RESEARCH ARTICLE

Heavy metal contamination in an urban stream fedby contaminated air-conditioning and stormwater discharges

Aisling O’Sullivan & Daniel Wicke & Tom Cochrane

Received: 21 April 2011 /Accepted: 1 October 2011 /Published online: 18 October 2011# Springer-Verlag 2011

AbstractPurpose Urban waterways are impacted by diffuse storm-water runoff, yet other discharges can unintentionallycontaminate them. The Okeover stream in Christchurch,New Zealand, receives air-conditioning discharge, while itsephemeral reach relies on untreated stormwater flow.Despite rehabilitation efforts, the ecosystem is still highlydisturbed. It was assumed that stormwater was the solecontamination source to the stream although water qualitydata were sparse. We therefore investigated its water andsediment quality and compared the data with appropriateecotoxicological thresholds from all water sources.Methods Concentrations of metals (Zn, Cu and Pb) in streambaseflow, stormwater runoff, air-conditioning discharge andstream-bed sediments were quantified along with flowregimes to ascertain annual contaminant loads. Metals wereanalysed by ICP-MS following accredited techniques.Results Zn, Cu and Pb concentrations from stormflowexceeded relevant guidelines for the protection of 90% ofaquatic species by 18-, 9- and 5-fold, respectively, suggest-ing substantial ecotoxicity potential. Sporadic copper (Cu)inputs from roof runoff exceeded these levels up to 3,200-fold at >4,000 μg L−1 while Cu in baseflow from air-conditioning inputs exceeded them 5.4-fold. There was an11-fold greater annual Cu load to the stream from air-conditioning discharge compared to stormwater runoff.Most Zn and Cu were dissolved species possibly enhancingmetal bioavailability. Elevated metal concentrations werealso found throughout the stream sediments.

Conclusions Environmental investigations revealed unsus-pected contamination from air-conditioning discharge thatcontributed greater Cu annual loads to an urban streamcompared to stormwater inputs. This discovery helpedreassess treatment strategies for regaining ecological integrityin the ecosystem.

Keywords Air-conditioning . Stormwater flow . Heavymetal contamination

1 Introduction

Waterways provide multi-functional assets to urban areas,including drainage (e.g., Shaw and Bible 1996; Tsihrintzis etal. 1995). In New Zealand, waterways are recognised asappreciating assets rather than simply drainage channels(Nolan 2005), while waterway rehabilitation is a priority formany urban residents. Local government guidelines inChristchurch, New Zealand (population, 369,000), affordrecognition and protection for six integrated waterway values(namely: culture, heritage, ecology, recreation, landscape anddrainage). These are considered in all new and retrofittedurban developments (CCC 2003). Significant efforts havebeen made to enhance urban waterways across New Zealand,and the Okeover stream in Christchurch serves as anationwide template for aesthetic enhancement retrofittedinto older suburbs. The stream was originally spring-fed butbecame disconnected from groundwater following urbandevelopments in the Canterbury region (Winterbourn et al.2007). The main flow for the stream’s perennial reach is nowdischarged (deep aquifer-derived) water that has been used toregulate climatic conditions in neighbouring universitybuildings, while the ephemeral reach is solely fed byuntreated stormwater. Although rehabilitated sections of the

Responsible editor: Vera Slaveykova

A. O’Sullivan (*) :D. Wicke : T. CochraneUniversity of Canterbury,Christchurch, Canterbury, New Zealande-mail: aisling.osullivan@canterbury.ac.nz

Environ Sci Pollut Res (2012) 19:903–911DOI 10.1007/s11356-011-0639-5

perennial reach are aesthetically appealing, ecological recov-ery has not been very successful (Blakely and Harding 2005;Winterbourn et al. 2007), and information on this streamwater and sediment quality is sparse.

It is widely reported that diffuse pollutants, especiallyheavy metals and sediments, draining from urban impervioussurfaces during storm events are efficiently transported tonearby waterways (Beasley and Kneale 2002; Egodawatta etal. 2009; Wicke et al. 2009). Consequently, the ecologicalhealth and integrity of receiving waterways are impaired(Waara and Färm 2008), while aesthetics and recreationaluses can be compromised. Principal stormwater metalcontaminants in New Zealand and elsewhere are zinc (Zn),copper (Cu) and lead (Pb), in particulate and dissolved forms(e.g., Brown and Peake 2006; Gobel et al. 2007; Zanders2005). Copper originates from brake linings and copper-panelled roofs (Karlen et al. 2002; Pennington and Webster-Brown 2008; Davie et al. 2001), while vehicle tyres andolder galvanised steel roofs are a prime source of zinc (Davieet al. 2001; Shedden et al. 2007; Zanders 2005). Simulta-neously, other contaminating water discharges can impact onurban waterways. Although it is reported that deterioratingpiping enhances copper concentrations in closed reticulatedwater systems (Lagos et al. 2000; Critchley et al. 2002;Royuela and Otero 2003; Merkel and Pehkonen 2006), openwaterway contamination by air-conditioning discharges isnot reported. When this water is discharged to neighbouringwaterways, metal contamination and hence ecotoxicity occur.

We measured water quality in the Okeover stream duringbaseflow originating from three air-conditioning dischargepoints, from a copper (alloy) roof and in stormflow runoffto discern metal contamination from these different dis-charges. The stream sediment was also sampled to assesshistorical accumulation of heavy metals. Understanding thewater and sediment signatures, along with estimated annualcontaminant loadings, can help to prioritise treatmentstrategies and to establish recovery options for regainingecological integrity in urban streams.

2 Methodology

2.1 Flow measurements

The Okeover stream follows a 1.5-km flat (0.4% slope)topographic gradient through the University of Canterbury,eventually merging with the Avon River. Cross sections of theupper stormwater-fed ephemeral reach (620m) were surveyedtwice in May 2006, and the downstream air-conditioning-fedperennial reach (970 m) was surveyed in December 2006 andDecember 2009. A simple (97.5° angle) sharp-crested v-notchweir was installed in the ephemeral reach channel bed 4.3 mdownstream from the primary (and uppermost) stormwater

culvert. Water levels were recorded directly off the weirlabelled with incremental depth levels and converted to flowmeasurements through a standard equation for this weir type.Flow was verified in subsequent rain events by measuringvelocity, rating it to the wetted cross section and comparing itto the depth of water flowing over the weir. In the perennialreach, flow gauging was conducted at five transect pointsduring baseflow and at the uppermost transect point during aDecember 2006 storm event. Stream velocity was measuredincrementally with a Global Water FP 101 Flow Probe at eachtransect point and converted to flow rates using the velocityintegrated method.

2.2 Water and sediment sampling

Water and sediment samples from the stream were manuallysampled following applicable sampling protocols given in theAustralian and New Zealand Environment and ConservationCouncil guidelines (ANZECC 2000). In compliance withthese guidelines, at least 10% of the samples were duplicatedfor Quality Assurance/Quality Control (QA/QC) purposes.Water samples were collected head-space free in high-density polyethylene (HDPE) sampling bottles. Cation–anionbalances were completed for baseflow water samples toensure accuracy of analytical measurements (within 5%differences). Blanks and spikes conducted by the laboratoriesensured instrument precision and always met the laboratoryQA/QC criteria of <10% difference while duplicate sampleshave <7% differences complying with a robust sampling,handling and instrument check process. Roof runoff wassampled in December 2009 from the downpipe outlet usingan automatic sampler (ISCO 6712C compact portablesampler, Teledyne Isco, USA). Total metal samples werepreserved with concentrated (69%) nitric acid (Fisher, traceanalysis grade) to reduce the pH to less than 2.0 (APHA2005). Dissolved metal samples were pre-filtered throughdisposable Waterra 0.45-μm filters before HNO3 acidifica-tion. Total suspended solids (TSS) samples were collected inseparate 1,000-mL HDPE containers. Individual compositesediment samples, each a mixture of three separate samples,were collected at five points along both transects using aplastic trowel that was washed between sampling points withdistilled water. Sediment was taken from the upper (5–10 cm) stream bed and transferred into appropriate samplingbags. Samples were stored and chilled at 2–5°C until beinganalysed within 2 days.

2.3 Chemical analyses

Total and dissolved metal concentrations were determined in2006 by R.J. Hill Laboratories (Hills Labs), an InternationalAccreditation New Zealand certified laboratory, but weremeasured at the University of Canterbury in 2009 (following

904 Environ Sci Pollut Res (2012) 19:903–911

procurement of the certified instrumentation). All metals (Zn,Cu and Pb) were analysed by inductively coupled plasmamass spectroscopy (ICP-MS) (Agilent) following Method3125B (APHA 2005). In 2009, total metal samples fordigestion were mixed thoroughly on a magnetic stir plate,while 25 mL of sample was transferred to a 50-mLpolypropylene centrifuge tube. After the addition of 5 mLconcentrated HNO3, tubes were placed in a heating block,and samples were boiled for 1 h. Cooled samples were thenfiltered through an encapsulated 0.45-μm PVDF filter(47 mm, Labserv) directly into the analysis tube andanalysed via ICP-MS. TSS were measured within 24 hfollowing Method 2540D (APHA 2005). Water quality wasmeasured for pH (YSI Model 60 pH field meter), dissolvedoxygen (DO), temperature (YSI 550A DO instrument) andturbidity (Hach Model 2100P portable turbidimeter) in fivereplicates at the time of sampling. Quality assuranceprotocols, including blanks, duplicates (10% of samples),spiked samples (5% of samples) and instrument calibrationwere conducted throughout each batch analysis.

2.4 Annual metal loads

The average annual metal load for stormflow was estimatedfrom three different sampling events in 2006 only (May 11and 20, and Dec. 20 as per Fig. 1 as no stormwater monitoringwas conducted in 2009). Annual metal loading (Ls, inkilograms per year) was calculated as a function of the eventmean concentration (Kp, in milligrams per cubic metre),annual precipitation (P, of 626 mm year−1) corrected forstorms that produce no runoff (8, of 0.85), catchment runoffcoefficient (C, of 0.72 in the upper ephemeral catchment)and catchment area (A, of 45 ha), as outlined in thenationally adopted design guidelines appropriate for Christ-church (CCC 2003) (Eq. 1). The annual load for all threeevents was initially calculated independently using Eq. 1,each calculation employing its own event mean concentra-tion (EMC) value. Each of the respective EMC values wascalculated from flow-weighted samples during the threesampling periods. The May 11 sampling event had 14samples, while the May 20 and Dec 20 events each had sixsamples to generate their EMC values.

Ls ¼ Kp � P � 8 � C � Að Þ=106 ð1Þ

Annual metal (Zn, Cu and Pb) loads for baseflow (Lb, inkilograms per year) were calculated from average total metalconcentrations (C, in grams per cubic metre) and net flowmeasurements (Q, in cubic metres per minute) downstreamof all three air-conditioning discharge points, assuming equalaverage flow on 365 days of the year (Eq. 2).

Lb ¼ C � Q � 5:27 � 105� �=103 ð2Þ

3 Results and discussion

3.1 Contaminated discharge

Average and peak flow for baseflow and stormflow eventsare given in Table 1, along with antecedent dry days,average rainfall intensities and peak intensities, which areimportant parameters in influencing contaminant discharge

0

100

200

300

400

500

600

0 50 100 150 200

mg

m-3

time since start of rainfall [min]

Total Zn in stormflow

12-May

24-May

20-Dec

ANZECC 90%

0

10

20

30

40

50

60

0 50 100 150 200

mg

m-3

time since start of rainfall [min]

Total Cu in stormflow

0

20

40

60

80

100

120

140

0 50 100 150 200

mg

m-3

time since start of rainfall [min]

Total Pb in stormflow

Fig. 1 Concentrations (milligrams per cubic metre) of total Zn (a), Cu(b) and Pb (c) throughout three storm events in 2006 compared withthe designated 90% concentration thresholds for the protection ofaquatic species (ANZECC 2000). Note different scales on the y-axisfor each figure

Environ Sci Pollut Res (2012) 19:903–911 905

(e.g. Egodawatta et al. 2009; Waara and Färm 2008; Wickeet al. 2009). Not all storm events were measured, so a robustmethod (employed by the Christchurch City Council) toestimate annual storm runoff for the ephemeral reach of thecatchment was used (CCC 2003). Annual stormwaterdischarge was calculated for typical Christchurch rainfalland impervious conditions (see Eq. 1), resulting in adischarge of 172,400 m3 year−1. Stream baseflow (measureddownstream of the last air-conditioning discharge point) was5.0 m3 min−1, with diurnal variability arising from the air-conditioning systems being switched off at night andupstream variability accounted for by fewer air-conditioning inputs. Overall, the 2,675,304 m3 year−1 ofstream baseflow from air-conditioning discharge is 15 timesgreater than discharge originating from stormwater runoff.

3.2 Metal concentrations

Average concentrations of dissolved (<0.45 μm) and total(sum of dissolved and particulate) key metals in baseflow,raw air-conditioning pipe effluent (before it entered thestream) and stormflow are given in Table 2. Concentrationsof the relevant effects-based ecotoxicological thresholds forprotecting 90% of freshwater species in New Zealand and

Australia (ANZECC 2000) are also provided to highlightpotential ecotoxicity.

3.2.1 Stormflow

Total (Zn, Cu, Pb) and dissolved (Zn and Cu) metalconcentrations in stormflow greatly exceeded the 90%ecotoxicological guidelines (Table 2), indicating that atleast 10% of the aquatic species in a freshwater ecosystemof this type would be negatively affected by the (untreated)stormwater runoff (ANZECC 2000). While there was little(3.5%) dissolved Pb in the stormwater, Zn was approxi-mately 56% dissolved at 153 μg L−1, while Cu was 25%dissolved at 4.0 μg L−1 (Table 2). The ecotoxicologicalthresholds were exceeded throughout all storm eventsmeasured, with distinct ‘first-flush’ phenomena apparentfor all metals, but not until at least 20 min after the rainfallcommenced, representing the lag time of entry to thereceiving waterway (Fig. 1a–c). EMC, a measure of flow-weighted concentration over an event and typically appliedto stormwater monitoring (Davie et al. 2001; Herngren etal. 2005; Lee et al. 2004), was calculated for the stormflow.Total EMCs were 18-fold (Zn), 9-fold (Cu) and 5.7-fold(Pb) above the 90% species protection threshold (Fig. 2).

Table 1 Flow conditions and storm rainfall intensities measured in the Okeover stream in 2006 and 2009

Condition Location on campus Average flow(m3 min−1)

Max. flow(m3 min−1)

Antecedent dryperiod (days)

Average rainfallintensity (mm h−1)

Peak rainfallintensity (mm h−1)

Stormflow Lower ephemeral reach(20 Dec 2006)

4.32±4.17 8.4 <2 10.5 19.2

Upper ephemeral reach(12 May 2006)

0.889±1.14 3.24 7 1.5 3.8

Upper ephemeral reach(24 May 2006)

8.25±2.11 10.11 5 4.4 5.6

Baseflow Transect of perennialreach (Dec 2006)

1.68±1.53 3.78

Transect of perennialreach (Dec 2009)

2.40±2.29 5.0

Table 2 Average baseflow (n=4 in 2006, n=10 in 2009), raw air-conditioning pipe effluent (2009, n=8) and stormflow runoff (2006, n=26)metal concentrations (in micrograms per litre±standard deviation, <d.l.—below instrument detection limit)

Metal 90% ANZECC[μg L−1]

Metal concentration (μg L−1)

Baseflow Air-conditioning pipe Stormflow

2006 2009 2009 2006

Total Diss. Total Diss. Total Diss. Total Diss.

Zn 15.0 7.8±4.0 8.3±4.1 15.3±9.8 13.5±3.2 12.4±0.5 12.2±4.4 271.0±39.0 153.0±54.0

Cu 1.8 6.8±3.6 5.0±4.3 12.4±7.1 5.1±1.9 8.8±3.6 5.9±3.4 16.0±12.0 4.0±2.0

Pb 5.6 0.28±0.20 <d.l. 1.65 ±0.84 0.07±0.02 0.54±0.09 0.08±0.03 26.0±7.0 0.90±0.30

Diss. dissolved

906 Environ Sci Pollut Res (2012) 19:903–911

Kayhanian et al. (2008) found that 90% of metal toxicitywas observed during the first 30% of the storm duration,with Zn and Cu being the prime toxicants, highlighting theneed to target these contaminants in initial stormwatervolumes.

Various catchment conditions, such as number ofantecedent dry days, surface type, vehicle condition anddriving behaviour, and rainfall characteristics, affect urbancontaminant wash-off and subsequent transport to receivingwaterways (Egodawatta et al. 2009; Herngren et al. 2005;Lee et al. 2004; Waara and Färm 2008; Wicke et al. 2009).

Variable metal concentrations in stormwater runoff mea-sured throughout the events monitored in this study (Fig. 1)may be explained by different precipitation conditionswithin the same catchment (e.g. Table 1). For instance,the 20 December 2006 storm event had a lower antecedentdry period of <2 days but high average (10.5 mm h−1) andpeak (19.2 mm h−1) rainfall intensities as flow increased to8.4 m3 min−1 (in 35 min), which constituted the first flush(Table 1). Metal concentrations during this event wereinitially higher than those of the other storm events (Fig. 1).By comparison, the 24 May 2006 event, with a greater 5-dayantecedent dry period, had comparatively lower average(4.4 mm h−1) and peak (5.6 mm h−1) rainfall intensities(Table 1), resulting in lower metal concentrations in therunoff (Fig. 1).

Metal concentrations from a copper (alloy) roof dis-charging directly into the Okeover stream were alsomeasured throughout a single storm event in December2009 to discern differences in metal levels between roofand pavement-derived stormwater. A distinct exponentialwash-off pattern after the start of the rainfall event(preceded by two antecedent dry days) was observed(Fig. 3). Total (of which 65% was dissolved) Cu concen-trations of 9,500 μg L−1 (during the ‘first-flush’ period)down to 960 μg L−1 (after 3 h of runoff) far exceeded roadand car park runoff stormflow levels (16 μg Cu L−1;Table 2). This corresponded to an exceedance range of 530 to5,300 times the 90% ANZECC ecotoxicological thresholds,highlighting substantial contamination from copper roofsduring rainfall events. (In another New Zealand study, copperroof runoff concentrations up to 7,690 μg Cu L−1 are reported,which concurred with this study) (Pennington and Webster-Brown 2008). While total Zn (range of 22–182 μg L−1, of

0

5

10

15

20

25

Stormflow Baseflow

Exc

eed

ence

Fac

tor

ANZECC (90%) Exceedence: water

Cu Pb Zn

Fig. 2 Exceedance magnitudes of the 90% levels for the protection ofaquatic species (ANZECC 2000) of Zn, Cu and Pb for baseflow andstormflow in 2006. Values were calculated from total EMC measuredin micrograms per litre ± standard deviation, n=26 (stormflow)and n=4 (baseflow)

0

2000

4000

6000

8000

10000

12000

0

100

200

300

400

500

600

Cu

[u

g/L

]

Zn

[u

g/L

]

roof runoff (copper roof)

total Zn

total Cu

0

10

20

mm

ra

in

Fig. 3 Concentrations(micrograms per litre) ofdissolved Cu and Zn fromuntreated roof runoff along withrainfall intensity (millimetres perhour) during a 3-h storm eventin November 2009 preceeded bytwo antecedent dry days

Environ Sci Pollut Res (2012) 19:903–911 907

which 90% were dissolved) levels in the copper-alloy roofrunoff also exceeded the 90% ANZECC guideline value of15 μg L−1, concentrations in road and car park runoff weresubstantially higher at 271±39 μg L−1.

3.2.2 Baseflow

Baseflow chemical signatures differed to those of stormwaterin that dissolved fractions of Zn and Cu in the former weremuch greater. For instance, Zn was predominantly (88%)dissolved in baseflow, but only 56% dissolved in stormflow,while Cu was between 41% (2009) and 74% (2006)dissolved in baseflow, but only 25% dissolved in stormflow(Table 2). However, actual dissolved concentrations of Cu inbaseflow were typically comparable to those of stormflow(4.0–5.1 μg L−1; Table 2), so only their relative speciationamounts differed. Zn and Cu were predominantly in thedissolved state, which is concerning: dissolved metal speciesare known to be more bioavailable and hence of a greaterecological threat to aquatic biota (Kayhanian et al. 2008;Walsh et al. 2005; Wood and Shelley 1999), which mayexplain the depauperate macroinvertebrate fauna found in thestream over a 6-year monitoring period by Winterbourn et al.(2007). In baseflow, most of the copper is likely to bebioavailable all the time due to its dominance in thedissolved state and continuous input from air-conditioningdischarge. The tendency of these metals to remain indissolved forms is reported elsewhere (Mosley and Peake2001; Sutherland et al. 2000).

It was surprising to discover that average total (6.8–12.4 μg L−1) and dissolved (5.0–5.1 μg L−1) copper levelsin baseflow (Table 2) consistently exceeded the 90%(1.8 μg L−1) ANZECC ecotoxicological guidelines by afactor of 5.35 and 2.8, respectively (Fig. 2). These resultsshow that substantial Cu contamination from a sourceindependent of stormwater runoff was occurring. Furtherinvestigation of ‘raw’ discharge from air-conditioning pipesprovided total Cu levels of 8.8±3.6 μg L−1 and dissolvedCu levels of 5.9±3.4 μg L−1 (Table 2). These were 4.9 and3.4 times greater than the 90% ecotoxicological protectionlevel (ANZECC 2000), respectively.

It is known that pipe age and dimensions, as well aswater sitting in the pipe (i.e. stagnant due to diurnal usage),cause electrochemical and thermodynamic changes to thecopper material, resulting in Cu dissolution (Lagos et al.2000). Fluctuations in water flow (such as operationalchanges) cause increased corrosion potential resulting frommore frequent thermal gradients, velocity changes, higheroxygen concentrations and mechanical pitting (Royuela andOtero 2003; Merkel and Pehkonen 2006). This probablyexplains the Cu contamination originating from the diur-nally operating air-conditioning pipes discharging to theOkeover stream. Critchley et al. (2002) and Zhang et al.

(2008) found that biofilm presence, especially duringovernight stagnation in pipes, enhanced Cu corrosion,which was attributed to a reduction in pH caused bymicrobial nitrification.

The Zn concentration in raw air-conditioning pipedischarge (at 12.4±0.5 μg L−1, 98% dissolved, Table 2)was marginally below the 90% ecotoxicological thresholdof 15 μg L−1, showing that discharge from air-conditioningconduits may also be of concern with respect to potentialZn toxicity. Lead concentrations were not an issue duringbaseflow conditions when compared to the ANZECC 90%ecotoxicological protection level of 5.6 μg L−1, although aremarkable increase between average total concentrations in2006 (0.28 μg L−1) and 2009 (1.65 μg L−1) was measured(Table 2). The large increase in baseflow metal concen-trations measured in 2009, compared with earlier samplingevents in 2006, was thought to have resulted from metalresuspension from stream-bed sediments that were depos-ited during earlier storm events. Although TSS concen-trations in baseflow were low, the pH was acidic (6.1–6.8),and it is believed that metal concentrations in resuspendedsediment (during peak discharge events) changed intodissolved forms. This explains both the increased dissolved(and hence, total) metal concentrations in the later samplingevent (Table 2).

3.3 Annual metal loads

Contaminant loading to the stream on a longer (i.e. annual)basis was calculated in order to estimate potential ecotox-icity from cumulative effects. Annual total metal contam-inant loads were calculated for stormflow (Eq. 1) in 2006and for baseflow (Eq. 2) in 2006 and in 2009 (Fig. 4).

0

10

20

30

40

50

60

Zn Cu Pb

Lo

ad [

kg]

Annual total metal load [kg]

Baseflow 2006

Baseflow 2009

Stormflow

Fig. 4 Annual average metal loads (kilograms per year) for stormflowin 2006 (n=26) entering the Okeover stream in the upper ephemeralcatchment (45 ha) and for baseflow in the perennial reach in 2006 (n=4)and 2009 (n=6)

908 Environ Sci Pollut Res (2012) 19:903–911

Annual Pb loading to the Okeover stream from storm-flow comprised 90% at 4.5 kg year−1 compared with0.5 kg year−1 Pb from baseflow (10%) in 2006 or4.3 kg year−1 in 2009 (Fig. 4). Although the use of lead-based paint (and fuels) was discontinued in New Zealand inthe mid-1980s, Pb is still found in stormwater runoff anddeposited sediments in Christchurch urban waterways (e.g.Zanders 2005; Wicke et al. 2009). Annual Zn loading fromstormflow comprised 75% at 46.7 kg year−1 compared with15.4 kg year−1 from baseflow (25%) in 2006 or40.1 kg year−1 in 2009 (Fig. 4). Annual Cu loading fromstormflow only comprised 17% at 2.8 kg year−1 comparedwith Cu from baseflow (83%) at 13.6 kg year−1 in 2006 or32.6 kg year−1 in 2009 (Fig. 4). This equated to a baseflowCu annual load of 4.9 times greater than stormflow (in2006). These data show substantial Cu contamination fromair-conditioning discharge compared to diffuse stormwaterinputs and the importance in monitoring both point andnon-point discharges. The relatively high annual baseflowmetal loads primarily result from much greater baseflowvolumes (2,675,304 m3) compared to stormwater volumes(172,400 m3).

3.4 Sediment metal concentrations

Metal concentrations in stream-bed sediments were com-pared with the Interim Sediment Quality Guidelines(ISQG), which are appropriate for these metals in NewZealand (McCready et al. 2006). Inherent variability wasobserved for all metals, despite each composite being amixture of three separate samples taken at each transectpoint within each reach. Metal-enriched sediments werefound throughout the stream, with concentrations exceedingISQG-low guidelines by up to a factor of 4 for Pb, 3.4 forZn and 1.3 for Cu in the ephemeral reach (Table 3).Sediment metal data for this stream transect reported earlierby Blakely and Harding (2005) suggest that substantialmetals are already present in the stream bed. Zinc wasbetween 364 mg kg−1 (perennial reach) and 677 mg kg−1

(ephemeral reach) in our 2006 investigations comparedwith 155 mg kg−1 throughout the stream transect in 2003(Blakely and Harding 2005). Copper was 51 mg kg−1

(perennial reach) and 82 mg kg−1 (ephemeral reach) in2006 and 64 mg kg−1 in 2003 (Blakely and Harding 2005).Additionally, Pb was 199 mg kg−1 in the ephemeral reach in2006 (but with substantial variability primarily due to1,100 mg Pb kg−1 at one transect point at the beginning ofthe ephemeral reach where the majority of stormwaterenters) and averaged 84 mg kg−1 for the whole stream in2003.

Copper did not appear to remain in the sediments to thesame extent as Zn since the ISQG-low levels were onlyexceeded by a factor of 0.8 (perennial reach) to 1.3-fold(ephemeral reach) (Table 3). This result was particularlyrevealing given the substantial annual copper loads to thestream, coupled with the fact that the perennial reachreceived most of the annual Cu load from contaminated air-conditioning discharge during baseflow conditions (Fig. 4).However, Cu entering the stream from the air-conditioningdischarge is predominantly in a dissolved state (40–73%)and is well reported to dominate as a dissolved species(Mosley and Peake 2001; Sutherland et al. 2000), likelyaccounting for lower sediment concentrations. By compar-ison, Pb associated more with particulate forms (see totaland dissolved lead concentrations in Table 2) and was fourtimes above the ISQG-low levels in the sediments of theephemeral reach in 2006 (Table 3), which may affectmacroinvertebrates that bury in sediments and/or ingestsediments to obtain nutrition. Other studies have shown thatmany urban aquatic biota are affected by contaminatedstream sediments (Paul and Meyer 2001; Suren et al. 2005).Furthermore, the in-stream sediments are probably anaccumulating source of metal contaminants with thepotential risk to remobilise under uncertain perturbations.The slightly acidic pH in this stream during all conditions(6.1–6.8) can facilitate Zn and Cu remobilisation of earlierstream-bed deposition (i.e. from storm inputs), leading toelevated concentrations during baseflow conditions.

3.5 Water quality and ecosystem health

In situ water quality parameters indicated degraded waterquality. Turbidity levels (46.7–59.7 nephelometric turbidityunits, NTU) in stormwater runoff exceeded the ecotoxico-

Table 3 Sediment metal concentrations (milligrams per kilogram) and exceedance factors in relation to ISQG guidelines (ANZECC 2000) alonga transect in the Okeover stream (n=5, median±standard deviation, n.m.=not measured)

Metal ISQG-low Ephemeral reach Perennial reach

[mg kg−1] [mg kg−1] Exceedance factor [mg kg−1] Exceedance factor

Zn 200 677±196 3.4±1.0 364±153 1.8±0.8

Cu 65 82±60 1.3±0.9 51±134 0.8±2.1

Pb 50 199±442 4.0±8.8 n.m. −

Environ Sci Pollut Res (2012) 19:903–911 909

logical thresholds (stipulated as 5.6 NTU, ANZECC 2000)ninefold, undoubtedly impacting ecological communities(Blakely and Harding 2005; Suren et al. 2005; Walsh et al.2005). TSS levels were below the detection limit in baseflowbut rose sharply during the first-flush in storm events to peakconcentrations of up to 412 gm−3, highlighting the contri-bution of sediments from stormwater runoff. TSS and totalmetal stormwater concentrations yielded Pearson’s correla-tion coefficients. Pb (r2=0.66) and Cu (r2=0.63) correlatedsomewhat to TSS but Zn (r2=0.19) did not.

Although dissolved oxygen levels were typically lowerduring storm events (7.75–8.26 gm−3 or 73.7–81.7%) thanduring baseflow conditions (8.3–9.0 gm−3), they did notreach critically low levels below 5 gm−3 or cause a changein redox conditions to become reducing (ORP measure-ments were always greater than +100 mV), so probably didnot lead to chronic biological impairment. Throughout thestudies, pH was slightly acidic (6.1–6.8), potentiallyenhancing metal dissolution and hence mobility, whichmay account for some Zn and Cu remobilisation from thesediments. Static pH tests performed by Hur et al. (2003)also revealed that copper leaching from brake wear debris ishighly pH-dependent, with greater leaching occurring atlower pH values. It is likely that the stream’s biologicalcommunities and hence ecological integrity were impactedfrom synergistic effects of Cu-contaminated air-conditioningdischarge, diffuse suspended solids and metals vectoredthrough sporadic stormwater inputs (especially during the‘first-flush’ period (e.g. Kayhanian et al. 2008) and complexremobilisation from contaminated in-stream sediments.

Winterbourn et al. (2007) found low MacroinvertebrateCommunity Index (MCI) and Semi-Quantitative MCI(SQMCI) values in this stream over a monitoring periodof 6 years (2000–2005). Poor water quality during heavyrainfall events and heavy metals in bed sediments weresuspected as contributing factors to the low indices, with adominance of species tolerant to poor water quality ordegraded habitat conditions. Data from our study validatedthese previous propositions but also discovered an addi-tional contaminating source of Cu and to some extent Zn,predominantly in a bioavailable form. This air-conditioningsource feeds the stream continuously during baseflow andis not diluted by other uncontaminated flow to mitigate theecotoxicological impact(s).

4 Conclusions

Continuous input of untreated stormwater runoff, subse-quent contaminated in-stream sediments and especiallycorroding air-conditioning pipes have caused significantwater quality impairment to the Okeover stream. Conse-quently, supporting ecological integrity to this urban stream

is an on-going challenge. Environmental investigations of thewaterway revealed that on an annual basis, air-conditioningdischarges, which provide the only baseflow to the stream,resulted in greater net metal contamination (especially of Cu)than diffuse stormwater inputs provide. There was between4.9- and 11.6-fold greater annual Cu load to the stream fromair-conditioning discharge compared to stormwater runoff,primarily attributed to much greater baseflow volumesentering the stream. Of priority now is to mitigate Cucontamination from point-source discharge of deterioratingair-conditioning conduits. These investigations highlightpotential contamination of waterways receiving air-conditioning discharge from conventional copper plumbingsystems that should be investigated.

Acknowledgements University of Canterbury Facilities Management,particularly Dr. Kate Hewson, provided logistical support for this study.The Christchurch City Council and Environment Canterbury paid formany of the analytical costs in 2006. Peter McGuigan in theEnvironmental Engineering Laboratory provided technical guidanceand support. Stuart Farrant, Eleanor Taffs, Will Jacobsen and IngridCooper assisted with field reconnaissance and sampling, and Dr. CreonUpton provided feedback on the text.

References

ANZECC (2000) Australian and New Zealand guidelines for fresh andmarine water quality, Australian and New Zealand Environment andConservation Council

APHA (2005) Standards methods for the examination of water andwastewater, 21st edn. American Public Health Association,Washington DC

Beasley G, Kneale P (2002) Reviewing the impact of metals andPAHs on macroinvertebrates in urban watercourses. Prog PhysGeog 26:236–270

Blakely T, Harding J (2005) Longitudinal patterns in benthiccommunities in an urban stream under restoration. New Zeal JMar Fresh 39:17–28

Brown J, Peake B (2006) Sources of heavy metals and polycyclicaromatic hydrocarbons in urban stormwater runoff. Sci TotalEnviron 359:145–155

CCC (2003) Waterways, wetlands and drainage guide; part B; design.Christchurch City Council, Christchurch

Critchley MM, Cromar NJ, McClure N, Fallowfield HJ (2002) Biofilmsin copper plumbing systems: sensitivity to copper and chlorine andimplications for corrosion. Wat Sci Technol 2(4):81–87

Davie AP, Shokouhian M, Ni SB (2001) Loading estimates of lead,copper, cadmium, and zinc in urban runoff from specific sources.Chemosphere 44:997–1009

Egodawatta P, Thomas E, Goonetilleke A (2009) Understanding thephysical processes of pollutant build-up and wash-off on roofsurfaces. Sci Total Environ 407:1834–1841

Gobel P, Dierkes C, Coldewey WC (2007) Storm water runoffconcentration matrix for urban areas. J Contam Hydrol 91:26–42

Herngren L, Goonetilleke A, Ayoko GA (2005) Understanding heavymetal and suspended solids relationships in urban stormwaterusing simulated rainfall. J Environ Manage 76:149–158

Hur J, Yim S, Schlautman MA (2003) Copper leaching from brakewear debris in standard extraction solutions. J Environ Monit5:837–843

910 Environ Sci Pollut Res (2012) 19:903–911

Karlen C, Wallinder IO, Heijerick D, Leygraf C (2002) Runoff rates,chemical speciation and bioavailability of copper released fromnaturally painted copper. Environ Pollut 120:691–700

Kayhanian M, Stransky C, Bay S, Lau SL, Stenstrom MK (2008)Toxicity of urban highway runoff with respect to storm duration.Sci Total Environ 389:386–406

Lagos GE, Reveco FD, Peters DC, Maggi C (2000) Exposure of thepopulation of Santiago, Chile, to copper contained in drinking water.In: Interdisciplinary perspectives on drinking water risk assessmentandmanagement, book series: IAHS publication, issue 260, 112–113

Lee H, Lau S-L, Kayhanian M, Stenstrom MK (2004) Seasonal firstflush phenomenon of urban stormwater discharges. Water Res38:4153–4163

McCready S, Birch G, Long E, Spyrakis G, Greely CR (2006) Anevaluation of Australian sediment quality guidelines. ArchEnviron Contam Toxicol 50:306–315

Merkel TH, Pehkonen SO (2006) General corrosion of copper indomestic drinking water installations: scientific background andmechanistic understanding. Corros Eng Sci Technol 41:21–37

Mosley LM, Peake BM (2001) Partitioning of metals (Fe, Pb, Cu, Zn)in urban run-off from the Kaikorai Valley, Dunedin, NewZealand. New Zeal J Mar Fresh 35:615–624

Nolan D (2005) Environmental and resource management law, 3rdedn. Lexis Nexis, Wellington

Paul M, Meyer J (2001) Streams in the urban landscape. Annu RevEcol Syst 32:333–365

Pennington SL, Webster-Brown JG (2008) Stormwater runoff qualityfrom copper roofing, Auckland, New Zealand. New Zeal J MarFresh 42:99–108

Royuela J, Otero E (2003) The assessment of short-term data of pipecorrosion in drinking-water. 2. Copper. Corros Sci 34:1595–1606

Shaw DC, Bible K (1996) An overview of forest canopy ecosystemfunctions with reference to urban and riparian systems. NorthwestSci 70:1–6

Shedden B, Batchelor J, Chizmar J (2007) Roof runoff study by NewZealand metal roofing manufacturers, South Pacific StormwaterConference, Auckland, 14 pp

Suren AM, Riis T, Biggs BJF, McMurtrie S, Barker R (2005)Assessing the effectiveness of enhancement activities in urbanstreams: I. habitat responses. River Res Appl 21:381–401

Sutherland RA, Tack FM, Tolosa C, Verloo M (2000) Operationallydefined metal fractions in road deposited sediment, Honolulu,Hawaii. J Environ Qual 29:1431–1439

Tsihrintzis VA, Vasarhelyi GM, Lipa J (1995) Multiobjectiveapproaches in fresh-water wetland restoration and design. WaterInt 20:98–105

Waara S, Färm C (2008) An assessment of the potential toxicity ofrunoff from an urban roadscape during rain events. Env SciPollut Res 15:205–210

Walsh C, Roy A, Feminella J, Cottingham P, Groffman P, Morgan R(2005) The urban stream syndrome: current knowledge andsearch for a cure. J N Am Benthol Soc 24:706–723

Wicke D, Cochrane TA, O’Sullivan AD (2009) Developing a rainfallcontaminant relationship model for Christchurch urban catchments.In: Proceedings of the 6th Pacific Stormwater Conference, NewZealand Water Association, Auckland, New Zealand, 29 April–1May, 10 pp

Winterbourn M, Harding JS, McIntosh A (2007) Responses of benthicfauna of an urban stream during 6 years of restoration. New ZealNat Sci 32:1–12

Wood T, Shelley M (1999) A dynamic model of bioavailability ofmetals in constructed wetland sediments. Ecol Eng 12:231–252

Zanders JM (2005) Road sediment: characterisation and implicationsfor the performance of vegetated strips. Sci Total Environ339:41–47

Zhang Y, Triantafyllidou S, Edwards M (2008) Effect of nitrificationand GAC filtration on copper and lead leaching in homeplumbing systems. J Environ Eng-ASCE 134:521–530

Environ Sci Pollut Res (2012) 19:903–911 911