipa Maria Santos Ferreira Supervisor: Doutor Ricardo Jaime Pereira Rosario´ dos Santos Members: Professora Doutora Ana Fonseca Galvao˜ Doutora Sara Luisa Proenc¸a Garcia Cordeiro Dias Teles

October 2012 Cover Image: c WEB2DNA Art Project R Baekdal

This document was produced using LATEX “Why is my District death-rate low?” Said Binks of Hezabad. “Well, drains, and sewage-outfalls are My own peculiar fad. learnt a lesson once, It ran Thus,” quoth that most veracious man: —

“It was an August evening and, in snowy garments clad, I paid a round of visits in the lines of Hezabad; When, presently, my Waler saw, and did not like at all, A Commissariat elephant careering down the Mall.

I couldn’t see the driver, and across my mind it rushed That that Commissariat elephant had suddenly gone musth.

I didn’t care to meet him, and I couldn’t well get down, So I let the Waler have it, and we headed for the town.

The buggy was a new one and, praise Dykes, it stood the strain, Till the Waler jumped a bullock just above the City Drain; And the next that I remember was a hurricane of squeals, And the creature making toothpicks of my ﬁve-foot patent wheels.

He seemed to want the owner, so I ﬂed, distraught with fear, To the Main Drain sewage-outfall while he snorted in my ear— Reached the four-foot drain-head safely and, in darkness and despair, Felt the brute’s proboscis ﬁngering my terror-stiffened hair.

Heard it trumpet on my shoulder — tried to crawl a little higher — Found the Main Drain sewage outfall blocked, some eight feet up, with mire; And, for twenty reeking minutes, Sir, my very marrow froze, While the trunk was feeling blindly for a purchase on my toes!

It missed me by a fraction, but my hair was turning grey Before they called the drivers up and dragged the brute away.

Then I sought the City Elders, and my words were very plain. They ﬂushed that four-foot drain-head and — it never choked again!

You may hold with surface-drainage, and the sun-for-garbage cure, Till you’ve been a periwinkle shrinking coyly up a sewer.

I believe in well-ﬂushed culverts....

This is why the death-rate’s small; And, if you don’t believe me, get shikarred yourself. That’s all.”

KIPLING, “Municipal” in Departmental Ditties.

Acknowledgments

I should not wish to say another word before crediting those who, magna cum laude, have directly contributed to the successful completion of this work. In the midst of a worldwide economical crisis, I ﬁnd myself thankfully indebted to:

The unyielding enthusiasm of my Supervisors — Filipa and Ricardo — who have with great success always kept me under the weather.

The invaluable work of the Lab Team — Silvia, Teresa and Mario — in particular Silvia, for her tireless patience to explain the ‘abc’ of Microbiology, which turned out to have no ‘a’ in it.

The unstoppable runs of the Strike Team — Cecilia and Teresa — for all the rainwater they collected... in their hair.

The endless support of the Operations and Logistics Team — Kika and my Parents — for lending their time, car and fridge, for stormwater keeping.

And, of course, the irreplaceable contribution of our Sponsors and Data Suppliers — the IGIDL and the LNEC — for giving us the rain we danced for.

iii

Abstract

Recent studies acknowledge the great impact that stormwater can have in receiving waters, sometimes as great as wastewater, especially in urban contexts. Despite being widely studied, pollution modelling of urban stormwater runoff has proven an extremely difficult challenge due to a significant spacial and temporal variability, and a short understanding of the accumulation processes of diffuse source pollutants during dry weather. Current water quality evaluation is based, among others, in the level of faecal contamination, which is measured through faecal indicator bacteria, such as E. coli and enterococci. As pollution mitigation measures are currently converging towards treat-at-source solutions, it seems urgent to pinpoint the source of the detected faecal pollution, in urban environments. Microbial source tracking methods are promising tools in finding the source of any contamination, but despite their fast development and numerous applications, they haven’t yet been used to track faecal pollution in urban stormwater runoff. This study tested water samples from three different catchments in the city of Lisbon for E. coli and enterococci, as well as the COD parameter. It further developed the characterization of stormwater runoff by testing the presence of faecal contamination from Humans, Dogs and Cats — three species considered relevant in the urban context — through the PCR technique targeting species-specific mitochondrial DNA from the selected species. This study demonstrates a high degree of faecal pollution in stormwater runoff, in the vast majority of cases being well above the legal limit for discharges, and thus highlighting the need for stormwater treatment at WWTPs prior to discharge. Also, faecal pollution of any human, canine or feline nature was found in most of the tested samples.

Keywords

Stormwater runoff; microbial source tracking; faecal contamination; polymerase chain reaction; mitochondrial DNA.

Resumo

Estudos recentes reconhecem o grande impacto que as aguas´ pluviais podem ter nos meios receptores, por vezes equiparavel´ ao das aguas´ residuais, especialmente em contexto urbano. Apesar de amplamente estudadas, modelar escorrenciasˆ pluviais em termos de carga poluente e´ bastante dif´ıcil, dada a grande variabilidade espacial e temporal dos poluentes, e uma fraca compreensao˜ dos fenomenos´ de acumulaçao˜ de poluentes de origem difusa em tempo seco. A actual avaliaçao˜ da qualidade da agua´ e´ baseada, entre outros, no n´ıvel de contaminaçao˜ fecal, que e´ medido atraves´ de indicatores fecais bacterianos, como a E. coli e os enterococos intestinais. As medidas de mitigaçao˜ da poluiçao˜ temˆ recentemente convergido para soluçoes˜ de tratamento na origem, pelo que se torna urgente avaliar a origem da poluiçao˜ detectada em ambiente urbano. Os metodos´ de microbial source tracking permitem identificar as origens de uma dada contamina- çao,˜ mas apesar do seu rapido´ desenvolvimento e numerosas aplicaçoes,˜ ainda nao˜ foram utilizadas para localizar a origem da poluiçao˜ fecal em escorrenciasˆ pluviais urbanas. Este estudo testou varias´ amostras de escorrenciasˆ pluviais, recolhidas em tresˆ bacias diferentes na cidade de Lisboa, quanto a` presença de E. coli, enterococos intestinais e quantificaçao˜ do parametroˆ CQO. Aprofundou ainda mais a caracterizaçao˜ da poluiçao˜ em escorrenciasˆ pluviais testando a contaminaçao˜ fecal quanto a esta ter sido originada por Humanos, Caes˜ e Gatos — tresˆ especies´ consideradas relevantes ao contexto urbano — atraves´ da tecnica´ de PCR aplicada a sequenciasˆ espec´ıficas de DNA mitocondrial para as especies´ indicadas. Este estudo demonstra os elevados n´ıveis de poluiçao˜ em escorrenciasˆ pluviais para os tresˆ parametrosˆ analisados (CQO, E. coli e enterococos), na grande maioria dos casos bastante acima do limite legal para a descarga de efluentes, salientando a necessidade de tratamento das aguas´ pluviais em ETAR, antes da descarga. Na maioria das amostras foi detectada poluiçao˜ fecal de origem quer humana, felina ou canina quer uma combinaçao˜ das tres.ˆ

Palavras Chave

Escorrenciasˆ pluviais; microbial source tracking; contaminac¸ao˜ fecal; polymerase chain reaction; DNA mitocondrial.

vii

Contents

1 Introduction 1 1.1 Whys, Whats and Hows ...... 2 1.2 Thesis Outline — The plot ...... 4

2 Framing the Work 5 2.1 Phenomena and Parameters ...... 6 2.1.1 Basic Concepts ...... 6 2.1.2 Monitoring Parameters ...... 7 2.1.3 Advanced Concepts ...... 9 2.2 Urban Water Pollution ...... 12 2.2.1 BMPs and current approaches ...... 12 2.2.2 (in)Sanitary Engineering? ...... 18

3 State of the Art 19 3.1 Urban Stormwater Runoff ...... 20 3.1.1 Characterization ...... 20 3.1.2 Urban Response ...... 24 3.1.3 Modelling Runoff ...... 26 3.2 Dealing with rainwater in Portugal ...... 30

4 Microbiology 33 4.1 Escheri. . . what? ...... 34 4.1.1 A Microscopic Overview of Pathogenic Organisms ...... 34 4.1.2 Bacteria ...... 35 4.1.3 Protozoa, Helminths and Viruses ...... 36 4.2 Microbial Source Tracking Methods ...... 37 4.2.1 The Point of Indicators ...... 37 4.2.2 Tracking the Source ...... 38 4.2.3 The Methods ...... 38 4.2.4 Current Issues and Future Research ...... 40 4.3 Case-Study Techniques ...... 41 4.3.1 Polymerase Chain Reaction (PCR) ...... 41

ix 4.3.2 Mitochondrial Markers ...... 43

5 Case-Study 45 5.1 Objectives ...... 46 5.2 The Basins ...... 46 5.2.1 Alcantara Basin ...... 47 5.2.2 Bairro das Ilhas ...... 49 5.2.3 Madalena ...... 50 5.2.4 WWTP in Alcantara ...... 50 5.3 The Laboratory ...... 51 5.3.1 Collected samples ...... 51 5.3.2 COD analysis ...... 52 5.3.3 E. coli and enterococci analysis ...... 53 5.3.4 DNA extraction procedures ...... 53 5.3.5 Single and nested PCR procedures — Primer design ...... 54 5.4 Processing Results ...... 55 5.4.1 Rain Data ...... 55 5.4.2 Crossing Data ...... 57

6 Conclusions and Future Work 65 6.1 Epilogue ...... 66 6.2 Prospective Sequels ...... 68

Bibliography 69

Appendix A Article submitted to ENASB A-1

x List of Figures

2.1 Effects of Imperviousness in Runoff ...... 6 2.2 Urban Diffuse Pollution Sources ...... 7 2.3 Coliform Hierarchy ...... 8 2.4 First Flush phenomenon...... 10 2.5 Detection of First Flush using dynamic EMC ...... 12 2.6 Absence of First Flush using dynamic EMC ...... 12 2.7 Performance of EU countries in the implementation of the Water Framework Directive . 14 2.8 Urban farming in Cuba ...... 15 2.9 Inﬁltration Trench ...... 15 2.10 Small Detention Pond ...... 16 2.11 Rainwater Tank Solution ...... 17 2.12 Garden Rooftop Solution ...... 17

3.1 Effects of Street Sweeping ...... 25 3.2 Bypass Solutions in WWTPs ...... 26 3.3 Porous Pavement in Lisbon ...... 30 3.4 Examples of Inﬁltration and Detention Solutions in Lisbon ...... 30 3.5 Stormwater solutions in Almada ...... 31 3.6 Alcantara WWTP ...... 32

4.1 Typical bacterial structure ...... 35 4.2 Salmonella typhimurium ...... 36 4.3 Escherichia coli ...... 37 4.4 Current MST techniques ...... 39 4.5 PCR Cycles example ...... 42 4.6 Electrophoresis apparatus and result ...... 44

5.1 Streets and Stormdrains in Lisbon ...... 46 5.2 Experimental Basins in Lisbon ...... 47 5.3 Alcantara Basin ...... 48 5.4 Ilhas Basin ...... 49 5.5 Madalena Basin ...... 50

xi 5.6 Sample Boxes in Stormdrains; Sediment Deposition ...... 51 5.7 Presence of E. coli using the Colilert kit ...... 53 5.8 Rain Data from the Experimental Campaigns for 2011 ...... 56 5.9 Rain Data from Experimental Campaigns for 2012 ...... 56 5.10 Maximum COD values per basin and per event ...... 57 5.11 Maximum EC values per basin and per event ...... 57 5.12 Maximum EF values per basin and per event ...... 57 5.13 Frequency of COD, E. coli and enterococci values ...... 59

6.1 Helix Bridge in Singapore ...... 66 6.2 The “Double Helix II” ...... 67 6.3 “The Double Helix Mutation of Increased Compassion” ...... 68

xii List of Tables

2.1 Maximum discharge loads for COD, EC and EF ...... 13

3.1 Typical values for Stormwater runoff ...... 20 3.2 EMC values obtained in Isfahan, Iran ...... 21 3.3 Road and Roof EMC values for Genoa, Italy ...... 22 3.4 EMCs of stormwater pollutants obtained in Korea ...... 22 3.5 SMC values for stormwater in Johor, Malaysia ...... 22 3.6 Mean concentrations for microbiological parameters in Portugal ...... 23

4.1 Common waterborne pathogenic microorganisms ...... 34

5.1 Collection points in Alcantara ...... 48 5.2 Collection points in Ilhas ...... 49 5.3 Collection points in Madalena ...... 50 5.4 Primers used for both single and nested PCR, for each species ...... 55 5.5 PCR steps and cycle conditions ...... 55 5.6 COD, EC and EF values for Alcantara ...... 58 5.7 COD, EC and EF values for Bairro das Ilhas ...... 58 5.8 COD, EC and EF values for Madalena ...... 58 5.9 Final data for the Alcantara catchment ...... 60 5.10 Final data for the Ilhas catchment ...... 61 5.11 Final data for the Madalena catchment ...... 62 5.12 Positive/Negative Count for mtDNA testing ...... 63

xiii

Abbreviations

BMP - Best Management Practice

BOD - Biochemical Oxygen Demand

COD - Chemical Oxygen Demand

CWA - Clean Water Act

DNA - Desoxyribonucleic Acid

E. coli - Escherichia coli

FIB - Faecal Indicator Bacteria

IGIDL - Instituto Geof´ısico Dom Luiz

IRLS - Iteratively Reweighted Least Squares

LNEC - Laboratorio´ Nacional de Engenharia Civil

RMSE - Monte Carlo Markov Chain

MPN - Most Probable Number

MST - Microbial Source Tracking

NC - Nash-Sutcliffe Criterion

OLS - Ordinary Least Squares

PBS - Phosphate Buffered Saline

PCR - Plymerase Chain Reaction

RMSE - Root Mean Square Error

TAE - Tris-Acetate-EDTA buffer

TMDL - Total Maximum Daily Load

TN - Total Nitrogen

TP - Total Phosphorus

TSS - Total Suspended Solids

UV - Ultraviolet

WFD - Water Framework Directive

WWTP - Wastewater Treatment Plant

xv xvi 1 Introduction

Contents 1.1 Whys, Whats and Hows ...... 2 1.2 Thesis Outline — The plot ...... 4

1 1. Introduction

1.1 Whys, Whats and Hows

The second half of the twentieth century saw the birth of a new environmental philosophy — or environmental ethics — which is still to this date amply debated. It aims to provide tools for right decisions concerning our relation with animals and the natural world, where traditional ethical theories seem to have come short. It argues that Nature should not only be preserved for its potential benefits, and that it has intrinsic values beyond its natural resources that include, for example, its beauty. It strives to define our place in Nature, and proposes the challenge of overcoming anthropocentrism, and adopting a more holistic view — that of the ecological systems. Of course, there have been some issues derived from this that were fiercely discussed, for example: is our place in nature equal to that of other animals? — beavers build dams that disrupt other habitats, and all animals pollute through their waste, so what are the proper limits to our ’natural’ intrusion into Nature? Why then should we have the special duty of preserving other species? — or the attribution of a moral status to other animals — are all animals equal and worthy of the same respect? Some agree that the great apes are more similar to humans and worthy of the same respect, while ants, fleas and bacteria are not. And what of natural places? Places do not experience pleasure or pain, why should it be wrong to change them? On the other hand, they have an aesthetic value for other animals and us (Moran, 2012). This somewhat describes the current tension between environmentalism and society, but today we do all agree — even if for different reasons — that nature should be preserved to the maximum possible extent. In the last quarter of the twentieth century most countries developed an increasing concern over urban pollution, trying to limit the impacts on natural environments of both solid waste and wastewater. Many solutions were found and legislated and, in the particular case of wastewater, they tried to preserve the quality of receiving waters. These legislations set the standard in water quality for human consumption, bathing, and preservation of marine life and coastal areas — the Federal Pollution Control Act 2002 (USA) and the EU Water Framework Directive 2000/60/EC (Europe). More recent directives control the discharge of dangerous substances from any human activity (e.g. Directive 76/464/EEC, Directive 91/271/EEC and Directive 91/676/EEC), ensure the monitoring of receiving waters through frequent sampling (e.g. Directive 79/869/EEC), and of course constantly increase the quality standards of drinking water (Directive 80/778/EEC and Directive 98/83/EC). The preservation of receiving waters depends strongly on the quality of wastewater and stormwater discharges. The latter is a direct result of precipitation over an urban basin, and it drags all pollutants accumulated during dry weather: rooftop pollutants, traffic pollutants, single discharges of waste and pollutants from several other sources. A number of different studies has found that pollutant loads are strongly related to factors of three different natures: geomorphology of the catchment (Butler and Memon, 1999; cit. in Ferreira 2006); local climacteric conditions or precipitation regime over the basin (Gnecco et al., 2005); and the type and intensity of land use (Gray, 2004). Despite the growing concerns in the preservation of receiving waters, and the increasing de-

2 1.1 Whys, Whats and Hows mand in water quality standards, there is no legislation concerning stormwater in particular, and no mandatory control over stormwater runoff. Also, the monitoring parameters that set the quality of discharges are more often than not, merely indicative of the chemical or organical presence of pollutants/microorganisms. For example in microbiological monitoring, the parameters referred in the legislation are indicators of bacterial faecal pollution (e.g. E. coli and enterococci), and no monitoring of viral contaminations is usually performed. On the other hand, and because of the impact of stormwater flooding in urban areas, there has been a great development in source-oriented techniques, trying to control stormwater runoff. Some pollution control criteria have been associated with these techniques, even with the lack of general legislation and the complexity of urban runoff modelling. Several solutions are presented as having a good pollution-control behaviour, such as sedimentation ponds (Marsalek et al., 2002), retention basins (Jensen et al., 2011), infiltration trenches (Hatt et al., 2011) or rainwater tanks (Khastagir and Jayasuriya, 2010). There has also been a steeper inclination towards the reuse of stormwater in urban contexts (Chanan et al., 2010), for example for watering gardens, flushing toilets, street cleaning, or even nearby agriculture; and although sometimes predicted to meet a certain negative attitude from local societies, it stands as a successful best management practice in the use of water resources (McArdle et al., 2011). It is common for urban runoff to contain organic matter from vegetal sources, like leaves, but also faecal microorganisms, like bacteria, from animal excreta. In this light, there has been a recent but great development in Microbial Source Tracking (MST) methods — and a valuable tool in implementing best management practices (BMPs). Tracking the source of microorganisms allows for a better understanding of the water cycle in general and urban stormwater behaviour in particular. It can also settle legal disputes and monitor the entities that are polluting receiving waters. The most recent MST methods are based on the premise that different intestinal systems select different microorganism populations, due to diet and digestive differences of their hosts (Santo Domingo et al., 2007). Nowadays, MST methods are more and more supported by the amplification of species-specific DNA sequences of the targeted organism — a recent, inexpensive technique named Polymerase Chain Reaction (PCR). The PCR technique allows an easy and fast reproduction of a targeted DNA sequence, which one must ensure belongs only to the targeted group, sub-group or species. This is the most appealing feature of the technique, because in samples containing diluted or scarce DNA, like water samples, it is possible to achieve a detectable quantity of a certain DNA, in a known proportion, and then compare it to a control DNA sequence. Through the analysis of stormwater samples collected in three urban catchments in Lisbon, the present study seeks to: (i) evaluate the quality of stormwater runoff in the city of Lisbon, with special focus on faecal contamination and COD levels; (ii) further develop microbial source tracking methods, in particular the use of mitochondrial DNA markers designed specifically for species common to the urban environment (humans, cats and dogs).; (iii) assess the origin of registered faecal pollution in the city of Lisbon. The experimental basins in Lisbon are in Alcantara, Bairro das Ilhas and Madalena street, and they

3 1. Introduction are quite dissimilar in their topographic, morphologic and land use characteristics. Alcantara is the most plural basin of the three. It has residential areas and areas more dedicated to commerce. There are streets with intense traffic, and both steep and flat streets. Most streets have trees in the sidewalk, intensifying the presence of vegetal debris. Bairro das Ilhas is a more residential area showing clear contrasts with Alcantara. Its streets are one-way roads, with small commerce and traffic, and some pedestrian exclusive areas. The Madalena street is in the historical part of Lisbon, showing some intense traffic and commerce, though not so intense as in Alcantara. It shows some steep slopes, is mostly impervious, and presents no vegetation. The collection campaigns took place from November 2011 to July 2012 with the objective of collecting stormwater samples in urban contextes. Stormwater was captured in plastic boxes and taken to the laboratory to be submitted to COD analysis and presence of faecal indicator bacteria: Es- cherichia coli and enterococci using detection kits. Also DNA was extracted from each sample, and specific sequences of mitochondrial DNA were reproduced using PCR techniques, in order to match with the species-specific control markers for Humans, Cats and Dogs.

1.2 Thesis Outline — The plot

The following chapters will take the reader through the different stages of this work. They will start at broader sceneries and progressing to more focused and specific subjects, always trying to incrementally raise the reader’s awareness on this theme, be it that he is already familiar with the discussed concepts or not. The next three chapters are of the same nature. Chapter 2 will try to explain some concepts used by modern Sanitary Engineering in an integrated view of the urban water cycle, along with discussing some of its best management practices and water policies and legislations. Chapter 3 is dedicated to stormwater: it tries to summarize what is happening worldwide in terms of stormwater pollution; and reflects on some mitigation measures, used both worldwide and in Portugal. Chapter 4 is entirely focused on filling the blanks in the civil engineer’s Biology, going through which microorganisms are potentially pathogenic or indicate the presence of pathogens, and how to track them and know their source, and which are just another brick in the wall. Chapter 5 changes the tone of this thesis and describes the experimental work, the collection of data and design of laboratory tests. It also produces and interprets the results, but the discussion, final outcome and future propositions for further studies are left for Chapter 6. In the Appendix A an article (in Portuguese) submitted to a Sanitary Engineering Conference — the 15th ENaSB (Encontro Nacional de Saneamento Basico)´ — and presented in Evora,´ Portugal, in September 2012 by the author(s) is also included, given its relation to the present work and relevance to the theme.

4 2 Framing the Work

Contents 2.1 Phenomena and Parameters ...... 6 2.2 Urban Water Pollution ...... 12

5 2. Framing the Work

2.1 Phenomena and Parameters

2.1.1 Basic Concepts

The first important idea is that the urban water cycle is a very complex system. While the process of distributing water and collecting wastewater is fairly controlled and well understood, the management of rainwater in an urban area can prove a challenging problem, and simplified models are more often than not inaccurate and unpromising. Rain does not fall uniformly over space and time, and data is only collected by rain gauges in specific geographical points. The behaviour of surface water is mainly subject to slope, obstacles, infiltration rate and surface smoothness, but these in turn are dependant of a wide range of different factors, for example: land use, soil humidity and porosity, and previous dry weather conditions have a direct influence over infiltration rate, thus reducing or boosting the flow of surface water. The ability of a basin to retain surface water is of the utmost importance in urban rainwater management, for it reduces the peak flow arriving at a given point downstream of the basin.

Figure 2.1: Effects of imperviousness in surface runoff. Adapted from USEPA (1999).

The hydrological basin of a given section is an area where every drop of rain that falls over it, eventually converges to that same section (a similar definition is given by Quintela (1996)). It is the basin that conveys the water downstream and its multiple characteristics can result in fast or slow concentration times. In an urban context, surface water is generally conveyed into the sewage system, and slow concentration times are essential in preventing the overcharging of collectors and/or WWTPs. Also, part of the rainwater infiltrates underground, creating an underground flow that converges more slowly to the same section. Another part is retained in the surface, whether in holes or flat areas or impervious surfaces, gradually infiltrating and/or evaporating. Impeviousness can have a strong im-

6 2.1 Phenomena and Parameters pact in surface water flows — or runoffs — in urban areas, as suggested by the U. S. Environmental Protection Agency (USEPA, 1999) in Figure 2.1. Rainwater passing through surfaces drags all sorts of particles with it, from small stones, leaves and sand to salts, dissolved metals and microorganisms. These can enter the runoff from diffuse sources, through washing or deposition, but also through point-sources such as domestic or industrial discharges. However, unlike diffuse sources, pollution from point sources is relatively easy to quantify because the point of entry is fixed, and flow rates and concentrations are usually known (Al Bakri et al., 2008; Choe et al., 2002). Figure 2.2 illustrates several urban diffuse sources of pollution. Pollutants will affect severely the use of rainwater, and ultimately, the way it is disposed of. Non- polluted rainwater can be used for soil irrigation or street washing, or even to dilute wastewater in unitary sewer systems with high peak flow discharges. On the other hand, heavily polluted rainwater needs to be treated in a WWTP before being released into receptor bodies. In particular, the runoff generated in residential and industrial areas is highly likely to contain hazardous materials such as oil components and heavy metals as well as floating materials, which by requiring proper treatment in WWTPs, should be preceded by surveys characterizing runoff pollution loads (Choe et al., 2002). A wide palette of parameters has been established to assert the quality of water, and these serve as the principal legal instrument in water management in general and in the implementation of BMPs in particular.

Figure 2.2: Different diffuse sources in urban pollution. Adapted from Dotto (2006).

2.1.2 Monitoring Parameters

The most popular parameters fall into two categories: physical-chemical and microbiological parameters. The ﬁrst include those studied by Gondim (2008): temperature; pH levels; turbidity; quantity of Dissolved Oxygen (DO); Total Suspended Solids (TSS); Biochemical Oxygen Demand (BOD); Chemical Oxygen Demand (COD); and also include the relation between Nitrogen and Phosphorous levels, and the quantity of heavy metals, such as zinc (Zn), lead (Pb) and copper (Cu), among others

7 2. Framing the Work

(e.g. Gnecco et al., 2005; Al Bakri et al., 2008; Choe et al., 2002). While temperature, pH levels, turbidity or suspended solids are familiar and everyday concepts, the rest perhaps need a word or two to help understand them.

Oxygen Demand, Biochemical Oxygen Demand and Chemical Oxygen Demand. The concept of Oxygen Demand is to quantify the amount of oxydable matter present in the water, by measuring the amount of consumed oxygen. BOD measures the oxidation of organic matter into an inorganic form. COD on the other hand is also capable of measuring the chemical consumption of dissolved oxygen, and thus give an idea of the presence of chemical non-biodegradable components. Gondim (2008) stresses the importance of the combined use of COD and BOD, for its relation indicates the biodegradability of efﬂuents in wastewater: the closer these two parameters are, the smaller the fraction of non-degradable components, so the more biodegradable the efﬂuent is likely to be.

Microbiological parameters - coliform bacteria. The microbiological parameters try to evaluate the presence of pathogenic microorganisms in water. But they do not always succeed. The assessment of Total Coliforms (TC) and Faecal Coliforms (FC) was rather popular in the past as a way of characterizing the presence of pathogens in water. It consists in measuring the fermentation of lactose in a 24h-48h period. However, there are bacteria from the soil or vegetal organisms that will also be included in Total Coliform count, so TC was widely dropped as a pathogen indicator (Macedo,ˆ 2005; cit. in Gondim, 2008). Instead, the count of coliforms was directed to a speciﬁc segment of bacteria species that come only from the digestive system of animal species, the Faecal Coliforms.

Figure 2.3: The Golden Rectangle of coliform bacteria hierarchy.

The quantiﬁcation of Escherichia coli (E. coli) and intestinal Enterococcus — both in the Faecal Coliform group — has become the most popular method of pathogen presence assessment, and

8 2.1 Phenomena and Parameters while these bacteria are usually non-pathogenic, they are generally accepted as good pathogen- indicators (McCarthy, 2009; Brownell et al., 2007). In fact, the U. S. Environmental Protection Agency recommended in 1986 the use of E. coli and enterococci as faecal indicators in the monitoring of freshwater (E. coli) and marine waters (enterococci), replacing faecal coliforms in general, based on studies showing a greater correlation between these indicators and gastrointestinal diseases than that of faecal coliforms (USEPA, 1986; cit. in Jeng et al., 2005). Figure 2.3 tries to ilustrate the relationship between pathogens and the more broader groups of coliform bacteria.

2.1.3 Advanced Concepts

This section will address some of the more complex phenomena related to urban stormwater runoff. A large number of studies were published in the last decade, all over the world, trying to shed some light over some of these advanced concepts, for example the first flush phenomenon, or methods for choosing basins and sampling procedures. During the ’70s several studies tried to establish successful urban runoff models, basin choice criteria and methods of sampling and measurement (Desbordes et al., 1980). Today there is still no particular experimental model that stands out to describe stormwater runoff in urban catchments, as results of pollution studies throughout the world tend to vary with local climate, pollutant sources, traffic volume and waste management strategies (Chow et al., 2011). Most models tend to relate pollutant concentrations to runoff volumes and precipitation regimes, but all need intense site-calibration (Dembel´ e´ et al., 2010; Dotto et al., 2010; Kanso et al., 2003). Authors agree, however, that there are certain tools and phenomena that describe to an acceptable extent the behavior of stormwater in an urban context, such as the first flush effect, the event mean contration (EMC), or the land use and roof runoff contribution and effects.

First Flush. The concept of first flush is that pollutant concentration in stormwater has a peak in a rainfall event that traduces the wash-off effect of rain over accumulated diffuse pollution. Deletic (1998; cit. in Gnecco et al., 2005) and Bertrand-Krajewski et al. (1998; cit. in Gnecco et al., 2005) agree that the first portion of the runoff volume contains the greater part of the pollutant load. However, there are different and more detailed definitions, based on the chosen approach to assess the first flush phenomenon Gnecco et al. (2005), like the ratio between pollutant load and volume of runoff has to grow faster than a preset threshold. Bach et al. (2010) chose to define first flush as the volume at which pollutant concentrations reach their ’background concentration’ (i.e. the statistically significant minimum). Because first flush is a very site-specific phenomenon (Chow et al., 2011), several studies have failed in their attempt to clearly identity the phenomenon in a specific catchment under studied rainfall events (e.g. Bach et al., 2010; McCarthy, 2009). This has been an extensively studied phenomenon, despite the inexistence of a model that can fully explain it. Chow et al. (2011), amongst countless others, have studied the First Flush phenomenon in the runoff from a commercial area in south Johor, Malaysia, over six storm events, sampling several times

9 2. Framing the Work

Figure 2.4: Cumulative mass and volume plots for the analysed pollutants. Taken from (Chow et al. 2011).

10 2.1 Phenomena and Parameters during the same storm event in order to evaluate the evolution of each analysed parameter over time.

The authors have measured BOD, COD, TSS, oil and grease (O&G), NO3-N, NO2-N, NH3-N, Soluble Phosphorus, and Total Phosphorus (TP). They detected a rapid growth of accumulated mass in every pollutant tested, as shown in Figure 2.4.

Event Mean Concentration. The event mean concentration (EMC) is deﬁned as the ratio between the total constituent mass discharged during an event, and the total runoff volume (Huber, 1993; cit. in Chow et al., 2011), as shown by the relation

M R Q(t)C(t)dt EMC = = (2.1) V R Q(t)dt where M is total mass of pollutant during the entire runoff (kg), V is total volume of runoff (m3), C(t) is time varying pollutant concentration (mg/l), Q(t) is the time varying flow (L/s), and t is elapsed time in the storm event (s). It is a useful tool for relating pollution loads between rainfall events with different intensities. These estimates could also be useful in legislating and allocating maximum pollutant loads. However, as an average descriptor of the event’s pollution loads, it does not explain the first flush phenomenon. Also, Dembel´ e´ et al. (2011) criticize that the most frequent EMC models are incapable of describing, even grosso modo, the build-up and wash-off processes which have been recognized as most influential in determining pollutant concentration and loads. Kim et al. (2007) propose a dynamic EMC, which describes an accumulated load relation in each instant of the rainfall event, rather than simply relating total mass and volume of runoff. This new approach is described in Equation 2.2.

Pt=t Q(t)C(t) dynamicEMC = t=0 (2.2) Pt=t t=0 Q(t)

Their experiment in a highway in the metropolitan area of Daejon, Korea, shows a rapid decline of the dynamic EMC — for TSS, COD, Total Nitrogen (TN) and Total Phosphorus (TP) — within the early 20-50 minutes of the event, reflecting the first flush effect. The following figures show examples of the results of dynamic EMCs in two different situations, one with the occurrence of first flush (see Figure 2.5) and the other without it (see Figure 2.6).

11 2. Framing the Work

Figure 2.5: Example of occurence of ﬁrst ﬂush using when using dynamic EMC. Taken from (Kim et al. 2007).

Figure 2.6: Example of absence of ﬁrst ﬂush using when using dynamic EMC. Taken from (Kim et al. 2007).

Site Mean Concentration. Of greater value is perhaps the site mean concentration (SMC) that is a volume-weighed average of EMCs for one speciﬁc site, and gives information of expected pollution load levels on that site, for different stormwater runoff volumes. On this matter Mourad et al. (2005) studied the minimum number of samples necessary to establish a reliable SMC, but weren’t able to propose a standard minimal number.

2.2 Urban Water Pollution

2.2.1 BMPs and current approaches

In the twentieth century there was a growing concern over water, and its quality and management. The concentration of people living in urban areas grew, and water consumption rose exponentially. This, allied to climate uncertainties and severe droughts, has become a main concern for world policies in the new millennium. Only in 2011, the drought in Horn of Africa, the worst in 60 years, left people virtually without food and water, as crops failed. The UN estimates 11.5 million people are still on the brink of starvation, 30’000 children have already died in southern Kenya alone and 500’000 refugees were forced out of their homes, and a frightening political instability has ensued in these countries. In Europe, the ﬁrst communitarian (EEC) water legislations were made in the ’70s, concerning monitoring and control of surface and underground waters for human consumption, bathing, and

12 2.2 Urban Water Pollution marine life — while in the United States there was the Clean Water Act 1972 (CWA) —. Later came more protective and preventive laws, and in the ’90s wastewater treatment was mandatory. Now, the EU Water Framework Directive (WFD — directive 2000/60/EC) establishes the Union’s actions on the ﬁeld of water policy, and the more recent EU Directive 2006/7/EC introduces the concept of ‘active management of bathing water sanitary quality’ (Soyeux et al., 2007). There are some other relevant directives, such as:

• Directive 76/464/EEC, controlling the dumping of dangerous substances (like heavy metals) into water bodies;

• Directive 79/869/EEC, concerning the analytic methods and sampling frequencies;

• Directive 80/68/EEC regarding the discharge of certain dangerous substances into groundwater and establishing systematic monitoring of the quality of such water;

• Directive 80/778/EEC which concerns quality standards for water intended for human consumption;

• Directive 91/271/EEC about the collection, treatment and discharge of domestic wastewater, wastewater from certain industries and mixture of wastewater.

• Directive 91/676/EEC concerning the protection of receiving waters against pollutants of agricultural nature, like certain nitrates;

• Directive 98/83/EC that sets microbiological, chemical and organoleptic standards for drinking water quality.

CWA policies are typically enforced through maximum daily loads into water bodies — the Total Maximum Daily Loads (TMDLs) — for each analyzed parameter. A TMDL establishes the mamixum pollutant load that a water body can receive without compromising water quality standards and provides the basis for establishing water quality controls (Simpson et al., 2002). On the other hand, WFD requires characterization (or risk assessment) of discharges to evaluate the likelihood that receiving water will achieve the appropriate quality goals (Ellis and Revitt, 2008). As an example, the legislated maximum discharge loads (WFD) in mg/l for COD, E. coli and enterococci in water discharges are displayed in Table 2.1. Through programs implementing these directives, only in Portugal a total of 422 coastal bathing waters and 86 inland bathing water have signiﬁcantly improved their quality (Gondim, 2008). In Fig- ure 2.7 the Commision of the European Communities (2007) evaluates the performance of the EU countries in implementing the Water Framework Directive.

Table 2.1: Maximum discharge loads for COD, E. coli and enterococci in water discharges.

COD E. coli Enterococci 150 mg/l 500 MPN/100ml 200 MPN/100ml

13 2. Framing the Work

Figure 2.7: Performance indicator per Member State regarding the ﬁrst stage of the implementation of the Water Framework Directive, including the EU-27 average. Adapted from Commission of the European Communities (2007).

With the continuous growth of urban tissue, and impervious areas, urban catchments usually generate a great volume of stormwater very quickly, when a rainfall event occurs. WWTPs are usually at a loss in managing these great volumes. On the other hand, Liebman et al. (2011) argue that “the problem is not that urban areas produce excessive quantities of stormwater. On the contrary stormwater is a resource. The problem redefined is that urban areas have a deficit of beneficial uses for the runoff they shed”. This puts the problem quite nicely, but perhaps too conveniently. But then again, why not? After all, the current solid waste treatment policies are similar when they propose recycling. The problems that arise when reusing storm or wastewater should be very similar to those encountered by recycling solid waste. Heavy metals and toxic substances and microorganisms should be checked before reuse and most solid particles retained as well. And of course, not every waste is a resource by itself, and perhaps in some cases only a small percentage is actually usable, but it’s a small percentage of water resources that will be saved. Already, in many arid and semi-arid regions, harvesting stormwater from rooftops is a common practice (Jensen et al., 2011). Liebman et al. (2011) also remind that bringing potable water for use in activities such as watering gardens, street cleaning and even farming is sometimes a more costly operation — as it usually needs to be brought from outside the urban area — than reusing harvested and treated stormwater — which is close at hand. To confirm this statement McArdle et al. (2011) concluded, in a study in Australia, that potable water would actually cost less for the consumer if stormwater was harvested and treated, though they predicted some antagonistic mood from the public towards this solution. Best management practices (BMPs) must differ significantly from traditional water management, in order to achieve a sustainable water management. Traditional management is understood as a more linear and meet-demand-and-dispose-of-produce process. Recent BMPs regard water management as a closed loop — the urban cycle — introducing initiatives like stormwater harvesting and reuse (Chanan et al., 2010). There is an enormous potential in using rainwater in agriculture, or (in a more

14 2.2 Urban Water Pollution

Figure 2.8: Urban farming in Cuba: the use of stormwater could signiﬁcantly lower irrigation costs in urban farming. Adapted from Liebman et al. (2011). urbanized context) garden irrigation and street washing. Also, there have been several successful case-studies all over the world, proposing different ways of harvesting and reusing rain, storm and wastewater (Jensen et al., 2011; Liebman et al., 2011; McArdle et al., 2011).

Figure 2.9: Inﬁltration trench. Image by Center for TMDL and Watershed Studies, Virginia Tech.

Popular stormwater mitigation methods are infiltration trenches (see Figure 2.9) and wetlands. Infiltration trenches seek to absorb part of the runoff in a catchment and send it underground, reducing both first flush peak flows and storage volumes. An additional objective of this measure is to filtrate the maximum pollutants possible, through sand filters and different layers of soil. Infiltration systems are widely used, and present several advantages (Moura et al., 2011):

• reduction of ﬂows and volumes in downstream sewers or surface waters;

• limitation of wash-off phenomena in urban areas that lead to reduction in pollutant loads;

• contribution to groundwater recharge;

• strong potential to include stormwater treatment solutions;

• facilitation of urban development in areas far from surface outlets;

15 2. Framing the Work

• allow for different uses and landscape integration.

In spite of posing as an interesting mitigation solution, there are still some challenges to be sur- passed, as recently studied, like Nitrogen removal (Hatt et al., 2009), excessive accumulation of pollutants in filters (Hatt et al., 2011), groundwater pollution risks (Moura et al., 2011) or efficiency decrease possibly due to clogging (Bergman et al., 2011; Freni et al., 2009). Moura et al. (2011) state that, because of these challenges, developing mechanisms to evaluate infiltration trenches is indis- pensable and urgent and they even propose a multicriteria decision-aid method to do this, while Freni et al. (2009) further state that infiltration trenches are more effective in quality rather than quantity control, and they still maintain a good performance considering the clogging and the absence of any pre-treatment structure. Wetlands seek to store large volumes of stormwater, and because they are much larger in area than trenches, they don’t require such high infiltration rates. The down side is that such large areas are not usually available in urban contexts, but urban planning now proposes that gardens and parks be designed to function as wetlands and sedimentation or detention ponds (see Figure 2.10) during wet weather and storm events (e.g. Jensen et al., 2011). Although some authors fear the entry of pollutants into ponds might affect its aquatic biosystems, Marsalek et al. (2002) studied the effect of sediments and pollutant retention in an on-stream pond, in Canada, and concluded that it was successfully mitigating impacts of urban runoff, with benthic taxa and organism counts being about the same upstream and downstream of the pond, which was later corroborated by a similar study conducted by Tixier et al. (2011).

Figure 2.10: Implementation of a small detention pond. Taken from Abbey Associates Inc. (www.abbey- associates.com/splash-splash).

Following a critic that wetland’s capacity to treat microbial loads has not been thoroughly evaluated (Mendez´ et al., 2009), several studies have ensued in this matter (Mendez´ et al., 2009; Sidhu et al., 2010). Hathaway et al. (2011) proposes that wetland design should promote hydraulic retention times, and minimization of stormwater velocity to promote sedimentation and degradation of microbes, such as E. coli. A not so popular measure is to use Rainwater tanks in stormwater harvesting. Although this solution has not been extensively studied in terms of stormwater impact reduction, it is a very attractive

16 2.2 Urban Water Pollution solution in water management in countries with difficult access to potable water. Khastagir and Jaya- suriya (2010) state that a 3000L tank is capable of reducing hydraulic loading by 75%, TSS by 97%, Total Phosphorus by 90%, and Total Nitrogen by 81%, thus making stored water fit to meet regular house demands such as toilet flushing, laundry use and garden watering. An example of a rainwater tank proposed by a construction company in Australia — House Building Australia — is shown in Figure 2.11.

Figure 2.11: Rainwater tank solution proposed by House Building Australia. Taken from House Building Australia (housebuildingaustralia.wordpress.com).

Even when not associated with water shortage and reuse, retaining roof runoff is an attractive solution in reducing both runoff volume and pollutant loads. Some architectural solutions propose garden rooftops that provide this runoff retaining function, while improving urban landscape. An example of a garden rooftop is the “Zero Energy Building” designed by Zoka Zola Architecture, shown in Figure 2.12.

Figure 2.12: Garden rooftop solution in a house by Zoka Zola Architecture. “Zero Energy Building” by Zoka Zola Architecture c 2003.

17 2. Framing the Work

2.2.2 (in)Sanitary Engineering?

Legislation on stormwater quality is still at a premature stage. There is virtually no regulation concerning the treatment and disposal of stormwater, only regarding water discharges in general. Most of stormwater solutions try to prevent only the more direct urban consequences of runoff, such as flooding or soil erosion, sometimes with attention to pollutant retention (e.g. detention basins, sedimentation ponds, infiltration trenches or stormwater tanks), but as the washing, transport and accumulation of pollutants and microbial pathogens is not well understood, these measures lack specific and definite design criteria. There is also no monitoring of stormwater quality, with the exception of that performed in WWTPs for the incoming mixture of storm and wastewater. In the context of urban stormwater and waste water management the assessment of the microbiological quality of discharges is extremely superficial. Urban water monitoring of microbiological features is generally reduced to faecal indicator bacteria (FIB). As explained before, the typical parameters used in water management are BOD, COD, and indicators such as Faecal Coliform bacteria (E. coli and enteroccoci). This is probably because most European countries have no greater concerns with faecal contamination because they lack coastal waters; instead, their concern is mainly focused on quality parameters monitoring agricultural pollutants like Nitrogen/Phosphorus ratios. Sanitary Engineering strives to guarantee minimum acceptable (i.e. legal) quality requirements for discharges, through use of different treatments in WWTPs. On the other hand, this type of microbiological assessment of water is absolutely poor for Health Sciences, because of its clear uselessness in epidemic or disease monitoring activities, for example. Also, there is absolutely no monitoring of other relevant waterborne pathogens such as virus, which could greatly improve both water quality and social response to viral outbreaks. Tracing pathogens back to their sources is very important for two main reasons: the first being the increase of mitigation potential of contamination effects, and the better understanding of the transport process; and the second being its use as a legal instrument in solving disputes when illegal or contaminated discharges are being made in receiving waters. The containment and source tracking of pathogens is a primary target for the Health sector, while current wastewater management policies only provide the quantification of the likely number of total pathogens, which is a meager assessment at best. However, of late, urban water management practices are visibly migrating from a treatment-just-before-discharge policy to a treat-at-source policy, both to meet more exigent legal requirements and to reduce total cost of WWTPs.

18 3 State of the Art

Contents 3.1 Urban Stormwater Runoff ...... 20 3.2 Dealing with rainwater in Portugal ...... 30

19 3. State of the Art

3.1 Urban Stormwater Runoff

3.1.1 Characterization

In this new millennium, several studies were conducted in order to assess and characterize in detail the quality of stormwater runoff. Table 3.1 shows the typical mean concentration values of TSS, BOD and COD in urban contextes, for some of these studies from across the world.

Table 3.1: Previous studies showing typical values for EMC or SMC (mg/l) for stormwater quality. * Peak load values and not EMC. **Typical values and not EMC.

Country Region TSS BOD COD Reference

Australia Melbourne 82 - - Bach et al. (2010)

Australia Sydney 110∼650 0.6∼16.9 - Al Bakri et al. (2008)

Canada Saskatoon 160∼210 - 55∼100 McLeod et al. (2006)

China Macau 319 - 201 Huang et al. (2007)

China Shenzhen 258∼1986 29∼280 59∼500 Luo et al. (2009)

France Paris 30∼75 8∼25 43∼113 Gromaire et al. (2001)

Iran Isfahan 149 - 649 Taebi and Droste (2004)

Italy Genoa 140 - 129 Gnecco et al. (2005)

Japan - 135∼242 36∼91 32∼85 Uchimura et al. (1997)*

Korea Chongju 275 - 509 Lee et al. (2003); cit. in Chow et al. (2011)

Malaysia Johor 364 95 311 Nazahiyah et al. (2007)

Malaysia Skudai 261 74 192 Chow et al. (2011)

Portugal Lisbon 390 32 203 Ferreira (2006)**

Portugal Lisbon 153 46 150 Gondim (2008)**

USA Texas 56 23 116 Baird et al. (1996); cit. in Chow et al. (2011)

In Australia, Al Bakri et al. (2008) tried to characterize the stormwater runoff in the city of Orange, a semi-urban catchment of 5800 ha about 250km west of Sydney. Stormwater quality was monitored in six sampling sites with ISCO automated samplers, for about two years. For TSS, registed EMC values reached over 650 mg/l in two sampling sites in dry weather and again in wet weather, though just for one site, and maximum EMCs were lower than those of dry weather period, ranging from 110 to 655 mg/l. In Canada, McLeod et al. (2006) conducted a study in four large urban catchments in the city of Saskatoon, trying to make each site different in land use, namely: new residential (241 ha), old residential (616 ha), commercial (74.6 ha) and light industrial (420 ha). Automated samplers were used in storm sewers, with readings recorded at 5 min intervals. SMC values obtained for TSS were: 210 mg/l in the commercial area, 190 mg/l in the old residential area and 160 mg/l in the new residential area; and for COD were: 75 mg/l in the commercial area, 100 mg/l in the old residential area and 55 mg/l in the new residential area. No results were used in the light industrial catchment

20 3.1 Urban Stormwater Runoff due to insufﬁcient data. In China, Huang et al. (2007) evaluated stormwater runoff from a catchment in the center of Macau. The catchment covered an area of 14 ha and impermeability of 60%, densely populated and mainly residential and commercial. Huang et al. (2007) registered SMC values of 318.6 mg/l for TSS, 201.4 mg/l for COD and for heavy metals: 55 µg/l for Zn, 3.2 µg/l for Pb and 4.9 µg/l for Cu. The authors veriﬁed as well that EMC values varied greatly between rainfall events. Also in China, Luo et al. (2009) evaluated runoff in an urban catchment in Shenzhen, in the central coastal area of southern Guangdong. The catchment has an area of about 119.2 ha and an occupation of 32% residential, 36% green land, 16% commerce, 6% industry, 9% communication areas and less than 1% of undeveloped land. The impervious coverage of the catchment is about 81%. The authors published EMC values that range between: 258∼1986 mg/l for TSS, 29∼280 mg/l for BOD, 59∼500 mg/l for COD, 2.05∼8.38 mg/l for TN and 0.03∼12.91 mg/l for TP. In 1999, Gromaire et al. (1999) concluded that roof runoff in Paris contributed more to urban runoff pollution than road runoff. Later, Gromaire et al. (2001) studied in the Marais catchment the quality of stormwaters in Paris, France. The catchment, located in the historical centre of Paris, has an area of about 42 ha, 90% of which are impervious. The area can be divided into roof surfaces (54.4%), streets (22.4%), courtyards (mainly impervious) and public squares and gardens (23.2%). Runoff was sampled on 20 different points, including roofs, streets and courtyards. EMC values obtained by the authors are shown as an interval between the 10th and the 90th percentiles: 30∼75 mg/l for TSS, 8∼25 mg/l for BOD, 43∼113 mg/l for COD. Taebi and Droste (2004) assessed characterized stormwater runoff in Iran, in an urban catchment in the city of Isfahan. The studied catchment was homogenous in occupation, with an area of about 560 ha, an impervious coverage of 35% and an average slope of 2%. SMC values for studied parameters and their maximum EMCs are shown in Table 3.2.

Table 3.2: EMC values obtained in Isfahan, Iran. Taken from Taebi et al. (2004).

TSS COD TN TP Pb Zn (mg/l) (mg/l) (mg/l) (mg/l) µg/l µg/l SMC 149 649 6.75 0.274 314 453 max EMC 467 2542 22.38 0.79 558 2386

Gnecco et al. (2005) analysed stormwater collected from an experimental catchment in the Uni- versity of Genoa in Genoa, Italy, discriminating between road and roof runoff. The site was a small residential basin with 2800 m2, with an impervious coverage of about 75%. The authors registered a substancial contribution of Zn from rooftops, while other parameters were smaller than those collected in stormwater. The most relevant results are shown in Table 3.3: Uchimura et al. (1997) conducted a survey in Japan over seven cities in order to assess and characterize urban stormwater runoff quality. The authors published results for one of the surveyed cities — T-city — with an urban catchment of 5 ha, resorting to automated samplers. Peak pollutant loads range between 135∼242 mg/l for TSS, 36∼91 mg/l for BOD, 32∼85 mg/l for COD, 3.2∼30.8

21 3. State of the Art

Table 3.3: Road and roof EMC values for Genoa. Taken from Gnecco et al. (2005).

TSS COD Cu Pb Zn Zn (roof) (mg/l) (mg/l) (µg/l) (µg/l) (µg/l) (µg/l) average EMC 140 129 19.4 13.2 81.1 446.7 max EMC 377 281 53.3 23.3 123.4 758.8 mg/l for TN and 0.26∼0.83 mg/l for TP. In Korea, Choe et al. (2002) reports EMC values for stormwater characterization in the city of Chongju, in several urban basins. The area has an yearly precipitation of 1225 mm, and is divided into three residential areas: with respective predominance of multi-family houses, single family houses and commercial buildings; and three industrial areas: each with a different industry type: metal, food and textile. The values of EMCs obtained for the different studied pollutants were as described in Table 3.4.

Table 3.4: EMCs of stormwater pollutants obtained in Korea. Adapted from Choe et al. (2002).

Sites TSS BOD COD TP Cr Cu Pb Fe (mg/l) (mg/l) (mg/l) (mg/l) µg/l µg/l µg/l µg/l Residential Multi-family 145.8 76.2 211.2 1.21 51 77 426 3910 Single-family 414.1 125.3 226.0 2.85 44 99 189 5930 Commercial 276.1 168.8 501.4 1.88 28 60 102 6020 Industrial Metal 88.3 58.8 118.4 2.60 67 44 157 3473 Food 90.7 34.2 71.7 1.30 80 45 85 3903 Textile 139.8 36.1 50.0 1.90 54 20 79 3900

A study conducted in Malaysia, by Nazahiyah et al. (2007), in an urban basin located in Skudai, Johor, also registered the ﬁrst ﬂush phenomenon and a great variability of EMCs between rainfall events. The study site was a low cost residential catchment of 3.34 ha, and a pervious coverage of 15% and an average slope of 2.53%. The authors comment on the severe degree of pollution of stormwater according to the Interim National Water Quality Standards for Malaysia, and registered SMC values as described on table 3.5:

Table 3.5: SMC and Maximum EMC values for stormwater in Skudai, Johor, Malaysia in 10 storm events. Taken from Nazahiah et al. (2007).

TSS BOD COD NO3-N NO2-N NH3-N TP Pb (mg/l) (mg/l) (mg/l) (mg/l) (mg/l) (mg/l) (mg/l) (µg/l) SMC 364 95 311 2.40 0.10 3.50 3.00 20 max EMC 1024 190 728 6.00 0.82 9.12 7.80 70

Also in Johor, Malaysia, Chow et al. (2011) studied the first flush in a commercial catchment. The catchment had an area of 34.21 ha, with about 98% of impervious surface. Stormwater was grab- sampled during storm events. The authors confirmed the presence of first flush phenomenon, as it was shown by the results in Figure 2.4, and obtained EMC values with high inter event variation, showing to be strongly influenced by storm size, which are shown in Table 3.1. In the north of Portugal, Costa and Ram´ısio (2012) made a relevant study of stormwater quality

22 3.1 Urban Stormwater Runoff in an urban catchment in the city of Guimaraes,˜ even though only the loads of certain nutrients and heavy metals were evaluated. The catchment is located in the historical centre of the city, with an area of 0.82 ha, with an impervious coverage of about 80%, 60% of which are rooftops and the remainder paved surfaces. The runoff coefficient of the catchment was estimated in 0.7. Samples were collected at the stormwater collector’s outlet that discharges into the Couros creek, with ISCO automated samplers. Obtained SMC values for heavy metals were: 240 µg/l for Fe with a maximum EMC of 330 µg/l; 40 µg/l for Zn with a maximum EMC of 80 µg/l; and 110 µg/l for Cu with a maximum EMC of 410 µg/l. Because pollutant concentration loads in urban stormwater is strongly related to the characteristics of the catchment, Zgheib et al. (2011) studied the influence of land use pattern on stormwater pollution. Water samples were collected at the outlets of three storm sewer networks in Paris and its outskirts with automated samplers. One site was a residential catchment on the suburbs of Paris and the other two were densely urbanized, one in the centre of Paris and the other also in the suburbs. The authors did not find significant differences in pollutant concentrations, but the study was one of the first describing priority substances in Parisian stormwater, like phenols and organotins, as well as heavy metals. Some of the more recent studies include microbiological assessment to urban runoff characterization. Table 3.6 shows some of these evaluations for Total Coliforms (TC), Faecal Coliforms (FC), E. coli (EC) and enterococci (EF). Special emphasis should be given to the Portuguese studies, since these provide good background data for comparison with the present study. Ferreira (2006) studied stormwater runoff collected from the Alcantara basin, an experimental urban catchment in Lisbon, Portugal. The catchment had an area of about 88 ha, and was divided into different land occupations, such as residential single-family, or multi-family, commercial areas with intense traffic and no vegetation or gardens. The water was grab sampled, and each collection point tried to describe or at least be associated with a certain land use. Although no EMC could be calculated, because only one sample was collected in each point per event, typical concentrations from the beginning of the event could be assessed, for BOD (23 mg/l), COD (203 mg/l) and TSS (390 mg/l). Also, Total Coliform counts were quantified, as seen in Table 3.6. Two years later, Gondim (2008) followed the same study in the same catchment. Typical values for TSS, BOD and COD confirmed the previous ones (see Table 3.1), but more microbiological parameters were quantified, namely Faecal Coliform, E. coli and enterococci, which are shown in Table 3.6:

Table 3.6: Mean concentrations for microbiological parameters in Portugal.

Country Region TC FC EC EF Reference (MPN/100ml) (MPN/100ml) (MPN/100ml) (MPN/100ml)

Portugal Lisbon 6.7E+06 - - - Ferreira (2006) Portugal Lisbon 2.7E+07 3.2E+05 3.2E+05 2.2E+05 Gondim (2008)

Although the values in the presented studies are of the same order of magnitude, it is difﬁcult to establish standards for stormwater pollution, because these differences are greatly inﬂuenced by

23 3. State of the Art climate conditions, types of land use and geomorphology. Also, there is no standard methodology or approach that is universally applied to such studies, and comparison is difﬁcult because of it. On the other hand, it shows the step-by-step stage of the development of pollution assessment approaches, which try to account for as much variables as possible.

3.1.2 Urban Response

The effect of the urban environment upon the quality of stormwater depends mainly on a set of factors with three different natures: geomorphologic characteristics of the catchment (Butler and Memon, 1999, cit. in Ferreira, 2006); climate conditions and precipitation regime over the basin (Gnecco et al., 2005); and the land use of the basin (Gray, 2004). To minimize WWTP overload events during rainfall events, there has been exhaustive research in methods devised to either control, check, deviate or minimize the pollutant charge and volume of stormwater. Most of recent solutions tend to adopt a treat-at-source perspective (see 2.2.2). There is an apparently effective method — that falls perhaps better in the preventive category — that is street sweeping. Street sweeping reduces the pollutants and solid particles or urban surfaces during dry weather, which would otherwise be washed off by stormwater during a rainfall event. Kang and Stenstrom (2008) are of the opinion that the most important factor in preventing pollutants discharge during storm events is the sweeper’s ability to pick up fine particles. The authors also state that little evidence is shown — despite numerous studies — that street sweeping directly improves stormwater quality, and despite the intuitive notion that larger amounts of swept material means less pollutants to be washed off. But they also state that “the contribution of street sweeping to environmental quality in urban areas should not be underestimated because of previous studies, which had insufficient statistical power to detect water quality improvements, had they existed”. The authors also provide a rather clear perspective into the expected — and intuitive — effects of street sweeping, which can be seen in Figure 3.1. Gromaire et al. (2000) studied two types of street sweeping in Le Marais catchment in Paris, both used at the time: manual gutter washing and pressurized water jet. Although their study showed that street cleaning waters appeared to be a minor pollutant source during dry weather, a significant amount of sediments was eroded inside the combined sewer, leading to believe that street sweeping can have an important role in reducing in-sewer pollution and thus reducing the first flush effect — during storm events — of these pollutants as well. Chang et al. (2005) proposed another method that combines sweeping and washing. The sweeper makes a pass, wetting the pavement, sweeping and sucking particles with a modified regenerative-air vacuum sweeper. The washer passes after the sweeper, washing away re-suspended particles. The authors have concluded that this combined method offers a measurable reduction in suspended solids. Another complementary approach to solve the stormwater quality problem is to address WWTPs as systems similar to supply chains. Like any supply chain system, there is an uncertainty associated with the supply of certain entities involved in the production process — like flows and volumes of raw storm/waste water, concentration of pollutants and microorganisms, available sludge and the previous

24 3.1 Urban Stormwater Runoff

Figure 3.1: Expected effects of street sweepin in pollutant loads versus a non-swept basin scenario. Taken from Kang and Stenstrom (2008).

flows that may still be retained in the treatment chain — which strongly influences the performance of treatment and quality of the final product — treated water discharges. As WWTPs are usually at a loss when strong storm events occur (see 2.2.1), and subsequent disruptions and disturbances are inevitable in this type of systems, solutions focusing on increasing their resilience might be also interesting. The purpose of resilience is to react effectively to the negative effects that disrupt supply chain systems. Several resilience strategies are based on redundancy, responsiveness, flexibility, robustness, velocity and collaboration between involved entities. Therefore, solutions should not only focus on possible disturbances and their sources, but on identifying failure modes and how to protect the system against them based on a predictive and preventive approach. (Carvalho et al., 2012) As an example of this philosophy, Ahnert et al. (2009) developed a model-based comparison between to solutions that intended to enhance WWTP capacity under stormwater conditions: the first is to increase inflow based on a dynamic capacity of the WWTP at the time — i.e. the actual capacity of the Plant in the moment the inflow arrives —, rather than simply discharging directly when the pre-set maximum inflow is reached (usually this design maximum is set for the worst-case scenario, which does not always occur); the second is to make the extra inflow bypass the activated sludge tank, and discharge it directly into the secondary clarifier. The second solution has three different options: the first (option A) enters the WWTP and is still collected in the storage tank, benefiting from sedimentation; the second (option B) goes further into the WWTP adding the passage through the sand-traps; the third option (option C) has the disadvantage to let inflow directly into the secondary clarifier — quite unlike the previous two, which would always meet some kind of primary treatment —,

25 3. State of the Art though some of it is recirculated with the return sludge stream “back” into the activated sludge unit. The author’s schematics of both solutions is presented in Figure 3.2. Ahnert et al. (2009) report that for the simulated time series, the model presents COD load re- ductions of 20% in discharges for both approaches, while total ammonia load reduction was of about 6% for the bypass solution and 11% for the inflow increase solution. While the bypass solution allows for higher inflows, the pollutant removal efficiency is not as good as the increased inflow solution. The authors also propose a combined approach as a way to minimise adverse effects and overall emissions to the receiving waters.

Figure 3.2: WWTP layout with three possible bypass solutions. Taken from Ahnert et al. (2009).

3.1.3 Modelling Runoff

Most stormwater quality models are either EMC models, SMC models or pollutograph models. Pollutograph models display the evolution of pollutant loads and runoff and rainfall volumes over time, during a storm event. They display more detailed information but require significantly more data collection to be properly applied (Dembel´ e´ et al., 2010). Most of these models predict pollutant loads for a given runoff volume, whose relation to rainfall quantities is made by model parameters obtained through regression analysis. One of the greatest challenges of stormwater modelling is to obtain the correct explanatory variables, since the build-up and wash-off processes of diffuse source pollutants is not well known. This complex nature of pollutant accumulation and wash-off, along with high temporal and spatial variations, generates technical difficulties in the development of accurate and reliable models (Dotto et al., 2010). Also, the calibration of models is rather difficult since there are no great quantities of runoff data and each data set is site-specific and event-specific. Calibration and sensitivity analysis is usually carried by random data generation algorithms, or frequencies analysis.

Types of models The most simple of models is the one obtained by simple regression and was presented already presented (see 2.1.3). There are other, more advanced, models that try to introduce the complexity of stormwater modelling while maintaining the inherent simplicity of models. Among the most common of these, are models based on multiple regression analysis, like the following one, used by Ferreira (2006). The basis of a multiple regression is that a set of explanatory variables, under direct

26 3.1 Urban Stormwater Runoff manipulation of a set of parameters, explain the dependent variable, like the equation that follows (Equation 3.1):

Y = a + b1X1 + b2X2 + b2X2 + ··· + bnXn (3.1)

where Y is the estimated variable, {X1,...,Xn} is the set of explanatory variables, a is a model parameter and {b1, . . . , bn} is also a set of model parameters but associated with the explanatory variables in such a way that bi matches Xi. Ferreira (2006) chose as a variable set the ones presented in the expression 3.2.

X1 = TSX2 = Imax X3 = Imean X4 = ln(TS) X5 = ln(Imax) X6 = exp(Imax) (3.2)

where TS is antecedent dry weather period; and Imax and Imean are both maximum and mean rainfall intensity. For the COD modelling, approaches using {X1,X2} and {X1,X2,X3} showed good adjustment. For TSS modelling, the approach {X1,X2} showed the best adjustment, and for BOD it was approach {X4} that showed the best adjustment. Another example of a multiple regression analysis model, is the one inspired on the Rational For- mula generally used in Hydrology, proposed by Che et al. (2003; cit. in Liu et al., 2005) and presented by equation 3.3:

Z τ0 Z t Z T Z t 1 1 0 Y = C(t) · h(t)dt dt + C(t) · h(t)dt dt (3.3) τ0 0 0 τ0 τ0 t−τ0

where Y is the runoff pollutant load per area unit; τ0 is the concentration time of the catchment; T is 0 the duration of the rainfall event; C(t) is the concentration of the pollutant in at instant t; and h(t) is the rainfall intensity at instant t. The first parcel of the equation describes the concentration increase before the peak flow and the second parcel the concentration decrease after the peak flow. For C(t), Che et al. (2003; cit. in Liu et al., 2005) propose an exponential decay law from an initial concentration

C0. This model presents three problems: the first is the aggravated weight of the second parcel when rainfall duration gets bigger than the catchment concentration time (T >> τ0); the second is that the change of model behaviour occurs for the peak flow, which may not coincide with the pollutant concentration peak derived from the first flush phenomenon; the third is the concentration seems to always decrease (even if at different rates), and in-flow pollutant build-up is not taken into account. Dembel´ e´ et al. (2011) propose a log-linear EMC model that — although it is a “back to simple regression” step — tries to introduce some empirical aspects into EMC models and reduce site depen- dency, mainly by breaking the model into two different behaviours and ensuring only one explanatory variable. The model is described by equation 3.4:

27 3. State of the Art

( b ln(X) + b , if X ≤ λ EMC = 1 2 (3.4) b3 X + b4, if X > λ

where X is the product between the rainfall depth and the antecedent dry weather period and b1, b2, b3 and b4 are model parameters. The model assumes a logarithmic increase of the pollutant EMC while X is below a threshold λ, until it reaches the peak of the first flush phenomenon, and from there the model assumes a decrease of EMC proportional to 1/X. The joining of both rainfall depth and antecedent dry weather period into one explanatory variable seems a risky step, for it requires the established relation to be an explanatory variable, and the choice of the relation itself poses some difficulties. For example, for an explanatory variable chosen by Dembel´ e´ et al. (2010), which defines X as the ratio between antecedent dry weather period and total depth, any increase in antecedent dry weather proportional to rainwater depth produces the same outcome in the model. As these examples do not presume to be exhaustive it must be said that there are still a wide number of models in the last few years that have tried to grasp, with different levels of success, the complexity of urban stormwater pollution modelling, and that will not be described here (e.g. Bach et al., 2010; Gromaire et al., 2011; McCarthy et al., 2010; Moura et al., 2011; Vezzaro et al., 2010).

Selection of explanatory variables The selection of explanatory variables is probably the most delicate step in stormwater quality modelling. Explanatory variables must usually meet two requirements: each explanatory variable (raw or transformed) must be linearly correlated to the dependent variable; and at least theoretically, neither statistical nor physical correlation should exist between explanatory variables (Dembel´ e´ et al., 2010).

Calibration, validation and sensivity analysis One popular way to calibrate model parameters is the Ordinalry Least Squares (OLS) method. This method chooses the value for a parameter that minimizes the sum of the squares of errors

(SOLS) between the observed and estimated dependent variable (Dembel´ e´ et al., 2010). This method is translated by the equation 3.5:

N X 2 SOLS = (Yk − Yˆk) (3.5) k=1 where Y and Yˆ are respectively the N observed and estimated values of the dependent variable. As all observations have the same weight, the OLS method is strongly inﬂuenced by outliers. And because all urban hydrological data sets show the presence of outliers, skewed distributions and large uncertainties in some measured variables, the OLS method may not provide a best linear unbiased estimate of parameters.

28 3.1 Urban Stormwater Runoff

An alternative is a robust regression method like the Iteratively Reweighted Least Squares (IRLS), which assigns to each observation a weight that tries to reduce the inﬂuence of outliers. The IRLS is described by the equation 3.6:

N X 2 SIRLS = Wk(Yk − Yˆk) (3.6) k=1

where Wk is the weight associated with each observation. Another two methods used to assess hydrological models, among many others, are the Root Mean Square Error (RMSE) and the Nash-Sutcliffe criterion (NC), described by equations 3.7 and 3.8, respectively:

v u N u 1 X RMSE = t (Y − Yˆ )2 (3.7) N k k k=1

PN (Y − Yˆ )2 NC = 1 − k=1 k k (3.8) PN 2 k=1 (Yk − µY )

where µY is the mean value of the N observations. Like OLS, RMSE is an unbiased estimate, provided the errors are normally distributed with no signiﬁcant outliers. The Nahs-Sutcliffe criterion is extensively used in hydrological modelling and corresponds to the variance of the model errors divided by the variance of the observations. It ranges from 1 (perfect model) to −∞ (completely unsuited model). In practice, NC values above 0.7 are considered as belonging to a satisfactory model. (Dembel´ e´ et al., 2010) As most stormwater quality models lack for adequate data amounts, there are widely accepted processes of generating plausible data for model calibration and/or veriﬁcation. One of these approaches are Bayesian statistics methods, in which posterior distributions are determined, given a likelihood data model and a prior distribution model. Another of these methods is the use of Monte Carlo Markov Chains (MCMC). Monte Carlo methods are a class of computational algorithms that rely on repeated random sampling to compute results. Examples of simple Monte Carlo processes are Random Walks algorithms. A Markov Chain is a stochastic process where the set of random variables model a certain process over time. In each iteration step, the generated value is strongly dependent on the previous one. When associated with random processes like Monte Carlo, these randomly generated values con- struct a Markov chain that converges to a desired distribution — the equilibrium distribution. Kanso et al. (2003) state that unlike traditional statistic theory, this method can readily cope with the nonlin- earity of a model.

29 3. State of the Art

3.2 Dealing with rainwater in Portugal

In Portugal most urban areas have been established before the controlling of stormwater became a concern in urban planning. Most stormwater in conveyed to WWTPs through the sewer systems. However, in developing urban areas — or through interventions in older ones — some solutions were used. The most popular solutions are the porous pavement, usually in parking lots, and green areas, serving either as infiltration trenches or wetlands, depending on their available area. In Figure 3.3 two examples of the use of porous pavement that can be seen in Lisbon. Golf resourts are planned with detention basins and sedimentation ponds; infiltration pits are usually built around trees in sidewalks. And garden design contemplates the use of infiltration basins and sometimes sedimentation ponds (see Figure 3.4).

Figure 3.3: Use of porous pavement in parking lots in Alcantara (Left), and Alges´ (Right), in Lisbon. Taken from Ferreira (1999).

Figure 3.4: Left: The use of gravel to facilitate inﬁltration in the Gulbenkian Foundation garden in Lisbon. Right: Detention Pond in the garden of the Faculty of Psychology, University of Lisbon. Image in the Public Domain.

Ferreira et al. (2004) proposed some solutions to address the stormwater quality problem in Al- mada (see Figure 3.5). One of the solutions was to build an inﬁltration basin in the catchment Vale Cavala, with an area of 280 ha, because stormwater could not be drained through gravity in this catchment. Water is collected throughout the basin in a stormwater drainage system with pipe diameters of 2000 mm, and ends in an inﬁltration basin, with an energy dissipation device.

30 3.2 Dealing with rainwater in Portugal

Figure 3.5: Stormwater solutions in Almada. (1) Campo da Bola; (2) Vale Cavala; (3) WWTP of Portinho da Costa; (4) Vale da Regateira. Adapted from Ferreira et al. (2004).

A second solution was the use of a stormwater reservoir. The catchment of Campo da Bola has elevations close to sea levels, and due to its closeness to the shore, and reduced slopes, the stormwater drainage system was constantly flooding in wet weather periods. The underground reservoir acts as a damping and storage device, where the stormwater is pumped from into the final emissary. Another solution was a retention-infiltration basin in the Regateira catchment (246 ha). In order not to overload the old stormwater systems downstream of this catchment, a dry retention-infiltration basin was built, over an area of 1.2 ha and a storage capacity of 6600 m3. Lança et al. (2012) studied the feasibility of collecting stormwater in public buildings for garden watering. Roof runoff was collected a public building in Coimbra into a tank then used to water a garden through water sprinklers. With regular monitoring of its quality, the author found that the quality of stormwater sprayed in the garden was good, and every parameter was below the applicable legislated limit. The WWTP in Alcantara — the most advanced in Portugal — serves the better part of Lisbon, with a population of about 800.000 habitants, over an area of 37 km2 and the collector system sums up to 22.4 km. It was presented with gardened roofing that promotes infiltration and retention. The retention is greatly increased with this new roofing solution, whereas the old one was completely impervious, and adds greatly to reduction of stormwater runoff through infiltration. This solution can be seen in Figure 3.6.

31 3. State of the Art

Figure 3.6: Alcantara WWTP in Lisbon.

32 4 Microbiology

Contents 4.1 Escheri. . . what? ...... 34 4.2 Microbial Source Tracking Methods ...... 37 4.3 Case-Study Techniques ...... 41

33 4. Microbiology

4.1 Escheri. . . what?

4.1.1 A Microscopic Overview of Pathogenic Organisms

Wastewater is an important vessel for pathogenic organisms to spread over a population. Animal and human faeces often contain infectious microorganisms that will contaminate surface and wastewater, and then the receiving waters, posing a serious health threat unless these pathogens are conveniently dealt with. This mitigation happens in wastewater treatment facilities, that must continuously monitor the organic pollutants of wastewater in order to apply an effective treatment. According to Metcalf & Eddy (2003) the principal pathogenic organisms belong to four broad categories: bacteria, protozoa, helminths and viruses.The most common pathogenic organisms found in raw wastewater are presented in Table 4.1, along with their assossiated diseases and symptoms.

Table 4.1: Pathogens commonly found in untreated wastewaters. Adapted from Metcalf & Eddy (2003).

Category Organism Disease Escherichia coli (E. coli) Gastroenteritis (diarrhoea) Campylobacter jejuni Gastroenteritis (diarrhoea) Legionella pneumophila Legionnaires’ disease Leptospira Leptospirosis Bacteria Salmonella Salmonellosis (food poisoning) Salmonella typhi Typhoid fever Shigella Shigellosis (bacilary dysentery) Vibrio cholerae Cholera (severe dehydration) Yersinia enterocolitica Yersinosis (diarrhoea) Balanditium coli Balantidiasis (diarrhoea, dysentry) Cryptosporidium parvum Cryptosporidiosis (diarrhoea) Protozoa Cyclospora cayetanensis Cyclosporasis (severe diarrhoea, vomiting) Entamoeba histolytica Amebiasis (diarrhoea with bleeding) Giardia lamblia Giardiasis (diarrhoea, nausea, indigestion) Ascaris lumbricoides Ascariasis (Roundworm) Enterobius vermicularis Enterobiasis (Pinworm) Helminths Taenia saginata Taeniasis (beef tapeworm) Taenia solium Taeniasis (pork tapeworm) Trichuris trichiura Trichuriasis (whipworm) Adenovirus Respiratory disease Enterovirus (polio, echo, coxsakie,...) Gastroenteritis, heart anomalies, meningitis Hepatitis A virus Infectious hepatitis Viruses Norovirus Gastroenteritis (vomiting) Parvovirus Gastroenteritis (vomiting) Rotavirus Gastroenteritis (vomiting)

A very important feature of microorganisms is their ability to form resistant forms, that can survive through very harsh conditions in extreme environments. As an example, some bacteria can create spores, an extremely resistant form that can endure heat and disinfecting chemicals, and remain dormant for decades. Some other microorganisms can form cysts (e.g. Giardia lamblia), oocysts (e.g. Cryptosporidium) or embryonated eggs (e.g. Ascaris lumbricoides). (Metcalf & Eddy, 2003)

34 4.1 Escheri. . . what?

4.1.2 Bacteria

The human digestive system is populated by many types of bacteria that are continuously shed in faeces. Although most of them are harmless, some of the excreted organisms can cause serious health problems. This implies a wide variety and concentration of bacteria in wastewater, both of pathogenic and non-pathogenic nature. Figure 4.1 represents the typical cell structure of a bacteria.

Figure 4.1: Bacterial Structure. By i-KOS Credentials.

One of the most common bacterial pathogenic found in wastewater throughout the world is the Escherichia coli (E. coli) an enteropathogen that can be present in high quantities in contaminated surface waters, and is transmitted to humans mostly by infected food and water. The E. coli usually circulates among individuals of a certain population without any symptoms — possibly due to im- munization granted by previous infection — affecting only external members visiting the community. This bacterium is on the genesis of the well renown warning to travelers not to drink the local water (Prescott et al., 1993). However, some E. coli strains can be quite dangerous, especially in the current global market, where food travels great distances. An example of this was the spring outbreak of E. coli — the E. coli O104:H4 strain — in Germany earlier this year, affecting about 3000 people, 800 of which suffered symptoms that included bloody diarrhea. The health authorities reported that the enteropathogen was spreading through infected cucumbers imported from Spain, and even though later it was found that the cucumbers had only been infected somewhere along the chain of transport, the whole incident still inﬂicted mass losses in the vegetable market in Europe — as tons of them had to be destroyed —, and resulted in generalized health panic and about 50 deaths. Another dangerous E. coli strain is the enterohemorrhagic E. coli O157:H7, because it frequently leads to hemorrhagic diarrhoea, and occasionally kidney failure, especially in young children and elderly people. Another very common pathogen found in wastewaters is the genus Salmonella. This group contains a wide variety of species that can cause disease in humans and animals. The most common of them is food poisoning, identiﬁed as salmonellosis, which is typically carried in eggs, while the worst is typhoid fever, caused by Salmonella typhi, which causes high fevers, diarrhoea and ulceration of the small intestine. Figure 4.2 represents a micrograph of a species of Salmonella. Another genus of bacteria, though less common, is the Shigella — with reported waterborne

35 4. Microbiology

Figure 4.2: Color-enhanced scanning electron micrograph showing Salmonella typhimurium (red) invading cul- tured human cells. Image by Rocky Mountain Laboratories, NIAID, U.S. NIH. outbreaks in swimming areas and infected drinking wells — which can cause bacillary dysentery, or shigellosis. Other harmful bacteria that could be present in wastewaters are Vibrio, Leptospira, Clostridium, Listeria and Yersinia, among others. Some of these bacteria can cause severe diseases such as cholera (Vibrio cholera), or Leptospirosis (Leptospira spp.) or listeriosis (Listeria). Listeriosis is a serious illness for humans, and may manifest as meningitis, or affect newborns due to its ability to penetrate the placenta. The prevention of listeriosis obliges cheese or meat-processing plants producing ready-to-eat foods, such as hot dogs and deli meats, to follow extensive sanitation procedures. Some of these other bacteria, even though they are not common in developed countries, are still prevalent in some parts of the world (Metcalf & Eddy, 2003).

4.1.3 Protozoa, Helminths and Viruses

There are other relevant microorganisms that ﬁt into the remaining three categories: Some pro- tozoans produce signiﬁcant impact on individuals with compromised immune systems, such as very young children, or old people, people undertaking cancer treatments or suffering from AIDS (Metcalf & Eddy, 2003). According to Metcalf & Eddy (2003), over the last century the presence of helminths infections has dramatically decreased due to the improvement of the sanitary system. This is the case over most developed countries as well. However, the constant immigration from countries where helminths are still endemic (worldwide, worms are one of the principal causative agents of human disease), should provoke a continuous protective response from the said sanitary systems, especially on what concerns wastewater treatment, since its way of transmission is through biosolid remains. There are also high levels of enteric viruses in waste and surface waters. These are continuously shed in an infected individual’s and cause a health risk to non-infected persons. From waterborne pathogenic viruses, Norovirus and rotavirus, for example, cause diarrhoeal disease. On the other

36 4.2 Microbial Source Tracking Methods hand, reoviruses and adenoviruses are usually associated with respiratory illnesses, gastroenteritis and eye infections (Metcalf & Eddy, 2003). Infection with diarrhoeal diseases can more common in cooler months, in warmer months of the year the infectious agent is rather of a bacterial nature, and has a signiﬁcant impact on children, more dramatically if undernourished, being the primary cause of childhood deaths in developing countries (Prescott et al., 1993).

4.2 Microbial Source Tracking Methods

4.2.1 The Point of Indicators

The reliability and accuracy of results strongly depends on the choice of the targeted microorganism. The indicator should accurately describe the quantity of faecal contamination in an aquatic environment. An ideal indicator should (Scott et al., 2002; Metcalf & Eddy, 2003):

• be thoroughly related to the presence of the pathogenic organism;

• have similar survival characteristics;

• be present in an equal or higher number than the pathogen;

• be easily detected and numerated;

• not reproduce outside the host organism;

• preferably be nonpathogenic.

Figure 4.3: Scanning electron micrograph of Escherichia coli, grown in culture and adhered to a cover slip. Image by Rocky Mountain Laboratories, NIAID, U.S. NIH.

Total and faecal coliforms have been the traditional microbial indicators throughout the world for determining the quality of waters (Noble et al., 2004; Petersen et al., 2005). However, in the last decades, studies have found signiﬁcant differences in the characteristics of coliforms and targeted pathogens, which limits their use (Brownell et al., 2007; McCarthy, 2009). Alternatively, studies (Grifﬁn

37 4. Microbiology et al., 2001; cit. in Scott et al., 2002; Brownell et al., 2007; McCarthy, 2009) have leaned into other microbes such as E. coli (see section 2.1.2 and Figure 4.3), enterococci and Clostridium perfigens (Scott et al., 2002). Recent studies try to evaluate the use of alternative faecal indicators such as anaerobes (genera Bacteroides and Bifidobacterium), viruses (e.g. Bacteroides fragilis phage and F-specific RNA coliphages) and faecal organic compounds (coprostanol). Savichtcheva and Okabe (2006) concluded that faecal anaerobes are generally unable to survive in aerobic conditions, so they are better suited to indicate recent faecal contamination. Also it is important to establish a reliable viral indicator, since bacterial indicators are not so well suited for viral pathogens detection. The authors state that there is an acceptable correlation between B. fragilis phages and coliphages and pathogenic enteroviruses. They also admit there could be advantages in using faecal organic compounds, especially in tropical climates, since most microbial indicators could multiply or be part of the local natural flora. Even if indicator organisms achieve these important changes and improvements, they still lack in pointing out the source of the detected faecal pollution, and understanding the origin of faecal pollution is paramount in assesing associated health risks as well as the actions necessary to prevent and mitigate them. (Scott et al., 2002)

4.2.2 Tracking the Source

Microbial Source Tracking is, as its name hints, tracing the origin of fecal pollution using microbiological, genotypic and phenotypic methods (Scott et al., 2002). Over the last decade there has been a strong tendency towards the development of Microbial Source Tracking (MST) methods, due to an increasing concern about the water quality, both to avoid the closure of recreational and bathing areas (e.g. U.S. and Canada), as well as to respect either international regulations (the case of European countries - European Union Directive 2006/7/EC) or home laws (U.S case - Clean Water Act 1972 and Federal Water Pollution Control Act 2002). According to Simpson et al. (2002), the majority of MST methods rely on fingerprint profiles of faecal bacteria, although viruses and protozoa also suggest to reliably discriminating between human and animal faecal sources. Following the authors’ report, it can be stated that studies using MST techniques are usually developed upon one of three approaches: (i) finding or choosing species indicative of the source: (ii) biochemical tests to differentiate sources; (iii) DNA fingerprinting using either genomic DNA or specific phylogenetic genes of faecal bacteria. There has been recently a strong tendency to rely on this last approach in source tracking, both to the great development of the methods and their general easy application, inexpensiveness and accuracy.

4.2.3 The Methods

Currently most MST methods fall into one of two broad categories: Culture-dependent and Culture- independent methods (Santo Domingo et al., 2007). The major difference between the two groups is that culture-based methods require the growth of microorganisms present in samples through microbial cultivation. If a method requires cultivation, then the target validation can be done either by

38 4.2 Microbial Source Tracking Methods comparing with a library (in these cases a compilation of microorganisms from different potential sources including those under study) or by phenotypic or genotypic analysis. An attempt to better illustrate these differences can be seen in Figure 4.4.

Figure 4.4: A rough classiﬁcation of current MST techniques. Adapted from USEPA (2005)

Examples of phenotypic analysis are the Antibiotic Resistance technique (that consists in indenti- fying the target through its resistance to different antibiotics) and Carbon Utilization Profile (e.g. API identification galleries). Simpson et al. (2002) criticize culture-based techniques on their tendency to underestimate the bacterial densities in environmental samples and their accuracy in identifying many bacterial isolates, when using conventional phenotypic characterization. Even though the described phenotypic methods are relatively simple and allow for the analysis of hundreds of isolates in a short period of time, Antibiotic Resistance presents some difficulties when applied to wide geographical areas and Carbon Utilization Profile lacks for reports of successful utilization in faecal source tracking. Some of the most recent MST methods are based on the detection of mitochondrial DNA. Different animals have different DNA sequences in the mitochondria so briefly, is it possible to distinguish them based on this characteristic. Cultivation-independent techniques use the Polymerase Chain Reaction (PCR) (see 4.3.1) and the quantitative-PCR (qPCR), that basically consist in fast reproduction of a DNA strain, very useful when one has but little DNA and multiple tests to perform. Santo Domingo et al. (2007) believe that currently MST methods are registering a transition toward cultivation-independent, library-independent techniques using PCR, and state some of the advantages of PCR techniques such as “having the potential of being sensitive, inexpensive, quantitative, and amiable to automation” adding that DNA can be preserved for future analysis and subject to multiple assays targeting multiple microbial targets. MST techniques that rely on DNA extraction usually target specific bacteria (e.g. Bacteroides, Bifi- dobacterium, etc. . . ) and Viruses (e.g. Enteroviruses, Adenoviruses and coliphages) (Santo Domingo et al., 2007). However, only a few years ago, Martellini et al. (2005) published the first report on the

39 4. Microbiology use of eukaryotic mitochondrial DNA (mtDNA) to both detect and differentiate faecal sources in water. It proved a simple and quick method that succeeded in differentiating between human, ovine, bovine and porcine from tissues, faeces, wastewaters, surface runoff and river water samples without cross reactivity with DNA from other species. The authors also argue that this method should prove better — sensitivity wise — when compared with the use of nucleic DNA, because each cell contains several copies of mtDNA, and it evolves much faster than its nucleic counterpart. Recent studies continue to corroborate the accuracy of mtDNA markers, developing new primers and sampling from different environments (Baker-Austin et al., 2010; Kortbaoui et al., 2009; Schill and Mathes, 2008). However, some authors Stoeckel and Harwood (2007); Balleste´ et al. (2010) suggest the combined use of several markers — despite the cost increase —, in a sort of source tracking “toolbox”, in order to overcome some slight inaccuracies of markers that might be caused by spatial and temporal variability or the still premature development of some markers.

4.2.4 Current Issues and Future Research

MST is a very useful tool when devising and implementing best management practices. However, over the last decade, several issues have been pointed out, primarily due to the lack of studies and publications on accuracy and performance of both existing and new methods (Santo Domingo et al., 2007). Simpson et al. (2002) believe that some of the major issues relate to the temporal and spatial variability of markers, as well as their detection limits, and the reproducibility of assays. Also, the running cost, technical expertise and time associated with certain MST’s sometimes conditions the use of potentially more accurate methods. The detection limit — sensitivity — of markers is extremely relevant to the reliability of results, and depends strongly on the approach chosen for source tracking. Simpson et al. (2002) suggest that a way to increase sensitivity of PCR methods would be targeting genes such as 16S rRNA, a gene that has several copies per cell. Gawler et al. (2007) studied the special variability of some markers between countries of the Altan- tic Rim, in Europe, and the United States, and concluded that the same markers could be use without prejudice to the sensitivity or specificity, although with significant variation in specificity from region to region. Most MST studies do not consider the influence of spatial and temporal variation, so it’s difficult to understand their impact in source tracking. Moreover, the authors find it reasonable to assume that the location of sampling sites will influence the reliability of results, considering the heterogeneity of watersheds. Also, as previously mentioned, some microorganisms will suffer season variations (see 4.2.1). As the basis of any new theory, the reproducibility of experiments is extremely relevant (Ullmo, 1969; Gawler et al., 2007), and Gawler et al. (2007) agree that the selected MST should give a similar answer when applied under standard conditions (e.g. on a particular watershed, the method should be able to always discriminate between animal sources common to that watershed). Some authors propose the use of traditional Faecal Indicators for source tracking, but library- dependent methods based on E.coli libraries can result in high source misclassification rates or the inability to classify many unknown source isolates (Stoeckel et al., 2004 cit. in Domingo et al., 2007). In addition to this, the density of pathogenic E. Coli and enterococci strains in the intestine tend to be

40 4.3 Case-Study Techniques

2-3 orders of magnitude less than those of their non-pathogenic counterparts, so methods tracking them also present immediate issues (Scott et al. 2005, cit in Domingo et al 2007). In spite of this Balleste´ and Blanch (2010) point towards non-traditional indicators, such as Bifidobacterium spp. and Bacteroides spp., as possible source-tracking indicators, because of their host-specificity. More than 100 different types of pathogenic viruses are excreted from human and animal faecal waste, and they often show good persistance in environmental waters. Also many viruses have a relatively stringent host association, making them excellent candidates for MST. Results so far, confirm that human-associated viruses have a high degree of host specificity (Fong and Lipp, 2005; cit. in Roslev and Bukh, 2011). The use of mtDNA has its limitations as well, for its sheding is not exclusively through faeces, but also other ways that include urine, blood, skin or saliva. However, faecal concentrations are superior to the rest, so this does not pose a serious problem. Roslev and Bukh (2011) also believe in the potential carryover of DNA from animals eaten by humans or other animals. Future studies in the use of mtDNA and further development of primers, should prove quite interesting. Recent studies show the possibility of designing highly species-specific primers (Baker-Austin et al., 2010; Kortbaoui et al., 2009; Schill and Mathes, 2008), but also assays that target larger groups of interest, such as mammals or other relevant sub-species with selected domesticated animals(Roslev and Bukh, 2011).

4.3 Case-Study Techniques

4.3.1 Polymerase Chain Reaction (PCR)

The polymerase chain reaction or PCR technique permits to synthesize large amounts of a DNA fragment. One of its major features is that it doesn’t require large quantities of DNA, to provide sufﬁcient material for accurately amplifying a speciﬁc target sequence. Among the most important agents of the polymerase reaction are primers. A primer is a strand of nucleic acid that serves as a starting point for DNA synthesis. They are required for DNA replication because the enzyme that catalyzes this process can only add new nucleotides to an existing strand of DNA. The polymerase starts replication at the 5’-end of the primer, and copies the opposite strand. After this step the synthesized chain is longer than the target sequence, because the primer has only set its beginning. Primers must come in pairs, one marking the beginning of the targeted sequence and the other marking the end. Later in the reaction the opposite primer will attach itself to the synthesized chain and start the polymerase from the end. The PCR is developed in three major steps (Prescott et al., 1993): (1) separation of complementary strands. The targeted DNA sequence is heat denatured so that separation takes place; (2) the fragmented DNA sequence is hydrogen bonded or annealed to primers, added in excess in this step so that the sequences won’t rather bond to each other; (3) DNA polymerase starts synthesizing in the reaction mixture, together with nucleoside triphosphates, in order to extend the primers and copy the targeted sequence. Each cycle doubles the amount of sequences present at the beginning of the process, so theoretically n cycles will provide 2n sequences of the targeted DNA. Which means 20

41 4. Microbiology

Figure 4.5: Example of single PCR cycles. In 3 cycles, the chain reaction produced 8 copies of the targeted sequence. Taken from Prescott et al. (1993).

42 4.3 Case-Study Techniques cycles will produce about 1 million copies and 30 cycles about 1 billion. Primer design is of the utmost importance to the outcome of the whole process, for a primer is responsible for choosing the sequence to reproduce. There are several studies (Martellini et al., 2005; Schill and Mathes, 2008; Kortbaoui et al., 2009; Baker-Austin et al., 2010) discussing their design and softwares able to design primers. The process described above is the marrow of the PCR technique, but there are more recent variants. The one described is more specifically known as conventional PCR. These newer versions of PCR, try to overcome several issues associated with this technique. For example the quantitative PCR (qPCR) quantifies the amount of DNA present, rather than just accusing presence/absence of the target sequence, the nested-PCR amplifies a more restricted sequence from a previously amplified one, and the multiplex PCR assays several sequences at one time, instead of a single one (Santo Domingo et al., 2007). To be sure, this technique is suffering a lot of evolution over the last years, and it remains a very promising tool in microbial source tracking, though in some cases it is the only availiable resource.

4.3.2 Mitochondrial Markers

It was said in the present study that mitochondrial markers have been successfully used in tracking mitochondrial DNA from exfoliated epithelial eukaryotes (see 4.2.3). Also the use of mtDNA is fairly recent and not thoroughly studied. Some of its evident advantages are: it has several copies per cell (like with the use of 16S rRNA); it carries sufficient sequence variation for a species-specific differentiation; it identifies the animal species directly, rather than identify host-associated bacteria or viruses. Source tracking using mtDNA is usually associated with PCR techniques in order to amplify the presence of the targeted sequence in a sample. After DNA extractions from samples, followed by amplification through one or more PCRs, the targeted sequence needs to be sided with the species- specific marker, for species identification. This is usually done visually, by staining sample and marker with ethidium bromide after performing an electrophoresis in agarose gels (Kortbaoui et al., 2009). Electrophoresis is a procedure which enables the sorting of molecules based on size and charge. Using an electric field, molecules (such as DNA) can be made to move through a gel (see Figure 4.6). Although the use of mtDNA in source tracking is relatively new, there have been remarkable devel- opments in mitochondrial markers that target numerous species. For example, Martellini et al. (2005) designed primers for amplification of faecal source mtDNA markes in humans, sheep, cows, and pigs in surface waters. Also Caldwell et al. (2007; cit. in Roslev and Bukh, 2011) developed primers for an assay targeting human, bovine and swine mt DNA in eukaryotic effluents, and Baker-Austin et al. (2010) did the same in order to target human, bovine, ovine and swine mtDNA from faecal pollution in surface waters and shellfish matrices. Schill and Mathes (2008) developed assays targeting mtDNA from dogs, cows, chicken, pigs, horses, Canada geese, white-tailed deer and humans in surface and ground water and influent wastewater, while Balleste´ et al. (2010) targeted humans, cattle and pigs also in wastewater.

43 4. Microbiology

Figure 4.6: Left: Electrophoresis apparatus. Photo by Jeffrey M. Vinocur. Right: An example result of an electrophoresis. Photo and diagram by ’Dr d12’.

44 5 Case-Study

Contents 5.1 Objectives ...... 46 5.2 The Basins ...... 46 5.3 The Laboratory ...... 51 5.4 Processing Results ...... 55

45 5. Case-Study

5.1 Objectives

Through the analysis of storwater samples, the present study seeks to: (i) evaluate the quality of stormwater runoff in the city of Lisbon, with special focus on faecal contamination and COD levels; (ii) further develop microbial source tracking methods, in particular the use of mitochondrial DNA markers designed speciﬁcally for species common to the urban environment (humans, cats and dogs).; (iii) assess the origin of registed faecal pollution in the city of Lisbon. Typical cobblestone sidewalks and experimented stormdrains can be seen in Figure 5.1. Sample data were collected from several points in three urban catchments in Lisbon: Alcantara (A), Bairro das Ilhas (I) and Madalena Street (M), during storm events between November 2011 and July 2012.

Figure 5.1: Left: Cobblestone sidewalks and tram lines in a Lisbon street. Right: The type of stormdrain used for sampling stormwater.

All storm and wastewater collected in central Lisbon, in particular in all three experimental catchments are conveyed to the Alcantara WWTP,which provides primary, secondary and tertiary treatment and, as mentioned before (see 3.2), has a wet weather ﬂow capacity of 6.6m3/s.

5.2 The Basins

The selected areas for collecting stormwater needed to provide a wide number and variety of samples. Their choice regarded several criteria such as land use diversity, concentration times, similar studies over the same area and accessibility, with some emphasis on the latter. The Alcantara basin had been subject to similar studies (Ferreira, 2006; Gondim, 2008), and for most collection points there were expected levels of pollution and faecal contamination. It is a very plural area in terms of land use and by itself would constitute a good source of samples. A total of six collection points were chosen. In January 2012 — a few months into the rain season —, there was an opportunity to widen the area for collecting samples, which resulted in the election of two more areas: the Madalena basin and the Bairro das Ilhas basin, the first located in the historical centre of Lisbon and being a touristic area with steep streets and intensive traffic, and the latter being mainly residential, with mainly local traffic and small buildings. The Madalena basin added three more collection points, while

46 5.2 The Basins the Bairro das Ilhas basin added six, summing a total of 15. Figure 5.2 pinpoints the location of the three experimental basins in the central Lisbon area, as well as the used rain gauges, whose data Instituto Geof´ısico Dom Luiz (IGIDL) and Laboratorio´ Nacional de Engenharia Civil (LNEC) were kind to supply (data from the rain gauge installed in Alcantara WWTP was also supplied by LNEC).

Figure 5.2: Location of the studied catchments, and rain gauges, in Lisbon. The leftmost basin is Alcantara (A), the topmost is Ilhas (I) and the one near the center is Madalena (M).

5.2.1 Alcantara Basin

This catchment was selected mainly because previous studies had already provided values to the levels of pollution and runoff characterization (Ferreira, 2006; Gondim, 2008), although no source tracking methods were ever applied. This experimental catchment, with an area close to 38.6 ha, is actually integrated in a larger one — the Alges-Alcantara basin — that flows to the Alcantara WWTP, in the west river-side of Lisbon. According to municipal records the Alges-Alcantara basin is divided into several sub-basins, of which E (previously D16 and D17 in Ferreira (2006)) matches the chosen experimental basin. On the other hand, the land use in this area is diverse: some areas have little impervious surfaces (e.g. the Agronomy faculty (ISA) of Technical University of Lisbon); others are residential areas mainly with houses with private gardens; and others have buildings and a high density of traffic and commerce (Ferreira, 2006). Typically the residential areas are located near the head of the basin whereas the more commercial areas, with more intense traffic, are located near its bottom. It is relevant to point out that this is an old industrial area of the city and a great part of the roads’ surface is dark basalt cobblestone, while the rest of them are regular asphalt. As to the sidewalks — like any other in Lisbon — is covered with white limestone. There are also several buses and trams running through this area and a taxi stop in Luis de Camoes Street. The commerce in this area is mostly street shops such as groceries, restaurants and cafes.

47 5. Case-Study

Figure 5.3: Location of sampling sites in Alcantara basin.

According to Ferreira (2006) the catchment has a unitary drainage system with the exception of ISA and a recent residential block that have a separate system and represent altogether about 20% of the whole system.Ferreira (2006) chose 6 points in this basin to collect runoff water from, of which ﬁve where maintained in this study. Later Gondim (2008) dropped some of these points, choosing another of the closest stormdrains, mainly due to practical issues such as difﬁcult access or the inability to produce enough sampling volume. One point that was altered was the one inside ISA, which was changed to one upstream of the original. The relation of the chosen stormdrains for this study to the previous ones in this area is gathered in Table 5.1.

Table 5.1: Collection points in the Alcantara basin, and comparison with previous studies.

Previous # Location Description Studies Little urban occupation and impervious- Next to south-east gate A1 ness; area with vegetation coverage and Gondim (2008) of ISA. trees; medium slope. In Joao de Barros St, Residential area, without commercial ac- A2 Ferreira (2006) next to Pedro Calmon St. tivity; streets with trees; smooth slope. Taxi stop in Luis de Residential area with buildings, intense Ferreira (2006); A3 Camoes St. traffic, bus lines; Near cafe; steep slope. Gondim (2008) Intense commercial activity and traffic; Ferreira (2006); A4 Bus stop in Calvario Sq. bus and tram lines; smooth slope. Gondim (2008) North side of Fontainhas Intense commercial activity and traffic; Ferreira (2006); A5 Sq. bus and tram lines; smooth slope. Gondim (2008) Square at the end of Co- Parking area, with high traffic; frequent A6 Ferreira (2006) zinha Economica St. flooding of stormdrain; flat slope.

48 5.2 The Basins

5.2.2 Bairro das Ilhas

The basin in Bairro das Ilhas — sub-basin L — is signiﬁcantly smaller than the previous one, and is mainly residential with little commercial activity, tight one-way streets and the average slope is generally smooth. The trafﬁc is of low intensity with areas of exclusive pedestrian access, and the only relevant green space is the Cesario´ Verde garden (4000 m2). It is also possible to see that some of the buildings’ rooftops drain directly to the pavement.

Figure 5.4: Location of sampling sites in Ilhas basin.

The total study area is of 6.25 ha, and six sampling stormdrains were selected, although one of them is most likely redundant due to its proximity to another point in the same conditions.

Table 5.2: Collection points in the Ilhas basin, and their general description.

# Location Description South-east corner of Cesario´ Verde Vegetation present; pervious areas; cobblestone sur- I1 garden. faces; little traffic, mostly pedestrian; medium slope. Cidade da Horta St, next to Ilha do Wide stairs; roof runoff drains directly to pavement; I2 Pico St (stairs). medium slope. Arroios St, corner with Ponta Del- I3 Somewhat higher traffic than the rest; steep slope. gada St. Completely impervious area; roof runoff drains into I4 Square at the end of Açores St. pavement; flat surface. Stairs connecting Ilha Terceira St. Narrow pedestrian area, very little pedestrian traffic; I5 and Cesario´ Verde garden. medium slope. I6 Same stairs as I5, but farther down. Same as I5.

49 5. Case-Study

5.2.3 Madalena

The third set of points is entirely in Madalena Street, right in the middle of the 18th century historical centre of Lisbon. The street has an intense commercial activity and trafﬁc, with bus lines and at the end of it, next to Martim Moniz Plaza trams as well. It has a steep slope, with its highest point around the middle length of the street, in Adelino Amaro da Costa Square. Three sample points were selected: one going up the street, at the top of it, in Adelino Amaro da Costa Square, and the last near the bottom of the street, were the trams join in.

Figure 5.5: Location of sampling sites in Madalena Street.

Table 5.3: Collection points in the Madalena basin, and their general description.

# Location Description Madalena St, upward, in the gutter Intense traffic and commercial activity; impervious M1 next to number 127. surfaces; steep slope. Madalena St, next to Adelino Intense traffic and commercial activity; impervious M2 Amaro da Costa Sq. surfaces; next to restaurants; steep slope. Madalena St, downward, corner Intense traffic and commercial activity; cobblestone M3 with Condes de Monsanto St. pavement; tram lines; steep slope.

5.2.4 WWTP in Alcantara

The Wastewater Treatment Plant in Alcantara is the most advanced in Portugal. A general description of its treatment processes will be given here, in order to reﬂect current treatment policies and technology in Portugal. As it was said, this WWTP, serves the better part of Lisbon, an equivalent population of about 800.000 habitants, over an area of 37 km2 and the collector system sums up to 22.4 km. It has a deodorization equipment capable of treating 160.000 m3/h of polluted air that effectively reduces unpleasant odours.

50 5.3 The Laboratory

The liquid phase reaches the entry work, through a solids retention well and a grate, to retain any supernatants with dimensions over 6 mm. Wastewater is then elevated in Archimedes bolts, for sand and grease removal. The primary treatment is based in the MULTIFLOTM and ACTIFLO R technologies. In dry weather the MULTIFLOTM decanters/thickeners process up to 3.3 m3/s with no reagent addition. In wet weather conditions, for flows over 3.3 m3/s, the ACTIFLO R kicks in to treat the flow surplus (up to an additional 3.3 m3/s). The ACTIFLO R technology is based in the addition of micro-sand for high density flocculation, and it is able to work under heavy heads such as 130m/h with high efficiency rates. The secondary treatment is made in fifteen biofilters — by BIOSTYRTM, where organic matter is eliminated and suspended solids removed in their majority, through the combined effects of adsorp- tion, hydrolysis and metabolization in the biofilm that develops at the filter’s surface. The effluents are then submitted to a tertiary treatment that seeks to disinfect them through UV lamps, in order to ensure the legal standards for microbiological parameters for discharges in bathing waters. The solid phase, or sludges, resulting from the primary and secondary treatments are thickened in the MULTIFLOTM equipments, dehydrated in four centrifuges and chemically stabilized with quicklime, making it a more attractive asset for agricultural uses.

5.3 The Laboratory

5.3.1 Collected samples

All samples were collected from boxes planted in stormdrains. Despite varying in size and depth, all boxes were put in such a way as to maximize the sample volume, and with a small enough opening as to try to prevent more water coming in once the box was full. Some were attached with wires to the concrete side of the stormdrain, some were balanced in small pine wood beams placed to support the box plus about 4 liters of stormwater. Per point were collected 2 liters in sterilized ﬂasks of 1 liter each.

Figure 5.6: Placing of sample boxes in stormdrains. Sediment deposition after a rainfall event.

51 5. Case-Study

The difﬁculties in the collection process are listed as follows, though most of which were shared by Ferreira (2006):

• The need to maintain a fast response to the rain event. As soon as it starts raining the teams need to go to the catchment to rapidly collect the samples and bring them to the lab. The problem is that there is no certainty when it is going to rain. The response is very difﬁcult in the weekend (lab closed) and during rush hours, because trafﬁc is chaotic.

• Impossible acess to some of the stormdrains, when a car is parked on top of it, or too near as to make it impossible to lift the heavy iron lid.

• the stormdrain does not produce enough sampling volume. It usually occurs when rainfall event is not strong, but it always happened in I1, for example.

• in point A3, taxi drivers confessed to pee in the stormdrain at late hours, thus possibly contaminating the sample or damaging the representativeness of stormwater pollution sources in the catchment. Also some cafes and restaurants threw food and washing discharges directly into stormdrains, while others only to the sidewalk or pavement.

• usually the sample boxes contained larger volumes of stormwater than those brought to lab (2 liters per point), so if the sample was not well homogenized, it could render the sample useless.

• There was also the risk of contaminating the sample with one’s own DNA, since the MSTs were also tracking human mtDNA, or transporting DNA from one sample to another. Team members always were gloves, and always cleaned them with alcohol before and after collecting each sample.

• E. coli and enterococci assessment was not always possible, due to large waiting periods for the arrival of the kits, and for most samples collected at weekends, the die-off rate was to great to get an acceptable representativity.

• There was also the risk of theft or damage of the equipment, since it was placed in public places to which everyone had access. At two different times some boxes disappeared.

Between rainfall events, an intervention was necessary both to clean the boxes of debris and solid particles and verify the conditions of the material. The boxes were washed with potable water, and emptied the day before a predicted rainfall event. Again, all team members were gloves in this phase as well.

5.3.2 COD analysis

COD assessment was made through the use of kits. if the sample was frozen, COD assessment could be done later. COD analysis was performed after all samples were collected using the COD cell test C4/25 (WTW) of the range 25-1500 mg/l. The stored samples were thawed and 3 mL were added to a cell test. The solution was carefully homogenized and incubated for 2 hours at 148oC in

52 5.3 The Laboratory a thermoreactor. The cell tests were cooled down for 10 minutes, shaken to homogenize the solution and left to cool down to room temperature protected from light. The results were read in a SpectroFlex 6600 (WTW) photometer.

5.3.3 E. coli and enterococci analysis

E. coli and enterococci assessment was made using a method which reveals the presence of the targeted bacteria by ﬂuorescence under UV light. The analysis were performed in the same day the samples were collected using Quanty-tray/2000 (IDEXX), Colilert-18 (IDEXX) and Enterolert (IDEXX) reagents, respectively. Samples were adequately diluted in 100 mL of sterile distilled water and a pack of the Colilert-18 or Enterolert reagent was added and diluted. The solution was poured into a Quanty-tray/2000 and sealed using a Quanty-tray sealer. The trays were incubated for 24h at 37oC for E. coli and 41.5oC for enterococci. After incubation, the tray was observed at UV light and the ﬂuorescent (positive) wells were counted. Figure 5.7 shows a good example of this procedure. E. coli and enterococci countings were obtained using the most probable number (MPN) table provided by the manufacturer.

Figure 5.7: Fluorescence indicates the presence of E.coli in water sample, using the Colilert kit. Taken from Rivera and Rock (2011).

5.3.4 DNA extraction procedures

DNA from faeces was extracted — in order to test the markers — using the QIAamp DNA Stool Kit (QIAgen) following the manufacturer’s instructions. About 220 mg of faeces were put into a 2 mL tube, added 1.6 mL of Buffer ASL and thoroughly homogenized by vortex. The mixture was centrifuged 1 minute at 14500 rpm and 1.4 mL of the supernatant was transferred into a new 2 mL tube. An InhibitEX tablet was added to the supernatant, mixed by vortex and incubated for 1 minute at room temperature. The sample was again centrifuged for 3 minutes at 14500 rpm, the supernatant transferred to a new 1.5 mL tube and centrifuged in the same conditions. 600 µL of the supernatant were transferred to a 2 mL tube containing 25 µL of proteinase K. 600 µL of Buffer AL were added to the mixture and mixed by vortex. The mixture was incubated 10 minutes at 70oC. After incubation,

53 5. Case-Study

600 µL of ethanol 96% were added and mixed by vortex. 600 µL of the lysate was transferred to a QIAamp spin column and centrifuged 1 minute at 14500 rpm. This step was repeated two more times until the total volume of lysate passed through the spin column. 500 µL of Buffer AW1 were added to the spin column and centrifuged at 14500 rpm for 1 minute. 500 µL of Buffer AW2 were added to the spin column and centrifuged at 14500 rpm for 3 minutes. The “empty” spin column was centrifuged 1 minute at 14500 rpm. 200 µL of Buffer AE were added to the spin column and left for 1 minute at room temperature. DNA was eluted by centrifugation at 14500 rpm for 1 minute and preserved at -80oC.

Extracting DNA from samples. DNA from concentrated samples was extracted using the QIAamp DNA mini Kit (QIAgen) following the instructions from the manufacturer. 200 µL of sample were added to a 1.5 mL tube with 20 µL of proteinase K. 200 µL of Buffer AL were added and mixed thoroughly by vortexing. The mix was incubated for 10 minutes at 56oC. 200 µL of ethanol 96% were added and mixed by vortex. The total lysate was transferred to a QIAamp Mini spin column and centrifuged 1 minute at 8000 rpm. 500 µL of Buffer AW1 were added to the spin column and centrifuged at 8000 rpm for 1 minute. 500 µL of Buffer AW2 were added to the spin column and centrifuged at 14500 rpm for 3 minutes. The “empty” spin column was centrifuged 1 minute at 14500 rpm. 200 µL of Buffer AE were added to the spin column and left for 1 minute at room temperature. DNA was eluted by centrifugation at 8000 rpm for 1 minute and preserved at -80oC.

5.3.5 Single and nested PCR procedures — Primer design

In order to identify the fecal pollution origin of the collected samples, mitochondrial DNA present in the samples was analyzed through nested PCR using specific primers for each animal in study: human, cat and dog. Human primers were the same described by Martellini et al. (2005). The mitochondrial DNA sequences of all animals in study were aligned using the ClustalW pro- gram and specific primers were obtained using the Primer Express software. Primers specificity was confirmed using BLAST. Primers were provided by Thermo Fisher Scientific.The primers sequences are shown in Table 5.4. PCR was performed in a Veriti 96 well thermal cycler (Applied Biosciences) using ilustra puRe- Taq ready-to-go PCR beads (GE Healthcare). Single PCR was performed in 25 µL volume using 0.4 pmol/µL of each primer, 5 µL of extracted DNA diluted to 10−1 and one PCR bead. Nested PCR was performed in the same conditions except that 1 µL of the single PCR reaction was used as tem- plate DNA and internal primers were used. In the case of Cat and Dog identification, nested PCR was performed simultaneously in the same reaction for both animals. 1 µL of each single PCR reaction and the four internal primers were used. PCR cycle conditions are shown in Table 2.

54 5.4 Processing Results

Table 5.4: Primers used for both single and nested PCR, for each species.

Amplicon length (bp) Human Single PCR primers Humito2-G 5’-AGCCCTTCTAAACGCTAATCCAAGCCT-3’ 659 Humito2-D 5’-CTTGTCAGGGAGGTAGCGATGAGA-3’ Nested PCR primers Humito11-G 5’-CCACTACTAGGCCTCCTCCTA-3’ 612 Humito11-D 5’-TAGCGATGAGAGTAATAGATAGGG-3’ Dog Single PCR primers Dogmito1-F 5’-ATGGCTCTAGCCGTTCGATTAAC-3’ 638 Dogmito1-R 5’-GGCTAGGAGGACTGAGGTGTTGAG-3’ Nested PCR primers Dogmito2-F 5’-CATTAGGATTCACAACCAACCTGTTA-3’ 236 Dogmito2-R 5’-AATAATGCCGGTAGGAGGTCAG-3’ Cat Single PCR primers Catmito1-F 5’-CCTGTCCACACTACTTGTACTCATCGC-3’ 539 Catmito1-R 5’-AGATGGTTGTTTAGGATGGCTACG-3’ Nested PCR primers Catmito2-F 5’-ATTTGATCCTATAGGGTCCGCC-3’ 350 Catmito2-R 5’-CCTATGAGCGACATGATGAAAGC-3’

Table 5.5: PCR steps and cycle conditions.

Temperature (oC) Step Human Dog and Cat Number of cycles Hold time (s) Pre incubation 94 94 1 300 Pre annealing 55 59 1 300 Ampliﬁcation Elongation 72 72 35 120 Denaturation 94 94 35 40 Annealing 55 59 35 60 Cooling 72 72 1 600

PCR products were observed by agarose gel electrophoresis in 2.5% SeaKem LE agarose (Lonza) gels. 10 µL of PCR product were loaded with 1 µL of 10x DNA loading buffer. 2 µL of 100 bp DNA ladder (New England Biolabs) were also loaded. Gels were run at 60 V using TAE buffer (1x). The DNA was stained by immersion in ethidium bromide solution. The resulting gel was visualized with the G: BOX (Syngene).

5.4 Processing Results

5.4.1 Rain Data

Rain data was kindly supplied by Laboratorio´ Nacional de Engenharia Civil (LNEC), and Instituto Geof´ısico Dom Luiz (IGIDL). LNEC has two udometers, one installed in its headquarters (LNEC-NES) and the other in the Alcantara WWTP (LNEC-ETAR), while IGIDL has only one and it is installed in its headquarters (see Figure 5.2).

55 5. Case-Study

Results obtained of instantaneous precipitation for 2011 and 2012, in each of the rain gauges — LNEC-NEC, LNEC-WWTP and IGIDL — are show in Figures 5.8 and 5.9. Also shown in the graphics are the dates of several campaigns, coupled with weekend denotation. Weekends were difﬁcult days to collect data, because the Lab was closed and there was no appropriate place to store the collected stormwater. The LNEC-WWTP rain gauge didn’t register any data from 04-11-2011 around 11a.m. to 28-11-2011 around 10a.m., possibly due to a malfunction in the equipment.

Figure 5.8: Rain data from experimental campaigns for 2011, in Lisbon.

Figure 5.9: Rain data from experimental campaigns for 2012, in Alcantara.

It is important to notice that all udometers from show relative accordance in precipitation in 2011, though the WWTP is further apart, while in the ﬁrst months of 2012 there are precipitation events registered on either one that are not captured in the other. Also noticeable is the difference between the udometers LNEC-NES and IGIDL, since they are in the same part of Lisbon, only about 2 km apart, which is possibly due to the use of different rain capture devices, different installation conditions or relevant spacial variability.

56 5.4 Processing Results

5.4.2 Crossing Data

After the COD analysis with the kits and the E. coli and enterococci assessment, the obtained results were crossed with the rain data, in order to have a qualitative idea of the pollutant concentration versus the collection catchment and/or the antecedent dry weather. Figures 5.10, 5.11 and 5.12 show the maximum values obtained in each sortie per storm event for the parameters COD, E. coli and enterococci, respectively.

Figure 5.10: Maximum COD values obtained per event and per catchment in the experimental campaigns.

Figure 5.11: Maximum E. coli values obtained per event and per catchment in the experimental campaigns.

Figure 5.12: Maximum EF values obtained per event and per catchment in the experimental campaigns.

57 5. Case-Study

After the COD analysis with the kits and the E. coli and enterococci assessment, a simple statistical analysis was made, in order to compare the obtain results with the previous ones obtained by Ferreira (2006) and Gondim (2008). The results obtained for the Alcantara basin, Ilhas basin and Madalena basin are presented in Table 5.6, Table 5.7 and Table 5.8, respectively.

Table 5.6: COD, E. coli and enterococci values for the Alcantara catchment.

Alcantara COD (mg/l) E. coli (MPN/100mL) Enterococci (MPN/100mL)

Average 440 5.1E+04 6.2E+04 Standard Deviation 350 6.5E+04 1.1E+05 skewness 0.51 1.85 3.56 maximum 1100 2.4E+05 4.8E+05 3rd Quartile 730 6.4E+04 5.6E+04 median 360 2.4E+04 2.9E+04 1st Quartile 140 1.1E+04 1.2E+04 Minimum 33 1.0E+02 8.0E+01 Amplitude 1100 2.4E+05 4.8E+05 Samples 19 18 18

Table 5.7: COD, E. coli and enterococci values for the Ilhas catchment.

Ilhas COD (mg/l) E. coli (MPN/100mL) Enterococci (MPN/100mL)

Average 100 6.0E+03 1.3E+04 Standard Deviation 76 1.4E+04 1.7E+04 skewness 2.35 3.29 1.71 maximum 350 4.8E+04 4.8E+04 3rd Quartile 120 4.4E+03 1.5E+04 median 70 1.4E+03 6.4E+03 1st Quartile 55 7.2E+02 2.4E+03 Minimum 45 1.3E+02 9.3E+02 Amplitude 300 4.8E+04 4.7E+04 Samples 17 12 12

Table 5.8: COD, E. coli and enterococci values for the Madalena catchment.

Madalena COD (mg/l) E. coli (MPN/100mL) Enterococci (MPN/100mL)

Average 340 2.6E+04 4.7E+04 Standard Deviation 290 6.9E+04 5.7E+04 skewness 1.00 3.26 1.17 maximum 940 2.4E+05 1.6E+05 3rd Quartile 500 4.5E+03 7.2E+04 median 240 2.1E+03 2.4E+04 1st Quartile 110 9.9E+02 3.3E+03 Minimum 76 1.7E+02 1.5E+03 Amplitude 860 2.4E+05 1.5E+05 Samples 14 12 12

For the Alcantara basin — the same basin studied by Ferreira (2006) and Gondim (2008) — the average for the COD parameter was considerably superior, of 440 mg/l, while Ferreira (2006) only

58 5.4 Processing Results obtained 203 mg/l and Gondim (2008) only 150mg/l, but still inferior to studies in other countries (see 3.1.1). The mean COD in the Madalena basin was superior as well, more than twice the average obtained by Gondim (2008). Only the Ilhas basin had a COD average that was below that of previous studies, similar to what Gondim (2008) obtained. E. coli and enterococci values range in all three basins between 6.0E+03 MPN/100mL and 5.1E+04 MPN/100mL, and between 1.3E+03 MPN/100mL and 6.2E+04 MPN/100mL, respectively. It is also noticed high standard deviation values, usually of the same order of magnitude as the averages. As expected, because of its land use and characteristics, the Ilhas catchment show lower levels of pollution than the other two. Madalena catchment shows intermediate levels, though closer to those registered in Alcantara. The generally high values registered for all basins are most likely related to the little number of samples and because it was a very dry rainy season.

Figure 5.13: Box-and-whiskers chart, showing the frequency distribution of COD, E. coli and enterococci concentration values, for each basin.

The average concentration values for COD, E. coli and enterococci — shown again in Figure 5.13 — are extremely relevant to explain the impact that untreated stormwater discharges have in receiving waters. With the exception of most values obtained in Bairro das Ilhas, concentrations clearly ex- ceed the legislated limit for COD discharge of 150mg/l, defined by DL-236/98 (Portugal). Also for E. coli and enterococci, the sampled stormwaters systematically exceeded the values legislated by Directive 2006/7/EC, which stipulates limits of 200 MPN/100mL and 500 MPN/100mL, for E. coli and enterococci respectively. The following tables present the results of mtDNA testing. It shows whether samples were positive or negative for human, dog or cat DNA presence, along with E. coli and enterococci values for that specific sample, allowing discrimination between points of the same basin. Table 5.9 presents DNA tests and final results for Alcantara, and Table 5.10 and Table 5.11 for Ilhas and Madalena basins, respectively.

59 5. Case-Study

Table 5.9: Final data for the Alcantara catchment. N/A - not availiable. ’+’ positive and ’-’ negative for speciﬁc DNA presence.

E. coli Enterococci PCR # Date (MPN/100mL) Human Dog Cat N/A N/A + - - test sample N/A N/A - - - 25-10-2011 N/A N/A - - - 15-03-2011 A1 N/A N/A N/A + + 30-03-2012 (morning) N/A N/A N/A - - 30-03-2012 (afternoon) 9,77E+04 4,84E+05 + + - 03-05-2012 1,00E+02 1,20E+03 - - - 19-04-2012 A2 N/A N/A - + - 26-04-2012 N/A N/A + - - 15-03-2012 N/A N/A N/A - - 16-03-2012 N/A N/A N/A - - 24-03-2012 N/A N/A N/A + + 30-03-2012 (morning) N/A N/A N/A - - 30-03-2012 (afternoon) A3 1,09E+03 8,00E+01 N/A - - 15-04-2012 1,41E+05 8,66E+04 - - - 19-04-2012 2,42E+05 6,49E+04 - + - 26-04-2012 3,13E+04 1,98E+04 - + - 03-05-2012 6,90E+04 1,45E+05 - + + 07-05-2012 N/A N/A + - - 25-10-2011 N/A N/A + - - 15-03-2012 N/A N/A N/A - - 15-03-2012 N/A N/A N/A - + 16-03-2012 N/A N/A N/A - - 30-03-2012 A4 4,84E+04 4,84E+04 N/A - - 15-04-2012 1,72E+04 2,61E+04 + - - 19-04-2012 2,01E+04 3,08E+04 - - - 26-04-2012 3,72E+04 6,10E+03 + + - 03-05-2012 2,82E+04 1,92E+04 - + + 07-05-2012 N/A N/A + + + test sample N/A N/A + - - 25-10-2011 N/A N/A N/A - - 03-02-2012 N/A N/A + - - 15-03-2012 N/A N/A N/A - - 16-03-2012 N/A N/A N/A - + 18-03-2012 A5 N/A N/A N/A + - 30-03-2012 N/A N/A N/A - + 30-03-2012 1,24E+03 1,16E+03 N/A - - 15-04-2012 1,66E+04 5,79E+04 - + - 26-04-2012 4,02E+03 2,75E+04 - + + 03-05-2012 1,72E+04 5,19E+04 + + + 07-05-2012 N/A N/A + - + test sample N/A N/A + - - 25-10-2012 N/A N/A + - + 15-03-2012 N/A N/A N/A - + 16-03-2012 A6 N/A N/A N/A + + 30-03-2012 N/A N/A N/A - - 30-03-2012 N/A N/A - - - 19-04-2012 1,41E+05 4,35E+04 + + - 26-04-2012 8,82E+03 9,28E+03 - + + 03-05-2012

60 5.4 Processing Results

Table 5.10: Final data for the Ilhas catchment. N/A - not availiable. ’+’ positive and ’-’ negative for speciﬁc DNA presence.

E. coli Enterococci PCR # Date (MPN/100mL) Human Dog Cat N/A N/A N/A + + 30-03-2012 (morning) I1 N/A N/A N/A - - 30-03-2012 (afternoon) N/A N/A N/A + - 30-03-2012 (morning) N/A N/A N/A - - 30-03-2012 (afternoon) I2 4,84E+04 4,84E+04 N/A - - 15-04-2012 N/A N/A - - - 19-04-2012 N/A N/A N/A + - 30-03-2012 (morning) I3 N/A N/A N/A + - 30-03-2012 (afternoon) 7,22E+03 1,45E+04 N/A - + 15-04-2012 N/A N/A N/A - - 30-03-2012 (morning) N/A N/A N/A - - 30-03-2012 (afternoon) 1,26E+02 9,30E+02 N/A - - 15-04-2012 I4 6,49E+04 2,42E+05 - - - 19-04-2012 N/A N/A - + + 26-04-2012 8,20E+02 2,52E+03 + + + 03-05-2012 1,24E+03 1,92E+03 - + + 07-05-2012 N/A N/A N/A - - 30-03-2012 (morning) N/A N/A N/A - - 30-03-2012 (afternoon) 3,98E+03 4,84E+04 N/A - + 15-04-2012 I5 N/A N/A - - - 19-04-2012 1,60E+03 1,46E+03 - + + 26-04-2012 4,00E+02 1,55E+04 - + + 03-05-2012 4,00E+02 7,98E+03 + + + 07-05-2012 N/A N/A N/A - - 30-03-2012 1,69E+03 9,77E+03 N/A - - 15-04-2012 I6 5,84E+03 3,42E+03 + + + 03-05-2012 8,20E+02 4,76E+03 + + + 07-05-2012

61 5. Case-Study

Table 5.11: Final data for the Madalena catchment. N/A - not availiable. ’+’ positive and ’-’ negative for speciﬁc DNA presence.

E. coli Enterococci PCR # (MPN/100mL) Human Dog Cat Date N/A N/A N/A - + 16-03-2012 N/A N/A N/A - - 24-03-2012 N/A N/A N/A - - 30-03-2012 (morning) N/A N/A N/A + + 30-03-2012 (afternoon) M1 1,70E+02 8,70E+03 N/A + + 15-04-2012 2,26E+03 3,87E+04 - - - 19-04-2012 2,01E+03 1,45E+03 - + - 26-04-2012 3,50E+03 1,54E+05 + + + 03-05-2012 1,48E+03 6,90E+04 + + + 07-05-2012 N/A N/A N/A - - 16-03-2012 N/A N/A N/A - + 18-03-2012 N/A N/A N/A - - 30-03-2012 (morning) N/A N/A N/A - - 30-03-2012 (afternoon) M2 4,84E+04 4,84E+04 N/A - - 15-04-2012 N/A N/A - - - 19-04-2012 2,31E+03 3,82E+03 - + - 26-04-2012 7,64E+03 5,24E+03 + + + 03-05-2012 1,24E+03 7,94E+04 - + + 07-05-2012 N/A N/A N/A - - 16-03-2012 N/A N/A N/A + + 24-03-2012 N/A N/A N/A - - 30-03-2012 (morning) M3 N/A N/A N/A - - 30-03-2012 (afternoon) 4,84E+04 4,84E+04 N/A - - 15-04-2012 2,42E+05 1,55E+05 - - - 19-04-2012

Some of the collection points inicially studied eventually show some systematic problems, such as:

• A1 — Accumulated a lot of dirt;

• A2 — difﬁcult access due to parked vehicles;

• A4 — frequently contained soup and/or milk;

• A5 — equipment disappeared twice during campaigns;

• I1 — sometimes desirable collection is not possible due to a deﬁcient installation of the box;

• I2 — sometimes ﬂooded and clogged;

• I3 — difﬁcult access due to parked vehicles;

• M2 — sometimes contained detergent and washing waters;

• M3 — equipment disappeared once before a campaign.

The information of Tables 5.9 to 5.11 reveals a singniﬁcant faecal contamination, based on the high levels of both faecal indicator bacteria (E. coli and enterococci). The PCR-based analysis allows to pinpoint the sources associated with that contamination, and also gives a geographical notion of

62 5.4 Processing Results its distribution. Results show that the three studied species contribute a relevant portion of this faecal contamination, having only 12% of the samples shown no presence of any of the three species. Also, 9% of total samples contained faecal contamination from all three species simultaneously. Faecal contamination with human source was detected in 44% of the tested samples, with canine source in 38% of the tests and with feline source in 35% of the tested samples (see Table 5.12). It was expected that dogs would be the highest source detected, but the fact that human faecal presence was higher can be a biased result due to the small amount of testing — total tests for human mtDNA were 50 samples, and for cat and dog were 100 samples.

Table 5.12: Count of Positive/Negative tests for Human, Dog and Cat mtDNA.

Human Dog Cat Alcantara Positive 15 52% 17 35% 15 31% Negative 14 48% 32 65% 34 69% total 29 49 49 Ilhas Positive 4 33% 12 44% 11 41% Negative 7 58% 15 56% 16 59% total 12 27 27 Madalena Positive 3 33% 9 38% 9 38% Negative 6 67% 15 63% 15 63% total 9 24 24 Total positive 22 44% 38 38% 35 35% negative 27 54% 62 62% 65 65% total samples 50 100 100

According to these results there is a strong faecal presence of human origin, especially in the Al- cantara basin, possibly due to intense pedestrian trafﬁc and commercial activities such as restaurants and cafes, but also to the relevant night life with discos. In Bairro das Ilhas there is the highest Cat and Dog rate of positives, which might reﬂect the quiet, residential occupation of the area. Madalena Street shows signs of lower animal faecal contamination than Ilhas, though the human ratio remains the same.

63 5. Case-Study

64 6 Conclusions and Future Work

Contents 6.1 Epilogue ...... 66 6.2 Prospective Sequels ...... 68

65 6. Conclusions and Future Work

6.1 Epilogue

The presence study was developed in order to evaluate the quality of urban stormwater runoff — through mean concentrations of COD, E. coli and enterococci — and also to determine the origin of the microbial faecal pollution load (between human, canine and feline sources) — through the use of mitochondrial DNA markers and the single and nested PCR techniques. DNA has become a very fashionable subject over the last decade, due to the increased education in school and research on the subject, to the ever more simple methods and applications related with it and to the spreading of its concept into other sciences and arts. One can without effort ﬁnd references to DNA in television series and movies, but also in art — as can e seen in Figures 6.1, 6.2, and 6.3.

Figure 6.1: The Helix Bridge in Singapore. photo by c Christopher Frederick Jones.

The knowledge from previous works, both in Portugal and worldwide, uncovers a great spatial and time variability of data, when it comes to assessing diffuse sources of pollution in stormwater runoff. A large number of phenomena is still not fully understood, and controlled, which makes modeling and calibration a rather difﬁcult process (Gondim, 2008). The methods of sampling were found satisfactory, but this work could greatly beneﬁt from larger and longer campaigns, and larger collecting teams. There is a risk that the manual collection option allowed interference in results, but on the other hand it ensured a satisfactory mixing and cleaning process, as well as input from in situ observations. The small opening in the box seemed to produce the desired effect of preventing dilution of suspended materials and deposition of sediments. Another downside of the collected data is the inability to calculate EMCs, and SMCs due to the collection of just one sample per point and per storm event. These tools would be of great use to compare with international literature, and to better assess the peak concentration for each parameter. It is known that the analysed concentration is in the beginning of the event, and it is assumed that these are not far from EMC values (Gondim, 2008) and peak concentration, but this can be a biased evaluation. The mitochondrial markers analysis was an innovative technique in microbial source tracking research in urban faecal pollution, and presents a great potential in understanding and preventing diffuse source pollution, as well as a promising tool in best management practices in general and urban

66 6.1 Epilogue water cycle management in particular. It will be an even greater asset with the widening of its use to other urban species such as rats, pigeons, and gulls. Though not in the same context, other authors have used this approach (mtDNA) and found high speciﬁcity and sensibility in their results (Balleste´ et al., 2010). It is concluded that most COD levels are high, and above legal limit for direct stormwater discharge — with mean concentrations ranging between 100 mg/l and 440 mg/l —, and that there is a strong faecal contamination — mean concentrations of E. coli between 6.0E+03 MPN/100ml and

5.1E+04 MPN/100ml, and enterococci between 1.3E+03 MPN/100ml and 6.2E+04 MPN/100ml. In fact in every chosen point of all three basins it was possible to detect at least once faecal contamination of one of the three sources. These high levels of the analysed parameters — the vast majority of which is well above legal limit for direct discharge — point towards the need of stormwater treatment prior to discharging in the Tagus River. It is noticed that, as expected, the Bairro das Ilhas catchment has the highest presence of faecal contamination from cats and dogs, because it is mostly a residential area and most pet animals belong to one of these two species. Alcantara shows the highest rate of human detections, probably because of the night life in this area. However it was very surprising to detect, in the end, more human faecal contamination in stormwater than either cats or dogs.

Figure 6.2: “Double Helix II” by Beverly St. Clair c 2004. Coloured fabric. photo by Susan Byrne.

67 6. Conclusions and Future Work

6.2 Prospective Sequels

Future evaluations of stormwater quality in Lisbon should be based on extensive sampling, enough to be as most statistically relevant as possible, in order to produce a sound analysis of stormwater pollution. This subject would benefit greatly with thorough computational modeling, that could study correlation of the several phenomena involved, based on a solid statistical analysis, and especially in what concerns correlation of stormwater pollution with geomorphologic effects and those of land use, with the help of GIS instruments that have greatly developed lately and are of easy access. EMCs and SMCs should be provided for comparison with international literature values, and for the better understanding of pollutant build-up, wash-off and transport. The first flush phenomenon should be thoroughly assessed, and adjusted to the several sampling sites, considering, and previous dry weather conditions. To complete the analysis of stormwater quality, an assessment of heavy metals present in stormwater should be conducted. The study of cases researching and proposing uses for stormwater as an urban resource should be very interesting, and when allied with a sound cost-benefit analysis could provide both political, economical and environmental solutions to address the two problems of disposing of stormwater and getting water for other uses. Specific legislation should be dedicated to the collection treatment and disposal of stormwater. Also a restructuration of microbiological parameters for water quality standards, which should be more diverse — as to include viruses for example — and to ensure the monitoring of both discharged and receiving waters under the light of the proposed parameters. Future studies should also focus on the use of mitochondrial markers for other urban species, and it would be very interesting to know a quantitative contribution of their faecal contamination.

Figure 6.3: “The Double Helix Mutation of Increased Compassion” by Franco Castelluccio.

68 Bibliography

Ahnert, M., Tranckner,¨ J., Gunther,¨ N., Hoeft, S., and Krebs, P. (2009). Model-based comparison of two ways to enhance WWTP capacity under stormwater conditions. Water science and technology, 60(7):1875–83.

Al Bakri, D., Rahman, S., and Bowling, L. (2008). Sources and management of urban stormwater pollution in rural catchments, Australia. Journal of Hydrology, (356):299–311.

Bach, P.M., McCarthy, D. T., and Deletic, A. (2010). The development of a novel approach for assessment of the ﬁrst ﬂush in urban stormwater discharges. Water science and technology, 61(10):2681– 8.

Baker-Austin, C., Rangdale, R., Lowther, J., and Lees, D. N. (2010). Application of mitochondrial DNA analysis for microbial source tracking purposes in shellﬁsh harvesting waters. Water Science & Technology, 61(1):1–7.

Balleste,´ E. and Blanch, A. R. (2010). Persistence of Bacteroides species populations in a river as measured by molecular and culture techniques. Applied and environmental microbiology, 76(22):7608–16.

Balleste,´ E., Bonjoch, X., Belanche, L. a., and Blanch, A. R. (2010). Molecular indicators used in the development of predictive models for microbial source tracking. Applied and environmental microbiology, 76(6):1789–95.

Bergman, M., Hedegaard, M. R., Petersen, M. F., Binning, P., Mark, O., and Mikkelsen, P. S. (2011). Evaluation of two stormwater inﬁltration trenches in central Copenhagen after 15 years of operation. Water Science & Technology, 63(10):2279.

Brownell, M. J., Harwood, V. J., Kurz, R. C., McQuaig, S. M., Lukasik, J., and Scott, T. M. (2007). Conﬁrmation of putative stormwater impact on water quality at a Florida beach by microbial source tracking methods and structure of indicator organism populations. Water research, 41(16):3747–57.

Carvalho, H., Cruz-Machado, V., and Tavares, J. G. (2012). A mapping framework for assessing Supply Chain resilience. International Journal of Logistics Systems and Management, 12(3):354– 373.

69 Bibliography

Chanan, A., Vigneswaran, S., and Kandasamy, J. (2010). Valuing stormwater, rainwater and wastewater in the soft path for water management: Australian case studies. Water science and technology, 62(12):2854–61.

Chang, Y.-M., Chou, C.-M., Su, K.-T., and Tseng, C.-H. (2005). Effectiveness of street sweeping and washing for controlling ambient TSP. Atmospheric Environment, (39):1891–1902.

Choe, J., Bang, K., and Lee, J. (2002). Characterization of surface runoff in urban areas. Water Science & Technology, 45(9):249–254.

Chow, M. F., Yusop, Z., and Mohamed, M. (2011). Quality and ﬁrst ﬂush analysis of stormwater runoff from a tropical commercial catchment. Water science and technology, 63(6):1211–6.

Commision of the European Communities (2007). Communication from the Commission to the Eu- ropean Parliament and the Council: First stage in the implementation of the Water Framework Directive 2000/60/EC. Technical report, Brussels.

Costa, D. F. and Ram´ısio, P. J. (2012). Estimac¸ao˜ de cargas poluentes de sistemas de drenage de aguas´ pluviais em ambiente urbano. In Proceedings of the 15th ENaSB conference, Evora,´ Portugal.

Dembel´ e,´ A., Bertrand-Krajewski, J.-L., and Barillon, B. (2010). Calibration of stormwater quality regression models: a random process? Water science and technology, 62(4):875–82.

Dembel´ e,´ A., Bertrand-Krajewski, J.-L., Becouze, C., and Barillon, B. (2011). A new empirical model for stormwater TSS event mean concentrations (EMCs). Water science and technology, 64(9):1926–34.

Desbordes, M., Deutsch, J.-c., and Hemain, J.-C. (1980). Urban runoff pollution in France: a national programme. In Proceedinds of th Helsinki Symposium, pages 29–36. IAHS.

Dotto, C. B. S., Kleidorfer, M., Deletic, A., Fletcher, T. D., McCarthy, D. T., and Rauch, W. (2010). Stormwater quality models: performance and sensitivity analysis. Water science and technology, 62(4):837–43.

Ellis, J. B. and Revitt, D. M. (2008). Deﬁning urban diffuse pollution loadings and receiving water hazard. Water Science & Technology, 57(11):1817–1823.

Ferreira, F. (2006). Modelac¸ao˜ e gestao˜ integrada de sistemas de aguas´ residuais. PhD thesis, Instituto Superior Tecnico. Universidade Tecnica de Lisboa.

Ferreira, F., Sousa, A., Morgado, M., and Saldanha Matos, J. (2004). Ilustrac¸ao˜ de soluc¸oes˜ nao˜ con- vencionais para o saneamento de aguas´ residuais. In Proceedings of the 11th ENaSB conference, Faro, Portugal.

70 Bibliography

Freni, G., Mannina, G., and Viviani, G. (2009). Stormwater inﬁltration trenches: a conceptual modelling approach. Water science and technology : a journal of the International Association on Water Pollution Research, 60(1):185–99.

Gawler, A. H., Beecher, J. E., Brandao,˜ J. a., Carroll, N. M., Falcao,˜ L., Gourmelon, M., Masterson, B., Nunes, B., Porter, J., Rince,´ A., Rodrigues, R., Thorp, M., Walters, J. M., and Meijer, W. G. (2007). Validation of host-speciﬁc Bacteriodales 16S rRNA genes as markers to determine the origin of faecal pollution in Atlantic Rim countries of the European Union. Water research, 41(16):3780–4.

Gnecco, I., Berretta, C., Lanza, L., and La Barbera, P. (2005). Storm water pollution in the urban environment of Genoa, Italy. Atmospheric Research, 77(1-4):60–73.

Gondim, R. B. (2008). Qualidade das aguas´ pluviais em meio urbano. PhD thesis, Instituto Superior Tecnico. Universidade Tecnica de Lisboa.

Gray, N. (2004). Biology and Wastewater Treatment. Imperial College Press, London, 2nd edition.

Gromaire, M. C., Garnaud, S., Ahyerre, M., and Chebbo, G. (2000). The quality of street cleaning waters: comparison with dry weather and wet weather ﬂows in a Parisian combined sewer system. Urban Water 2, pages 39–46.

Gromaire, M. C., Garnaud, S., Gonzalez, A., and Chebbo, G. (1999). Characterization of urban runoff pollution in Paris. Water Science & Technology, 39(2):1–8.

Gromaire, M. C., Garnaud, S., Saad, M., and Chebbo, G. (2001). Contribution of different sources to the pollution of wet weather ﬂows in combined sewers. Water research, 35(2):521–533.

Gromaire, M. C., Robert-Sainte, P., Bressy, A., Saad, M., De Gouvello, B., and Chebbo, G. (2011). Zn and Pb emissions from rooﬁng materials - modelling and mass balance attempt at the scale of a small urban catchment. Water Science & Technology, 63(11):2590.

Hathaway, J. M., Hunt, W. F., Graves, a. K., Bass, K. L., and Caldwell, A. (2011). Exploring fecal indicator bacteria in a constructed stormwater wetland. Water Science & Technology, 63(11):2707.

Hatt, B. E., Fletcher, T. D., and Deletic, A. (2009). Pollutant removal performance of ﬁeld-scale stormwater bioﬁltration systems. Water science and technology, 59(8):1567–76.

Hatt, B. E., Steinel, A., Deletic, A., and Fletcher, T. D. (2011). Retention of heavy metals by stormwater ﬁltration systems: breakthrough analysis. Water science and technology, 64(9):1913–9.

Huang, J.-l., Du, P.-f., Ao, C.-t., Lei, M.-h., Zhao, D.-q., Ho, M.-h., and Wang, Z.-s. (2007). Charac- terization of surface runoff from a subtropics urban catchment. Journal of environmental sciences (China), 19(2):148–52.

Jeng, H. A., Englande, A. J., Bakeer, R. M., and Bradford, H. B. (2005). Impact of urban stormwater runoff on estuarine environmental quality. Estuarine Coastal and Shelf Science, 63:513–526.

71 Bibliography

Jensen, M. B., Cederkvist, K., Bjerager, P. E. R., and Holm, P. E. (2011). Dual Porosity Filtration for treatment of stormwater runoff: ﬁrst proof of concept from Copenhagen pilot plant. Water science and technology, 64(7):1547–57.

Kang, J.-H. and Stenstrom, M. K. (2008). Evaluation of street sweeping effectiveness as a stormwater management practice using statistical power analysis. Water science and technology, 57(9):1309– 15.

Kanso, A., Gromaire, M. C., Gaume, E., Tassin, B., and Chebbo, G. (2003). Bayesian approach for the calibration of models: application to an urban stormwater pollution model. Water Science & Technology, 47(4):77–84.

Khastagir, a. and Jayasuriya, L. N. N. (2010). Impacts of using rainwater tanks on stormwater harvesting and runoff quality. Water science and technology : a journal of the International Association on Water Pollution Research, 62(2):324–9.

Kim, L.-H., Jeong, S.-M., and Ko, S.-O. (2007). Determination of ﬁrst ﬂush criteria using dynamic EMCs (event mean concentrations) on highway stormwater runoff. Water Science & Technology, 55(3):71.

Kortbaoui, R., Locas, A., Imbeau, M., Payment, P., and Villemur, R. (2009). Universal mitochondrial PCR combined with species-speciﬁc dot-blot assay as a source-tracking method of human, bovine, chicken, ovine, and porcine in fecal-contaminated surface water. Water research, 43(7):2002–10.

Lança, I., Silva-Afonso, A., and Pimentel-Rodrigues, C. (2012). Aproveitamento de aguas´ pluviais em edif´ıcios - estudo da segurança de um sistema de aproveitamento para fins de rega. In Proceedings of the 15th ENaSB conference, volume 234370094, Evora,´ Portugal.

Liebman, M. B., Jonasson, O. J., and Wiese, R. N. (2011). The urban stormwater farm. Water Science & Technology, 64(1):239.

Liu, Y., Che, W., and Li, J. (2005). Monitor-based evaluation of pollutant load from urban stormwater runoff in Beijing. Water Science & Technology, 52(9):191–197.

Luo, H., Luo, L., Huang, G., Liu, P., Li, J., Hu, S., Wang, F., Xu, R., and Huang, X. (2009). Total pollution effect of urban surface runoff. Journal of environmental sciences, (21):1186–1193.

Marsalek, J., Rochfort, Q., Grapentine, L., and Brownlee, B. (2002). Assessment of stormwater impacts on an urban stream with a detention pond. Water Science & Technology, 45(3):255–263.

Martellini, A., Payment, P., and Villemur, R. (2005). Use of eukaryotic mitochondrial DNA to differentiate human, bovine, porcine and ovine sources in fecally contaminated surface water. Water Research, 39(4):541–8.

McArdle, P., Gleeson, J., Hammond, T., Heslop, E., Holden, R., and Kuczera, G. (2011). Centralised urban stormwater harvesting for potable reuse. Water science and technology, 63(1):16–24.

72 Bibliography

McCarthy, D. T. (2009). A traditional ﬁrst ﬂush assessment of E. coli in urban stormwater runoff. Water Science & Technology2, 60(11).

McCarthy, D. T., Deletic, A., Mitchell, V. G., and Diaper, C. (2010). Sensitivity analysis of an urban stormwater microorganism model. Water science and technology, 62(6):1393–400.

McLeod, S., Kells, J., Asce, M., and Putz, G. J. (2006). Urban Runoff Characterization and Load Estimation in Saskatoon, Canada. Journal of Environmental Engineering, 132(11):1470–1481.

Mendez,´ H., Geary, P.M., and Dunstan, R. H. (2009). Surface wetlands for the treatment of pathogens in stormwater: three case studies at Lake Macquarie, NSW, Australia. Water science and technology, 60(5):1257–63.

Metcalf & Eddy, I. (2003). Wastewater Engineering: treatment and reuse. McGraw Hill, 4th edition.

Moran, J. (2012). Three Challenges for Environmental Philosophy. Philoshophy Now, (88).

Moura, P., Barraud, S., Baptista, M. B., and Malard, F. (2011). Multicriteria decision-aid method to evaluate the performance of stormwater inﬁltration systems over the time. Water science and technology : a journal of the International Association on Water Pollution Research, 64(10):1993– 2000.

Mourad, M., Bertrand-Krajewski, J.-L., and Chebbo, G. (2005). Sensitivity to experimental data of pollutant site mean concentration in stormwater runoff. Water Science & Technology, 51(2):155– 162.

Nazahiyah, R., Yusop, Z., and Abustan, I. (2007). Stormwater quality and pollution loading from an urban residential catchment in Johor, Malaysia. Water science and technology, 56(7):1–9.

Noble, R., Lee, I., and Schiff, K. (2004). Inactivation of indicator micro-organisms from various sources of faecal contamination in seawater and freshwater. Journal of Applied Microbiology, 96(3):464– 472.

Petersen, T. M., Rifai, H. S., Suarez, M. P., and Stein, A. R. (2005). Bacteria Loads from Point and Nonpoint Sources in an Urban Watershed. (October):1414–1426.

Prescott, L. M., Harley, J. P., and Klein, D. A. (1993). Microbiology. Wm. C. Brown Publishers, 2nd edition.

Quintela, A. C. (1996). Hidrologia e Recursos Hidricos. Instituto Superior Tecnico, Lisbon.

Roslev, P.and Bukh, A. S. (2011). State of the art molecular markers for fecal pollution source tracking in water. Applied microbiology and biotechnology, 89(5):1341–55.

Santo Domingo, J. W., Bambic, D. G., Edge, T. a., and Wuertz, S. (2007). Quo vadis source tracking? Towards a strategic framework for environmental monitoring of fecal pollution. Water research, 41(16):3539–52.

73 Bibliography

Savichtcheva, O. and Okabe, S. (2006). Alternative indicators of fecal pollution: relations with pathogens and conventional indicators, current methodologies for direct pathogen monitoring and future application perspectives. Water research, 40(13):2463–76.

Schill, W. B. and Mathes, M. V. (2008). Real-time PCR detection and quantiﬁcation of nine potential sources of fecal contamination by analysis of mitochondrial cytochrome b targets. Environmental science & technology, 42(14):5229–34.

Scott, T. M., Rose, J. B., Jenkins, T. M., and Farrah, S. R. (2002). Microbial Source Tracking : Current Methodology and Future Directions. Applied and Environmental Microbiology, 68(12):5796–5803.

Sidhu, J. P. S., Toze, S., Hodgers, L., Shackelton, M., Barry, K., Page, D., and Dillon, P. (2010). Pathogen inactivation during passage of stormwater through a constructed reedbed and aquifer transfer, storage and recovery. Water science and technology, 62(5):1190–7.

Simpson, J. M., Domingo, J. W. S., and Reasoner, D. J. (2002). Critical Review Microbial Source Tracking : State of the Science. Environmental Science & Technology, 36(24):5279–5288.

Soyeux, E., Blanchet, F., and Tisserand, B. (2007). Stormwater overﬂow impacts on the sanitary quality of bathing waters. Water science and technology : a journal of the International Association on Water Pollution Research, 56(11):43–50.

Stoeckel, D. M. and Harwood, V. J. (2007). Performance, design, and analysis in microbial source tracking studies. Applied and environmental microbiology, 73(8):2405–15.

Taebi, A. and Droste, R. L. (2004). Pollution loads in urban runoff and sanitary wastewater. The Science of the total environment, 327(1-3):175–84.

Tixier, G., Rochfort, Q., Grapentine, L., Marsalek, J., and Lafont, M. (2011). In search of effective bioassessment of urban stormwater pond sediments: enhancing the ’sediment quality triad’ approach with oligochaete metrics. Water science and technology : a journal of the International Association on Water Pollution Research, 64(7):1503–10.

Uchimura, K., Nakamura, E., and Fuijita, S. (1997). Characteristics of stormwater runoff and its control in Japan. Water Science & Technology, 36(8-9):141–147.

Ullmo, J. (1969). La Pensee´ Scientiﬁque Moderne. Flammarion, Paris.

USEPA (1999). Preliminary Data Summary of Urban Storm Water Best Management Practices. Tech- nical report, U.S. Environmental Protection Agency, Ofﬁce of Water.

Vezzaro, L., Eriksson, E., Ledin, a., and Mikkelsen, P. S. (2010). Dynamic stormwater treatment unit model for micropollutants (STUMP) based on substance inherent properties. Water science and technology : a journal of the International Association on Water Pollution Research, 62(3):622–9.

Zgheib, S., Moilleron, R., and Chebbo, G. (2011). Inﬂuence of the land use pattern on the concentrations and ﬂuxes of priority pollutants in urban stormwater. Water science and technology : a journal of the International Association on Water Pollution Research, 64(7):1450–8.

74 A Article submitted to ENASB

A-1 CARACTERIZAÇÃO DE QUALIDADE DE ESCORRÊNCIAS PLUVIAIS DA CIDADE DE LISBOA – ORIGEM DA POLUIÇÃO MICROBIOLÓGICA

Gaspar QUEIROZ 1; Filipa FERREIRA 2; Ricardo SANTOS 3; Sílvia MONTEIRO 3; José S. MATOS 4

RESUMO Em resposta à crescente preocupação com a poluição urbana e a qualidade dos meios receptores, e atendendo ao recente desenvolvimento de métodos de Microbial SourceTracking (MST), destinados a aferir a origem da contaminação fecal presente nos efluentes (humana e animal), este estudo incide sobre a caracterização da qualidade de escorrências pluviais em três bacias urbanas de Lisboa. O desenvolvimento do estudo tem por objectivo: (i) avaliar a qualidade das escorrências pluviais em bacias urbanas, com ênfase na contaminação fecal e no parâmetro CQO; (ii) desenvolver a técnica de marcadores de DNA mitocondrial para espécies típicas de meio urbano (humanos, cães e gatos); (iii) avaliar a origem da poluição fecal existente, através da utilização dos marcadores de DNA mitocondrial desenvolvidos especificamente para as espécies animais referidas. Foram recolhidas amostras de escorrências pluviais e analisados os seguintes parâmetros: CQO, Escherichia coli, Enterococos intestinais e DNA mitocondrial específico de diferentes organismos (utilizando a técnica de PCR), para determinação da origem da contaminação fecal. Concluiu-se que os níveis de CQO são elevados (concentrações médias de 102 a 439 mg/l) e que existe uma forte contaminação fecal (concentrações médias de E.Coli . entre 1.3x10 3 e 6.2x10 4 NMP/100 ml), sobretudo de origem humana e animal (canina e felina). Estudos futuros focar-se-ão na pesquisa de contaminação de outros animais citadinos (pombos, gaivotas e ratos) e na contribuição quantitativa de cada um dos animais poluidores em cada uma das amostras.

Palavras-chave : escorrências pluviais; marcadores de DNA mitocondrial; origem da contaminação fecal; poluição microbiológica.

1 Aluno do Mestrado em Engenharia Civil do Instituto Superior Técnico, [email protected] 2 Professora Auxiliar do Instituto Superior Técnico, Av. Rovisco Pais, 1049-001 Lisboa, [email protected] 3 Investigador, Instituto Superior Técnico, Laboratório de Análises, Instituto Superior Técnico, Av. Rovisco Pais, 1049-001 Lisboa, [email protected] e [email protected] 4 Professor Catedrático do Instituto Superior Técnico, Av. Rovisco Pais, 1049-001 Lisboa, [email protected]

A-2

1 INTRODUÇÃO Nas últimas décadas, os países desenvolvidos têm demonstrado uma preocupação crescente com a poluição urbana. Esta preocupação abrange essencialmente três áreas de monitorização, controle e intervenção: a qualidade do ar, a qualidade da água e a gestão e reciclagem dos resíduos sólidos. Os dois grandes factores que contribuem para a poluição da água em meio urbano são as águas residuais, por um lado, e as escorrências pluviais, por outro. Estas últimas são resultado directo da precipitação sobre a bacia urbana e arrastam os poluentes que se acumulam em tempo seco, transportando-os até à rede de águas residuais. Vários estudos têm revelado que a carga poluente é dependente de um conjunto de factores de três naturezas diferentes: uns prendem-se com as características geomorfológicas da bacia de drenagem (Butler e Memon, 1999, citado por Ferreira, 2006); outros com as condições climatéricas ou regime de precipitação produzido sobre a bacia (Gnecco et al. , 2005); e ainda outros com o tipo de ocupação da bacia (Gray, 2004). Em meio urbano, é comum as escorrências superficiais conterem matéria orgânica proveniente de resíduos vegetais, como folhas de árvores, e bactérias de origem fecal provenientes de diferentes dejectos animais. Estima-se que cerca de 17 g/(m 2.ano) de dejectos caninos sejam depositados nas zonas pavimentadas urbanas, e que a carga poluente produzida pelos cães de uma cidade como Manchester, no Reino Unido (que em 2001 tinha uma população de cerca de 2,5 milhões de habitantes), seja equivalente à produzida por 25 a 30 000 habitantes (Gray, 2004). Em Nova York, os cães depositam diariamente nas ruas da cidade 68 toneladas de dejectos e 405 000 litros de urina, grande parte dos quais acabam por ser arrastados durante o tempo de chuva (Feldman, 1974; citado por Gray, 2004). A preocupação com a qualidade dos meios aquáticos, quer no que respeita a problemas de saúde e custos de encerramento de zonas balneares e de aquacultura por fraca qualidade da água, quer no que respeita ao cumprimento das Directivas internacionais (nos países europeus - European Union Directive 2006/7/EC, COD/2002/0254), tem crescido de importância nos últimos anos. Actualmente, a qualidade das águas é baseada na presença de bactérias indicadoras de contaminação fecal ( Escherichia coli (EC) e Enterococos intestinais (EI)). No entanto, estas bactérias não permitem a identificação da fonte da contaminação existente nos aquíferos, facto que é de extrema relevância para perceber e debelar o problema de contaminação fecal logo desde a sua origem. O conceito de que a origem da contaminação pode ser rastreada utilizando métodos microbiológicos, genotipicos, fenotipicos e químicos é denominado de Microbial Source Tracking. Todos os métodos utilizados em Microbial Source Tracking baseiam-se na premissa de que diferentes organismos apresentam características diversas que permitem a sua distinção. Diversas metodologias utilizadas no diagnóstico das fontes de poluição têm como base a técnica de PCR ( Polymerase Chain Reaction ) que é baseada na amplificação de sequências específicas de DNA de um determinado organismo. Este tipo de métodos, que não estão dependentes de métodos culturais, é muito importante, dada a rapidez de obtenção de resultados (Domingo et al. 2007) e a sua elevada especificidade e sensibilidade. Os estudos mais recentes testam diversos marcadores genéticos, entre os quais marcadores mitocondriais específicos, com o objectivo de estabelecer a espécie geradora da contaminação fecal registada (e.g ., Ballesté et al., 2010). Estes marcadores têm sido aplicados em diversos países e a recursos hídricos no geral, nomeadamente: bacias

A-3 hidrográficas não urbanas, na Áustria (Reischer et al ., 2011) e Califórnia, EUA (Kildare et al. , 2007); em rios, no Canadá (Martellini et al ., 2005; Kortbaoui et al ., 2009) e em Espanha (Ballesté e Blanch, 2010); águas superficiais e costeiras, no Reino Unido (Baker-Austin et al., 2010). Baker-Austin et al. (2010) conseguiram identificar correctamente a origem da contaminação fecal em 85% (17/20) das amostras. Têm também sido aplicados em contextos mais urbanos, designadamente: para avaliar a presença de poluição fecal humana em linhas de água naturais, em tempo de chuva, em bacias servidas por fossas sépticas individuais, na Austrália (Ahmed et al ., 2008); e para demonstrar a presença de ligações indevidas em colectores pluviais que descarregam constantemente em tempo seco, na Califórnia, EUA (Sercu et al ., 2009). Os estudos citados referem o bom desempenho dos marcadores mitocondriais, bem como a crescente precisão e assertividade dos resultados. Ahmed et al. (2008) desenharam dois marcadores genéticos específicos para humanos (HF183 e HF134), em amostras de linhas de água naturais, em tempo de chuva, com elevados níveis de indicadores de contaminação fecal (EC, na ordem de 1,2 a 5,1 x10 6 NMP/100 ml, e EI, entre 4,5 a 5,6 x10 5NMP/100 ml), conseguindo detectar contaminação fecal humana em 40% dos testes para três diferentes locais e três eventos num dos marcadores e 70% dos testes com o outro. Contudo, este tema continua a ser pouco desenvolvido em Portugal, o que resulta num fraco espólio de informação sustentada. Neste sentido, o presente trabalho pretende colmatar esta carência de investigação e conhecimentos, servindo assim as entidades gestoras dos sistemas de drenagem urbanos. Através da análise de amostras de escorrências pluviais, pretende-se determinar as características e origem da poluição fecal depositada em meio urbano, utilizando marcadores de DNA mitocondrial que permitem a identificação das espécies que originaram a poluição captada. Assim, pretende-se identificar na poluição fecal das escorrências pluviais da zona em estudo, as seguintes espécies: humanos, cães e gatos.

2 METODOLOGIA 2.1 Selecção da bacia experimental e sua caracterização De forma a avaliar a qualidade das escorrências pluviais na cidade de Lisboa, foram seleccionadas três bacias experimentais que se consideram representativas dos diferentes tipos de ocupação e utilização do espaço urbano da cidade. Assim, foram recolhidas amostras nas sub-bacias de Alcântara, Rua da Madalena e Bairro das Ilhas, respectivamente localizadas nas bacias E, L e KJL (tal como ilustrado na Figura 1), servidas pelo sistema de drenagem afluente à ETAR de Alcântara. A escolha da bacia experimental de Alcântara deveu-se a dois principais factores: a existência de estudos anteriores sobre esta bacia (Ferreira, 2006; Gondim, 2008) e portanto de dados referentes a poluição em escorrências pluviais urbanas; bem como o facto da bacia reunir diversos tipos de ocupação de solo, de intensidade de tráfego e de características geomorfológicas, potenciando o interesse dos resultados obtidos. A bacia compreende-se entre o Instituto Superior de Agronomia (ISA) e a Avenida da Índia, no bairro de Alcântara. Apresenta diversas zonas de ocupação distintas: uma grande zona residencial, parte com moradias e pequenos jardins, parte com prédios com pátios interiores impermeáveis; zonas de pequeno comércio, com lojas, cafés e restaurantes; nas zonas de

A-4 cota mais baixa verifica-se um tráfego intenso, com muitas carreiras de autocarros e alguns elétricos. Embora, à excepção do ISA, a bacia seja quase totalmente impermeável, algumas ruas são bastante arborizadas e pavimentadas com calçada e uma parte dos telhados (um quarto a um terço) drena directamente para a rua. Os pontos de amostragem escolhidos, a que se referem a Figura 2 e o Quadro 1, pretendem mostrar esta diversidade e ao mesmo tempo assegurar uma correlação estreita com os obtidos por Ferreira (2006) e Gondim (2008).

A5 A4

Figura 2 – Localização dos pontos de amostragem na sub-bacia de Alcântara.

A-5

Quadro 1 – Localização dos pontos de amostragem, na sub-bacia de Alcântara. # Localização Características Junto do portão sudeste do Instituto Reduzida ocupação e impermeabilização; zona de cobertura A1 Superior de Agronomia. vegetal e árvores; declive médio. Na rua João de Barros, junto do Zona residencial de moradias, sem actividade comercial; ruas A2 cruzamento com a Rua Pedro Calmon. arborizadas; declive baixo. Na praça de Táxis da Rua Luís de Zona residencial de prédios, de tráfego intenso; café junto do A3 Camões. ponto de recolha; declive elevado. Na paragem de autocarro do Largo do Intensa actividade comercial e tráfego pedonal e rodoviário; A4 Calvário. declive baixo. Intensa actividade comercial e tráfego pedonal e rodoviário; A5 No lado norte Largo das Fontainhas. declive baixo. No largo no final da Rua da Cozinha Zona de estacionamento, com trafego elevado e declive A6 Económica praticamente nulo; entupimentos do sumidouro.

A bacia experimental do Bairro das Ilhas, na freguesia de São Jorge de Arroios, é significativamente menor que a anterior, sendo caracterizada pela sua ocupação residencial sem actividades comerciais significativas, arruamentos estreitos, geralmente de sentido único, e declividade não acentuada. A zona em estudo é um bairro delimitado pela rua Visconde de Santarém a Norte, pela calçada de Arroios a Este, pela rua Pascoal de Melo a Sul e pela rua Dona Estefânia a Oeste, como é possível ver na Figura 3. O tráfego é reduzido e tem zonas de acesso exclusivamente pedonal. É uma zona essencialmente sem vegetação, exceptuando o jardim Cesário Verde. No Quadro 2 apresenta-se a localização e características dos pontos de amostragem seleccionados no Bairro das Ilhas.

O terceiro conjunto de pontos de amostragem situa-se na Rua da Madalena, em plena Baixa Pombalina. Este arruamento apresenta uma forte actividade comercial e tráfego rodoviário intenso, com passagem regular de autocarros e, no fundo da rua, junto ao Martim Moniz, também de elétricos. O seu declive é acentuado, com ponto alto no Largo Adelino Amaro da Costa (Largo do Caldas). Na Figura 4 e no Quadro 2 apresenta-se a localização e características dos pontos de amostragem seleccionados na Rua da Madalena.

I1 I2

Figura 3 – Localização dos pontos de amostragem na sub-bacia do Bairro das Ilhas.

A-6

Quadro 2 – Localização dos pontos de amostragem na sub-bacia do Bairro das Ilhas. # Localização Características Na esquina Sudeste do jardim Cesário Zona com vegetação presente, e zonas permeáveis; I1 Verde. pavimento em calçada e pouco trafego; declive médio. R. Cidade da Horta, junto às escadas Zona pedonal larga (escadas) com drenagem directa das I2 da R. Ilha do Pico escorrências de coberturas; declive médio. Calçada de Arroios, no cruzamento com I3 Zona de tráfego superior às restantes; declive elevado. R. de Ponta Delgada Zona de estacionamento totalmente impermeável; I4 Largo no final da R. dos Açores. escorrências de coberturas; declive desprezável. Escadas que ligam a R. Ilha Terceira Zona pedonal estreita (escadas) de pouco tráfego pedonal; I5 com o jardim Cesário Verde. declive médio. Nas mesmas escadas que I5, mas Zona pedonal estreita (escadas) de pouco tráfego pedonal; I6 mais abaixo. declive médio.

Figura 4 – Localização dos pontos de amostragem na sub-bacia da Rua da Madalena.

Quadro 3 – Localização dos pontos de amostragem na Rua da Madalena. # Localização Características R. Madalena a subir, no sumidouro Zona impermeável de intenso tráfego e actividade comercial; M1 junto ao nº127. declive elevado. Zona impermeável comercial, de trafego intenso; restaurantes M2 R. Madalena, junto ao Largo do Caldas. próximos; declive baixo. R. Madalena, cruzamento com R. Zona impermeável de intenso tráfego e actividade comercial; M3 Condes de Monsanto. passagem de elétricos; declive elevado.

2.2 Descrição do trabalho experimental O trabalho de campo incidiu na recolha de amostras de escorrências pluviais, à entrada de sumidouros, originadas por eventos pluviométricos consecutivos em diversos pontos de amostragem situados na bacia de Alcântara (A), no Bairro das Ilhas (I) e na Rua da Madalena (M). As campanhas foram realizadas durante os meses de Abril a Julho de 2012. As escorrências foram captadas através de caixas interceptoras localizadas no interior dos sumidouros. Estes dispositivos, ilustrados na Figura 5, dispõem de uma pequena abertura, que permite a entrada das escorrências pluviais, apresentam reduzidas dimensões e uma capacidade de quatro litros.

A-7

Figura 5 – Exemplo de montagem do dispositivo de recolha de escorrências pluviais (à esquerda) e deposição de sedimentos em tempo seco (à direita).

De uma campanha para a seguinte, é necessária uma intervenção de limpeza e verificação das condições do material, pois fica depositada nos sumidouros uma quantidade elevada de partículas sólidas e de detritos (tal como se visualiza na Figura 5), procedendo-se então a uma lavagem da caixa com água. Este procedimento deve ser feito com luvas, de modo a não contaminar a caixa com DNA humano, devendo ser assegurado anteriormente à ocorrência de um evento de precipitação. Ao ocorrer um evento pluviométrico, a equipa de trabalho desloca-se para a zona em estudo de forma a recolher as amostras (são recolhidos dois frascos de 1 litro de capacidade por ponto de amostragem) e fazê-las chegar ao laboratório rapidamente. Procede-se então à determinação dos seguintes parâmetros: CQO (através da utilização de kits), EC e EI (através dos métodos Colilert e Enterolert, respectivamente, com procedimentos tais como descritos em Rivera e Rock (2011)) e DNA mitocondrial especifico de diferentes organismos (tal como descrito no subcapítulo 2.3), para determinação da origem da contaminação fecal presente.

2.3 Técnica de marcadores de DNA mitocondrial para espécies típicas de meio urbano As amostras foram concentradas por centrifugação e o DNA foi extraído através do kit QIAamp DNA Mini Kit (Qiagen). A partir do DNA extraído foram efectuados ensaios de PCR para a determinação da presença de DNA proveniente de humanos, cães, gatos, ratos, pombos e gaivotas. Os resultados foram visualizados num gel de agarose (2.5%) após coloração com brometo de etídeo.

3 ANÁLISE E APRESENTAÇÃO DE RESULTADOS Os eventos pluviométricos ocorridos durante o período das campanhas experimentais, de Março a Maio de 2012, foram analisados com base na informação udográfica proveniente da estação climatológica do Instituto Geofísico Infante Dom Luiz (IGIDL), em Lisboa. Na Figura 6 apresenta-se graficamente a precipitação horária registada (incluindo a realização de campanhas experimentais). Verifica-se que os picos de precipitação mais significativos ocorreram no início e meio de Maio, atingindo intensidades médias horárias de cerca de3 e 7 mm/h, respectivamente

A-8

Precipitação instantânea em Lisboa 8 7 6 5 4 3 2 Precipitação Precipitação (mm) 1 0 01-03-2012 16-03-2012 31-03-2012 15-04-2012 30-04-2012 15-05-2012 30-05-2012

Campanhas fins de semana Precipitação Instantânea

Figura 6 – Precipitação horária registada durante o período de realização das campanhas experimentais (de Março a Junho).

Para precipitações manifestamente reduzidas – tipicamente inferiores que 0.5 mm/h – não foram realizadas campanhas de amostragem uma vez que tais precipitações não originavam escorrências em quantidade suficiente para recolha e/ou análise laboratorial. Os resultados de qualidade das escorrências pluviais, no que se refere aos parâmetros físico-químicos e microbiológicos, foram submetidos a um tratamento estatístico com vista à análise da sua variabilidade. No Quadro 4 apresenta-se a síntese dos principais parâmetros estatísticos descritivos relativos à qualidade das escorrências pluviais dos locais monitorizados. A representação gráfica dos extremos, quartis e mediana, que permite aferir relativamente à variabilidade e simetria da amostra, é ilustrada pela Figura 7 (para EC e EI optou-se por uma escala logarítmica).

Quadro 4 – Síntese dos principais parâmetros estatísticos descritivos relativos à qualidade das escorrências pluviais das bacias experimentais. Alcântara Ilhas Madalena CQO E. coli Enterococos CQO E. coli Enterococos CQO E. coli Enterococos (mg/l) (NMP/100 ml) (NMP/100 ml) (mg/l) (NMP/100 ml) (NMP/100 ml) (mg/l) (NMP/100 ml) (NMP/100 ml) Conc. médias 439 5.1E+04 6.2E+04 102 6.0E+03 1.3E+04 341 2.6E+04 4.7E+04 Desvio padrão 348 6.5E+04 1.1E+05 76 1.4E+04 1.7E+04 291 6.9E+04 5.7E+04 Assimetria 0.51 1.85 3.56 2.35 3.29 1.71 1.00 3.26 1.17 máximo 1 091 2.4E+05 4.8E+05 346 4.8E+04 4.8E+04 937 2.4E+05 1.6E+05 3º Quartil 728 6.4E+04 5.6E+04 117 4.4E+03 1.5E+04 502 4.5E+03 7.2E+04 mediana 358 2.4E+04 2.9E+04 70 1.4E+03 6.4E+03 240 2.1E+03 2.4E+04 1º Quartil 138 1.1E+04 1.2E+04 55 7.2E+02 2.4E+03 111 9.9E+02 3.3E+03 Mínimo 33 1.0E+02 8.0E+01 45 1.3E+02 9.3E+02 76 1.7E+02 1.5E+03 Amplitude 1 058 2.4E+05 4.8E+05 301 4.8E+04 4.7E+04 861 2.4E+05 1.5E+05 Nº dados 19 18 18 17 12 12 14 12 12

A-9

feitoCQO no excel:(mg/l) E. coli (NMP/100 ml) Enterococos (NMP/100 ml) 1200 1.E+06 1.E+06

1000 1.E+05 1.E+05

800 1.E+04 1.E+04

1.E+03 600 1.E+03

400 1.E+02 1.E+02

200 1.E+01 1.E+01

0 1.E+00 1.E+00 Alcântara Ilhas Madalena Alcântara Ilhas Madalena Alcântara Ilhas Madalena

Figura 7 – Gráficos tipo box-and-whiskers representativos da distribuição dos valores obtidos de CQO, E. coli e Enterococos, para as três bacias experimentais.

Verifica-se que a concentração média em CQO na sub-bacia de Alcântara é de 439 mg/l, apresentando valores de 102 e 341 mg/l para as sub-bacias do Bairro das Ilhas e da Rua da Madalena. No que se reporta aos parâmetros microbiológicos, nomeadamente a EC, as concentrações médias variam entre 1.3 x10 3 e 6.2 x 10 4 NMP/100 ml, nas três sub-bacias. O parâmetro EI apresenta valores similares, entre 1.3x10 3 e 6.2 x 10 4, nas sub-bacias em estudo. Os desvios padrão são, tipicamente, da ordem de grandeza das concentrações médias registadas. Tal como seria de esperar atendendo aos usos e características das três sub-bacias em análise, as escorrências pluviais do Bairro das Ilhas apresentam, globalmente, os menores níveis de poluição. As escorrências pluviais da Rua da Madalena assumem valores intermédios, embora mais parecidos com os registados em Alcântara. Os elevados valores médios referentes às concentrações em CQO, EC e EI merecem relevo, uma vez que evidenciam o significativo impacto no meio receptor, em resultado da descarga de escorrências pluviais. À excepção da maioria dos valores de CQO obtidos no Bairro das Ilhas, as concentrações excedem, claramente, o valor limite de emissão definido pelo Decreto-Lei nº 236/98, de 150 mg/l para CQO. De acordo com a Directiva 2006/7/CE, os valores dos parâmetros microbiológicos correspondentes à classe de qualidade boa para águas costeiras e de transição (de 200 e 500 NMP/100 ml, respectivamente para EI e EC), são sistematicamente excedidos pelas amostras analisadas.

Nos Quadros 5 a 7 apresentam-se, para cada ponto de amostragem das três sub-bacias consideradas, as concentrações obtidas para os parâmetros microbiológicos e os resultados referentes aos marcadores PCR para humanos, gatos e cães. Os resultados apresentados nos quadros representam as amostras com resultados positivos (+), negativos (-) e também os que por alguma razão (falta de amostra, contaminação exógena, entre outras) não foi possível obter resultados (não avaliado – n.a.). Alguns dos pontos de amostragem inicialmente considerados vieram a demonstrar apresentar problemas pelas seguintes razões:

A-10

• A4 – frequentemente continha sopa e leite; • I1 – por vezes não possibilita uma recolha eficaz, por deficiência do sistema instalado; • I2 – algumas vezes inundado e entupido; • I3 – difícil acesso, viaturas estacionadas.

Quadro 5 – Resultados dos marcadores PCR para os pontos de amostragem da sub-bacia de Alcântara. E. coli Enterococos PCR Data(NMP/100mL) Humano Cão Gato 1ª Amostragem n.a. n.a. + - - 25-10-2011 n.a. n.a. - - - 15-03-2011 n.a. n.a. - - - A1 30-03-2012 (Manhã) n.a. n.a. n.a. + + 30-03-2012 (Tarde) n.a. n.a. n.a. - - 03-05-2012 9,77E+04 4,84E+05 + + - 19-04-2012 1,00E+02 1,20E+03 - - - A2 26-04-2012 n.a. n.a. - + - 15-03-2012 n.a. n.a. + - - 16-03-2012 n.a. n.a. n.a. - - 24-03-2012 n.a. n.a. n.a. - - 30-03-2012 (Manhã) n.a. n.a. n.a. + + 30-03-2012 (Tarde) n.a. n.a. n.a. - - A3 15-04-2012 1,09E+03 8,00E+01 n.a. - - 19-04-2012 1,41E+05 8,66E+04 - - - 26-04-2012 2,42E+05 6,49E+04 - + - 03-05-2012 3,13E+04 1,98E+04 - + - 07-05-2012 6,90E+04 1,45E+05 - + + 25-10-2011 n.a. n.a. + - - 15-03-2012 n.a. n.a. + - - 15-03-2012 n.a. n.a. n.a. - - 16-03-2012 n.a. n.a. n.a. - + 30-03-2012 n.a. n.a. n.a. - - A4 15-04-2012 4,84E+04 4,84E+04 n.a. - - 19-04-2012 1,72E+04 2,61E+04 + - - 26-04-2012 2,01E+04 3,08E+04 - - - 03-05-2012 3,72E+04 6,10E+03 + + - 07-05-2012 2,82E+04 1,92E+04 - + + 1ª amostragem n.a. n.a. + + + 25-10-2011 n.a. n.a. + - - 15-03-2012 n.a. n.a. + - - 03-02-2012 n.a. n.a. n.a. - - 16-03-2012 n.a. n.a. n.a. - - 18-03-2012 n.a. n.a. n.a. - + A5 30-03-2012 n.a. n.a. n.a. + - 30-03-2012 n.a. n.a. n.a. - + 15-04-2012 1,24E+03 1,16E+03 n.a. - - 26-04-2012 1,66E+04 5,79E+04 - + - 03-05-2012 4,02E+03 2,75E+04 - + + 07-05-2012 1,72E+04 5,19E+04 + + +

A-11

Quadro 5 – Resultados dos marcadores PCR para os pontos de amostragem da sub-bacia de Alcântara. (cont.) E. coli Enterococos PCR Data(NMP/100mL) Humano Cão Gato 1ª amostragem n.a. n.a. + - + 25-10-2012 n.a. n.a. + - - 15-03-2012 n.a. n.a. + - + 16-03-2012 n.a. n.a. n.a. - + A6 30-03-2012 n.a. n.a. n.a. + + 30-03-2012 n.a. n.a. n.a. - - 19-04-2012 n.a. n.a. - - - 26-04-2012 1,41E+05 4,35E+04 + + - 03-05-2012 8,82E+03 9,28E+03 - + +

Quadro 6 – Resultados dos marcadores PCR para os pontos de amostragem da sub-bacia do Bairro das Ilhas. E. coli Enterococos PCR Data(NMP/100mL) Humano Cão Gato 30-03-2012 (Manhã) n.a. n.a. n.a. + + I1 30-03-2012 (Tarde) n.a. n.a. n.a. - - 30-03-2012 (Manhã) n.a. n.a. n.a. + - 30-03-2012 (Tarde) n.a. n.a. n.a. - - I2 15-04-2012 4,84E+04 4,84E+04 n.a. - - 19-04-2012 n.a. n.a. - - - 30-03-2012 (Manhã) n.a. n.a. n.a. + - I3 30-03-2012 (Tarde) n.a. n.a. n.a. + - 15-04-2012 7,22E+03 1,45E+04 n.a. - + 30-03-2012 (Manhã) n.a. n.a. n.a. - - 30-03-2012 (Tarde) n.a. n.a. n.a. - - 15-04-2012 1,26E+02 9,30E+02 n.a. - - I4 19-04-2012 6,49E+04 2,42E+05 - - - 26-04-2012 n.a. n.a. - + + 03-05-2012 8,20E+02 2,52E+03 + + + 07-05-2012 1,24E+03 1,92E+03 - + + 30-03-2012 (Manhã) n.a. n.a. n.a. - - 30-03-2012 (Tarde) n.a. n.a. n.a. - - 15-04-2012 3,98E+03 4,84E+04 n.a. - + I5 19-04-2012 n.a. n.a. - - - 26-04-2012 1,60E+03 1,46E+03 - + + 03-05-2012 4,00E+02 1,55E+04 - + + 07-05-2012 4,00E+02 7,98E+03 + + + 30-03-2012 n.a. n.a. n.a. - - 15-04-2012 1,69E+03 9,77E+03 n.a. - - I6 03-05-2012 5,84E+03 3,42E+03 + + + 07-05-2012 8,20E+02 4,76E+03 + + +

A-12

Quadro 7 – Resultados dos marcadores PCR para os pontos de amostragem da sub-bacia da Rua da Madalena. E. coli Enterococos PCR Data(NMP/100mL) Humano Cão Gato 16-03-2012 n.a. n.a. n.a. - + 24-03-2012 n.a. n.a. n.a. - - 30-03-2012 (Manhã) n.a. n.a. n.a. - - 30-03-2012 (Tarde) n.a. n.a. n.a. + + M1 15-04-2012 1,70E+02 8,70E+03 n.a. + + 19-04-2012 2,26E+03 3,87E+04 - - - 26-04-2012 2,01E+03 1,45E+03 - + - 03-05-2012 3,50E+03 1,54E+05 + + + 07-05-2012 1,48E+03 6,90E+04 + + + 16-03-2012 n.a. n.a. n.a. - - 18-03-2012 n.a. n.a. n.a. - + 30-03-2012 (Manhã) n.a. n.a. n.a. - - 30-03-2012 (Tarde) n.a. n.a. n.a. - - M2 15-04-2012 4,84E+04 4,84E+04 n.a. - - 19-04-2012 n.a. n.a. - - - 26-04-2012 2,31E+03 3,82E+03 - + - 03-05-2012 7,64E+03 5,24E+03 + + + 07-05-2012 1,24E+03 7,94E+04 - + + 16-03-2012 n.a. n.a. n.a. - - 24-03-2012 n.a. n.a. n.a. + + 30-03-2012 (Manhã) n.a. n.a. n.a. - - M3 30-03-2012 (Tarde) n.a. n.a. n.a. - - 15-04-2012 4,84E+04 4,84E+04 n.a. - - 19-04-2012 2,42E+05 1,55E+05 - - - Com base na informação apresentada nos Quadros 5 a 7, evidencia-se uma contaminação fecal significativa explicitada nos elevados valores encontrados para as duas bactérias comumente utilizadas como indicadores de contaminação fecal (EC e EI). A análise, efectuada por PCR, permite aferir quais os diferentes agentes responsáveis por esta contaminação. Até à data, foram avaliados marcadores para humanos, cães e gatos e, em duas amostras (não incluídas nos quadros anteriores), para pombos. Os resultados indicam uma contribuição bastante relevante dos animais estudados, com 57% de amostras positivas para pelo menos um dos animais e 9% das amostras positivas para os três animais (humanos, cães e gatos). As duas amostras analisadas para os marcadores de pombo (recolhidas nos pontos de amostragem A3 e A4 a 7 de Maio) foram positivas. Atendendo aos resultados actualmente apurados, existe uma forte componente fecal de origem humana na sub-bacia de Alcântara, que se pode dever ao facto desta abranger zonas residenciais, com intenso tráfego pedonal e vida nocturna relevante. É importante também referir que, no Bairro das Ilhas, a percentagem de resultados positivos para gatos e cães é superior quando comparada com os outros locais de ensaio. Isto é justificado por ser uma zona fundamentalmente residencial, o que implica a existência de um maior número de

A-13 animais de companhia, aumentando, consequentemente, a sua influência para os resultados positivos. A Rua da Madalena apresenta uma contaminação fecal de origem humana superior à animal, possivelmente por ter uma intensa actividade comercial, embora a diferença seja mais ténue que no caso de Alcântara.

4 PRINCIPAIS CONCLUSÕES O presente estudo foi desenvolvido com os objectivos principais de avaliar a qualidade das escorrências pluviais em meio urbano (concentrações médias de CQO, EC e EI para diversos eventos pluviométricos) e, adicionalmente, avaliar a origem da carga microbiológica poluente, identificando a proveniência dos dejectos animais (espécie animal, nomeadamente humanos, cães e gatos), através da técnica de marcadores de DNA mitocondrial. A análise dos marcadores mitocondriais constitui uma técnica inovadora na pesquisa da origem da poluição fecal em meio urbano, com grande potencial de aplicação na redução da poluição depositada em tempo seco, especialmente com o alargamento da detecção da origem a outras espécies comuns ao meio urbano actual, nomeadamente ratazanas, pombos ou gaivotas. Embora não se encontrem utilizando a mesma aplicação destas técnicas na literatura, estudos anteriores utilizando este tipo de abordagem (DNA mitocondrial) já foram efectuados em diferentes tipos de amostra, tendo como resultados uma alta especificidade e sensibilidade (Ballesté et al., 2010). Concluiu-se que os níveis de CQO são elevados (concentrações médias de 102 a 439 mg/l) e que existe uma forte contaminação fecal (concentrações médias de EC entre 1.3x10 3 e 6.2x10 4 NMP/100 ml; concentrações médias de EI entre 1.3x10 3 e 6.2 x 10 4), sobretudo de origem humana e animal. De facto, é possível detectar a presença de contaminação fecal de origem humana, canina ou felina em cada ponto de amostragem da bacia de Alcântara, do Bairro das Ilhas e da Rua da Madalena. Estudos futuros focar-se-ão na pesquisa de contaminação de outros animais citadinos (pombos, gaivotas e ratos) e na contribuição quantitativa de cada um dos animais poluidores em cada uma das amostras.

AGRADECIMENTOS Os autores agradecem ao Instituto Geofísico Dom Luiz pela cedência de dados de precipitação, sem os quais não teria sido possível efectuar parte do tratamento de dados apresentado.

REFERÊNCIAS BIBLIOGRÁFICAS Ahmed, W., Stewart, J., Powell, D., & Gardner, T. (2008).Evaluation of Bacteroides markers for the detection of human faecal pollution. Letters in Applied Microbiology , 46 (2), 237-42. Baker-Austin, C., Rangdale, R., Lowther, J., & Lees, D. N. (2010). Application of mitochondrial DNA analysis for microbial source tracking purposes in shellfish harvesting

A-14 waters. Water Science and Technology: a journal of the International Association on Water Pollution Research , 61 (1), 1-7. Ballesté, E.,Blanch, A. R. (2010). Persistence of Bacteroides species populations in a river as measured by molecular and culture techniques. Applied and environmental microbiology , 76 (22), 7608-16. Ballesté, E., Bonjoch, X., Belanche, L. a, & Blanch, A. R. (2010). Molecular indicators used in the development of predictive models for microbial source tracking. Applied and environmental microbiology , 76 (6), 1789-95. Domingo, J.W.S., Bambic, D.G., Edge, T.A., Wuertz, S.(2007). Quo vadis source tracking? Towards a strategic framework for environmental monitoring of fecal pollution.Water Research 41, 3539-3552. Ferreira, F.(2006). Modelação e gestão integrada de sistemas de águas residuais. Dissertação de Doutoramento. Instituto Superior Técnico. Universidade Técnica de Lisboa. Lisboa. Gondim, R. B.(2008). Qualidade das águas pluviais em meio urbano . Instituto Superior Tecnico. Universidade Tecnica de Lisboa. Gnecco, I., Beretta, C., Lanza, L.G., La Barbera, P.(2005). Storm water pollution in the urban environment of Genoa, Italy.Atmospheric Research 77, 60-73. Gray,N.(2004). Biology and Wastewater Treatment.2 nd edition.Imperial College Press. London. Kildare, B. J., Leutenegger, C. M., McSwain, B. S., Bambic, D. G., Rajal, V. B., &Wuertz, S. (2007). 16S rRNA-based assays for quantitative detection of universal, human-, cow-, and dog-specific fecal Bacteroidales: a Bayesian approach. Water research , 41 (16), 3701-15. Kortbaoui, R., Locas, A., Imbeau, M., Payment, P., &Villemur, R. (2009).Universal mitochondrial PCR combined with species-specific dot-blot assay as a source-tracking method of human, bovine, chicken, ovine, and porcine in fecal-contaminated surface water. Water Research , 43 (7), 2002-10. Elsevier Ltd. Martellini, A., Payment, P., &Villemur, R. (2005). Use of eukaryotic mitochondrial DNA to differentiate human, bovine, porcine and ovine sources in fecally contaminated surface water. Water Research , 39 (4), 541-8. Reischer, G. H., Kollanur, D., Vierheilig, J., Wehrspaun, C., Mach, R. L., Sommer, R., Stadler, H., e Farnleiter, A. H. (2011). Hypothesis-driven approach for the identification of fecal pollution sources in water resources. Environmental science & technology , 45 (9), 4038- 45. Rivera, B., Rock, C. (2011). Microbial Source Tracking : Watershed Characterization and Source Identification. Tucson, Arizona. Sercu, B., van de Werfhorst, L., Murray, J., & Holden, P. (2009). Storm Drains are Sources of Human Fecal Pollution during Dry Weather in Three Urban Southern California Watersheds. Environmental Science & Technology , (43), 293-298.

A-15