Titanium in UK rural, agricultural and urban/industrial rivers: Geogenic and anthropogenic colloidal/sub-colloidal sources and the significance of within-river retention

Science of the Total Environment 409 (2011) 1843–1853

Contents lists available at ScienceDirect

Science of the Total Environment

j ourna l homepage: www.e lsev ie r.com/ locate /sc i totenv

Titanium in UK rural, agricultural and urban/industrial rivers: Geogenic andanthropogenic colloidal/sub-colloidal sources and the significance ofwithin-river retention

Colin Neal a, Helen Jarvie a, Philip Rowland b,⁎, Alan Lawler b, Darren Sleep b, Paul Scholefield b

a Centre for Ecology and Hydrology, Wallingford, Crowmarsh Gifford, Wallingford, OXON, OX10 8BB, UKb Centre for Ecology and Hydrology, Lancaster, Lancaster Environment Centre, Library Avenue, Bailrigg, Lancaster, LA1 4AP, UK

⁎ Corresponding author. Tel.: +44 1524 595800; fax:E-mail address: [email protected] (P. Rowland).

0048-9697/$ – see front matter © 2010 Elsevier B.V. Aldoi:10.1016/j.scitotenv.2010.12.021

a b s t r a c t
a r t i c l e i n f o
Article history:Received 28 July 2010Received in revised form 2 December 2010Accepted 9 December 2010Available online 24 February 2011

Keywords:RiversTitaniumColloidNanoparticleRibbleWyre

Operationally defined dissolved Titanium [Ti] (the b0.45 μm filtered fraction) in rivers draining rural,agricultural, urban and industrial land-use types in the UK averaged 2.1 μg/l with a range in average of 0.55to 6.48 μg/l. The lowest averages occurred for the upland areas of mid-Wales the highest just downstream ofmajor sewage treatment works (STWs). [Ti] in rainfall and cloud water in mid-Wales averaged 0.2 and 0.7 μg/l,respectively. Average, baseflow and stormflow [Ti] were compared with two markers of sewage effluent andthus human population: soluble reactive phosphorus (SRP) and boron (B). While B reflects chemicallyconservative mixing, SRP declined downstream of STW inputs due to in-stream physico-chemical and biologicaluptake. The results are related to colloidal and sub-colloidal Ti inputs from urban/industrial conurbationscoupled with diffuse background (geological) sources and within-river removal/retention under low flows as aresult of processes of aggregation and sedimentation. The urban/industrial inputs increased background [Ti] byup to eleven fold, but the total anthropogenic Ti input might well have been underestimated owing to within-river retention. A baseline survey using cross-flowultrafiltration revealed that up to 79% of the [Ti] was colloidal/nanoparticulate (N1 kDa i.e. Nc. 1–2 nm) for the rural areas, but as low as 28% for the urban/industrial rivers. Thisraises fundamental issues of the pollutant inputs of Ti, with the possibility of significant complexation of Ti in thesewage effluents and subsequent breakdownwithin the rivers, as well as the physical dispersion of fine colloidsdown to the macro-molecular scale. Although not directly measured, the particulate Ti can make an importantcontribution to the net Ti flux.

+44 1524 61536.

l rights reserved.

© 2010 Elsevier B.V. All rights reserved.

1. Introduction

Over the past decade there has been increasing research intocolloids and nanoparticulates in surface waters (Lead and Wilkinson,2006;Wiggington et al., 2007; Ju-Nam and Lead, 2008). This has comeabout in part with the development of new and sophisticatedmethods of analysis and the needs of studying trace metal sources,fluxes and environmental fates (Andersson et al., 2006; Stolpe andHassellov, 2007; Stolpe et al., 2010). The research has taken on greatersignificance with the expansion in global industrial nanomaterialproduction, potential issues concerning ecosystem damage andprofound knowledge gaps concerning the environmental behaviour,fate and impacts of manufactured nanomaterials (Nowak and Bucheli,2007; Handy et al., 2008). This is because synthetic nanoparticles haveremarkable surface characteristics that make their reactivity tobiological and inorganic environmental systems potentially high(Wiggington et al., 2007). While the research continues to expand

in this area, there is considerable need to developmethods tomeasurenanoparticles in the natural environment and to distinguish ‘back-ground’ geogenic sources from anthropogenic (manufactured)sources and to understand their environmental pathways, dissolvedand colloidal fractionation, bioavailability and risk within theenvironment.

Here, consideration is given to a hydrogeochemically andindustrially important transition metal, titanium (Ti) that is used asTiO2 in paint whiteners, semiconductors, photo catalysts, wastewatertreatments and sunscreens (Ju-Nam and Lead, 2008). Whilst in bulkform TiO2 is largely non-toxic, in nanoparticulate form, TiO2 hasspecific biological properties linked to its photo catalytic activity andhas been shown to be toxic to fish, linked to oxidative stress and otherphysiological impacts (Federici et al., 2007). Titanium is one of themost abundant elements in the earth's crust (it is 9th of all elementsand second only to iron within the transition metals) and it is acommon constituent of rocks, soils and sediments (Skrabal, 1995;Skrabal and Terry, 2002; Greenwood and Earnshaw, 2005). Itpredominantly occurs as resistate and weathering residues as oxides(rutile and anatase, TiO2), mixed-oxide (ilmenite, FeTiO3) andsilicates (e.g. the clay minerals) as well as being important

http://dx.doi.org/10.1016/j.scitotenv.2010.12.021

mailto:[email protected]

http://dx.doi.org/10.1016/j.scitotenv.2010.12.021

http://www.sciencedirect.com/science/journal/00489697

1844 C. Neal et al. / Science of the Total Environment 409 (2011) 1843–1853

constituents in rock forming silicate minerals (e.g. pyroxenes andamphiboles). Titanium is mobile at high temperatures and pressuresunder metamorphic and magmatic conditions (Ayers and Watson,1993; Van Baalen, 1993).

Despite its high abundance, the concentrations of Ti in filteredsurface waters is low due to the predominance of an oxidation state ofIV and a high surface charge density that results in a strong affinity foroxygen. The concentrations of Ti in filtered waters is typically at theppb to sub-ppb level (Van Baalen, 1993; Skrabal, 1995; Skrabal et al.,1992; Skrabal and Terry, 2002) and within estuarine to marineenvironments its concentration can be particularly low. Indeed, therecan be major loss of riverine Ti from the filtered fraction as the waterbecomes brackish and through to the continental edge and to theopen seas. Recent thermodynamic analysis on TiO2 solubility(Schmidt and Vogelsberger, 2009) showed truly dissolved concentra-tions of Ti of around 0.05 μg/l within the pH range 4 to 10: this pHrange covers the spectrum of most UK surface waters (Neal andRobson, 2000). Despite this, Ti can still occur at supposedly over-saturated concentrations in natural surface waters and this to a largepart can be explained by the presence of colloidal Ti that can passthrough standard membrane filters (typically 0.45 μm) that are usedin environmental monitoring studies (Kennedy et al., 1974). Indeed,with the use of smaller size filters, greater removal occurs and the lossof Ti during estuarine mixing strongly points to the presence andflocculation of colloidal material (Skrabal, 1995; Skrabal and Terry,2002; Skrabal et al., 1992). Thus, the computed oversaturation maywell link to an overestimation of “truly dissolved” Ti species in solution.Given the high surface charge density, truly dissolved Ti occurspredominantly as oxide/hydroxide complexes for most surface waters(Turner et al., 1981; Schmidt andVogelsberger, 2009), but thepresence oforganic complexes may be important. Given the high crustal abundanceof Ti plus its low solubility, then Ti will be largely transported throughriver system with the suspended sediment (SS) and bedload materials.

For the study, an overview is provided based on monitoringprogrammes for the upper River Severn in mid-Wales and the Ribbleand Wyre, within the industrial heartland of northwest England.These sites cover a spectrum of rural, agricultural and urban/industrialtypologies. We focus on the Ti concentration in the b0.45 μm fractionand denote it as [Ti] and we provide a comment on the particulateloading which has not been directly measured within the study. Thehypothesis tested is that in areas of high population/pollution, [Ti] arehigher relative to pristine/rural backgrounds. To do this, Ti concen-trations are compared with twomarkers of sewage effluent, boron (B)and soluble reactive phosphorus (SRP — essentially orthophosphate)to evaluate effluent sources andwithin-river retention of Ti. Boron is achemically conservative tracer of sewage effluent inputs (Neal et al.,2010a) and so B concentrations, [B], in river water indicates the levelof effluent contamination and the degree of dilution. SRP provides asimilar marker, but with one critical difference: SRP is progressive lostfrom the river water during downstream transport, as a result ofphysico-chemical and biological uptake processes and SRP concentra-tions, [SRP], declines (Neal et al., 2010b). The use of SRP (whencoupled with B) is a novel approach to test for in-river retention. Thisapproach is employed because there is the potential for colloidalaggregation and sedimentation processes (Boncagni et al., 2009;Domingos et al., 2009; Keller et al., 2009). Indeed, the ecotoxicity ofnanoparticles and exposure of organisms in different aquaticenvironmental compartments (e.g. river water or bed sediments)has been directly linked to the nature and extent of their colloidalstability (Valzeboer et al., 2008; Alvarez et al., 2009).

The paper provides new spatial and temporal data that covers themore general aspects of seasonal, hydrological and urban/industrialsources of Ti for a wide range of catchment typologies. It is intended tocomplement and stimulate new research on colloid/nanoparticles forTi. Further, the upper Severn, Ribble and Wyre studies provides along-term base for the Centre for Ecology and Hydrology and the

Natural Environment Research Council within the context of“National Capability”. It is hoped that the work will stimulate newand parallel research at these long-term catchment/basin observato-ries where high quality environmental information is collected, and athand to be shared. Within this context, the results of a baseflowsurvey for the Ribble and Wyre provide an indication of the colloidaland sub colloidal size range within the [Ti].

2. Study area

2.1. Mid Wales

The mid-Wales study focuses on the headwater areas of the riverSevern and the associated Plynlimon experimental catchments (Nealet al., 1997a,b). The area is comprised of moorland and coniferafforested upland areas with no habitation and minimal agriculture(very low density sheep farming in the moorland areas). The bedrockis Ordovician and Silurian slate, shale, mudstone and grit with acovering of acidic soils (peat, podzol and gley). For these sites, rainfall,cloud water and five river water sites were monitored. There are twomain tributaries to the upper Severn, the Afon Hafren and the AfonHore, and both have been sampled at upper and lower sites along theriver. The upper Hafren represents acid moorland (peat) drainage,while the lower Hafren has the upper Hafren input plus a lower half ofSitka spruce plantation, on podzolic soils, that has been progressivelythinned/harvested. The Afon Hore has similar soils and vegetation,although the upper Horemonitoring site represents around 50% forestcover, much of which was harvested in 2008/9. The lower Horerepresents plantation forest that was harvested/clear-felled duringthe mid 1980s. The fifth river monitored is the Nant Tanllwyth thatrepresents Sitka plantation on gley soils — the forest was 50% clear-felled in 1996.

For all plantation sites, following clear felling or thinning, therewas either replanting with saplings or natural reseeding.

2.2. Ribble/Wyre

The northwest England study deals with two river systemsentering the Irish Sea, the Ribble and Wyre. Their basins drainupland areas of outstanding natural beauty (part of the YorkshireDales and the trough of Bowland). In the case of the Wyre, there isthe market town of Garstang in its lower reaches, while the Ribbleincludes part of the urban/industrial heartlands of Lancashire andhistoric towns such as Burnley, Accrington, and Wigan. The geologyof the Ribble is predominantly carboniferous rocks. However on thefringe of the Wyre and Ribble there are Triassic age sandstones andbetween these two there are relatively thin strata of Permianmudstones. Within the Carboniferous group, the rocks decrease inage southwards through a sequence of limestone, calcareousmudstone and siltstone, to millstone grit and the lower coalmeasures of coal bearing rocks including mudstone, siltstone andgrit. Apart from in the headwater areas of mainly millstone grit, theDarwen, Douglas and Calder catchments are confined primarily tothe coal measures as of course the towns in the area largely cameabout in the industrial revolution due to two sources of power, coaland water.

Across the Ribble and Wyre basins, 26 sites were monitored.Fig. 1 provides a map of the study area and the location of thesampling points. Correspondingly, Table 1 provides a list of sites, theseparation between rural, urban/industrial and agricultural typolo-gies and the site identifiers for the other tables, and graphs. WithinTable 1, average data for B and SRP are presented to broadly showthe degree to which the effluent inputs increase in importancedown gradient and the extent of SRP loss relative to B: Table 2provides B and SRP data for the upper Hafren for a backgroundcomparison.

Fig. 1. Location map. Note that the sampling points are marked in sequence as Tables 1and 2. However, in this figure, the prefix C, D, R and W are used to distinguish theCalder, Douglas, Ribble and Wyre basin sequences.

Table 1Monitoring sites in order of basin and downstream sequence for the Ribble and Wyre. Withithe increase in urban/industrial inputs and the SRP illustrates both the urban/industrial inp

River Location No. Site

Calder Basin in downstream sequence

Pendle Barrowford 1 TributaryColne Barrowford 2 TributaryCalder Altham Br 3 Main-stemHyndburn Hyndburn Br 4 TributaryCalder Whalley 5 Main-stem

Douglas Basin in downstream sequence

Yarrow Chorley 1 TributaryDouglas Adlington 2 Main-stemDouglas Standish 3 Main-stemDouglas Parbold 4 Main-stemTawd Hoscar 5 TributaryEller Br Briars Lane 6 TributaryDouglas W B Br 7 Main-stem

Ribble Basin in downstream sequence

Dunsop Dunsop 1 TributaryLoud Mytham Br 2 TributaryHodder L Hodder 3 TributaryRibble Gisburn 4 Main-stemRibble Gt Mitton 5 Main-stemCalder Whalley 5 TributaryRibble Ribchester 7 Main-stemDarwen Roach Br. 8 Tributary

Wyre Basin in downstream sequence

M Wyre Marshaw Br 1 TributaryT Wyre Stoops Br 2 TributaryWyre Abbeystead 3 Main-stemWyre Garstang Br 4 Main-stemCalder Calderbridge 5 TributaryBrock Bilsborough 6 TributaryWyre St Michaels 7 Main-stem

1845C. Neal et al. / Science of the Total Environment 409 (2011) 1843–1853

2.2.1. The RibbleFor the Ribble, three sites were monitored on the main-stem to

represent, the upper, mid and lower reaches with increasing urban/industrial inputs especially from the tributary of the Calder that entersthe Ribble between the middle and lower monitoring sites. Fourtributaries were also monitored, the rural Hodder and the urban/industrial impacted Calder, Darwen and Douglas: 3, 5, 1 and 7 sitesrespectively.

For the Calder and Douglas a series of sites were studied rangingfrom the clean headwater areas through the urban/industrial centreswithin the valleys and to the agricultural areas of the lowlands in thecase of the Douglas. In terms of urban inputs and the industrial towns,the Calder is impacted by Colne, Burnley, Padiam and Accrington, theDarwen by Darwen and Blackburn and the Douglas by Wigan.

2.2.2. The WyreAn upland–lowland sequence of seven sites was established. These

sites included two headwater tributaries (the Tarnbrook andMarshaw Wyre) and two other tributaries (the Brock and Calder)that drain to the lower Wyre. The lower tributaries have headwatersin the Trough of Bowland and drain through farming areas within thelowland. N.B. there are two rivers named “Calder”, one in the Ribbleand the other in the Wyre Basin.

3. Monitoring and chemical analysis

For the mid-Wales studies, monitoring of Ti of rainfall, cloudwater and stream water began in October 1998. The sampling was

n the table, average concentrations of boron and SRP are included. The boron illustratesuts and within-river decay. The average represents the mean for the full dataset.

Typology Avg B (μg/l) Avg SRP (μg/l)

Rural 17 22Urban/industrial 51 176Urban/industrial 48 142Urban/industrial 68 44Urban/industrial 54 200

Rural/urban 25 36Rural/urban 34 208Urban/industrial 44 188Urban/industrial 84 38Agricultural 74 72Agricultural 54 346Urban/industrial 94 1222

Rural 9 4Rural 19 68Rural 13 19Rural 17 58Rural 18 45Urban/industrial 54 200Rural 27 113Urban/industrial 71 550

Rural 9 6Rural 9 6Rural 9 8Rural 13 20Rural 16 28Rural 16 46Rural/urban 18 167

Table 2Average, baseflow and stormflow concentrations of boron, SRP and dissolved Ti for rainfall, cloud water and stream chemistry for the upper Severn. Conifer harvesting times1=gradual thinning of forest, 2=clear felling of lower half of catchment inmid 1980s (ourmonitoring deals with forest reestablishment), 3=felling 2008/9, 4=felling in 1996. Theaverage represents the mean for the full monitoring period. The baseflow and stormflow averages represent the average for the bottom and top 10% of flows through the fullmonitoring period. The rainfall and cloud water data are highly variable and skewed to high concentrations at lower flows. A better comparison with the streams is to use the flowweighted averages for the inputs: 0.16 and 0.48 μg/l, for rainfall and cloud water, respectively.

Site Typology B (μg/l) [Ti] (μg/l) [Ti] ratio

Average Average Baseflow Stormflow Storm/Base

Atmospheric inputsRainfall Rainfall 5.2 0.27⁎ 0.26 0.23 0.88Cloud water Cloud water 23.9 0.33* 1.07 0.12 0.11

RiversU Hafren Moorland 6.5 0.55 0.58 0.57 0.98L Hafren 50% forest1 5.6 0.57 0.61 0.6 0.98U Hore 50% forest2 8.1 0.65 0.75 0.67 0.89L Hore 77% forest3 8.0 0.6 0.66 0.56 0.85Tanllwyth 100% forest4 8.9 0.69 0.68 0.77 1.13

Stream averages and ranges across the riversAverage 7.4 0.61 0.66 0.63 0.83Minimum 5.6 0.55 0.58 0.56 0.11Maximum 8.9 0.69 0.75 0.77 1.13


weekly, with bulk collection for rainfall and cloud water, andinstantaneous grab samples of river water. The cloud water wascollected using a lidded harp type system. For trace elementdetermination, the samples were acidified to 1% concentrated highpurity concentrated nitric acid (as a preservative) and kept in pre-cleaned (10% hydrochloric acid followed by high purity water rinses)plastic bottles.

For the Ribble and Wyre study monitoring began on the 18thFebruary 2008 on a fortnightly basis for one year. As the distancescovered were large, the monitoring programme was split, with halfthe sites monitored one week and the other half the next week. Afterthe first year, the monitoring was monthly and in this case the samesplit of monitoring was used with samples being collected with afortnightly stagger.

For both studies similar methods of sample processing andchemical analysis was used. Filtration employed 0.45 μm cellulosenitrate filter circles and was undertaken on the day of sampling.Inductively Coupled Plasma-Mass Spectrometry (Perkin Elmer ElanDRC II) and Inductively Coupled Plasma-Optical Emission Spectrom-etry (Perkin Elmer DV4300) was used to determine [Ti] and [B],respectively. Soluble reactive phosphorous was analysed colorimet-rically using a Seal AQ2 discrete analyser using molybdenum bluechemistry. Suspended sediment concentration [SS] was determinedgravimetrically by passing one litre of water through aWhatman GF/Cfilter and drying the filters to 105 °C.

For the analysis, instruments were calibrated on the day of useusing a range of standard solutions. Values exceeding the topcalibration standard were diluted and the analysis was repeated.Analysis was validated using values from the analysis of internal QCsamples and regular proficiency testing samples supplied by Aqua-check ltd. The analysis was conducted using mainly methodsaccredited by UKAS to ISO 17025.

In the case of the upper Severn catchments, SRP concentrationswere usually low and often below the detection limit. In this case, theaverage concentrations were set to 5 μg/l and this reflects the lowvalues encountered: details of [SRP] for the catchment are provided inNeal et al. (2003).

For [Ti] measurements, there were around 420 data-points eachfor rainfall and cloudwater and around 520 data-points for each of thePlynlimon stream sites monitored. Correspondingly, there werearound 43 data-points for each of the Ribble and Wyre stream sitesmonitored.

3.1. Cross flow ultrafiltration of baseflow waters

Within the broader context of our hydrochemical studies, a seriesof measurements were taken to examine if ultra-fine particles arepresent in the operationally defined dissolved (b0.45 μm) fraction ofsurface waters for a wide suite of water quality determinands usingcross flow ultrafiltration. For this, a baseflow survey of selected Ribbleand Wyre sites was undertaken during 8–12 September 2008. Allsamples were processed by ultrafiltration within 24 h of samplecollection and primarily on the day of sampling. Cross-FlowUltrafiltration was undertaken using a 1 kDa regenerated cellulosemembrane (Millipore Pellicon 2), to a concentration factor of 15,based on the establishedmethods of Wilding et al. (2005) and Liu andLead (2006). The 1 kDa membrane was tested with Vitamin B12(1.3 kDa) and the results showed a retention coefficient of 0.99. Fivewater samples were taken to cover clean to polluted environmentsacross a range of [Ti]. Two rural upland sites (the TarnbrookWyre andthe Dunsop), an agricultural lowland river (Eller Brook) an industrialimpacted river (Hynburn Brook) and the most sewage effluentcontaminated river (Douglas at Grimshaw Green), were studied.Given the time consuming nature of the filtration procedure, only onesample could be processed each day.

4. Methodological considerations

4.1. Filter size considerations

The filtration size cut-off of 0.45 μm is not universally used toprovide an operationally defined dissolved fraction as other sizes aresometimes used and colloidal fractionmay even be defined in terms ofa upper-bound 1 μm diameter cut-off (Skrabal and Terry, 2002; Leadand Wilkinson, 2006). For our study, a “0.45 μm dissolved fraction” isused in relation to [Ti]. We note that the filters may remove orenhance colloidal formation. In order to minimise such effects, severalaliquots of water were filtered to wash the filters before samples werecollected and the filters were changed before blockage of the filtersoccurred. With regard to ultrafiltration, a 1 kDa membrane was usedto provide an operational separation betweenmolecules and colloidalfractions. For environmental monitoring studies this cut-off isgenerally assumed to correspond approximately to a size exclusionof 1 to 2 nm (Wilding et al., 2005; Liu and Lead, 2006), although thissize exclusion will depend upon the species, surface charge, radius of

Table 3A summary of averages, baseflow and stormflow averages for the Ribble and Wyre aswell as the Douglas and Calder sub-basins. Included in the table is S/B, the ratio of thestormflow to baseflow average concentrations for [Ti]. Calder site 5 is the same asRibble site 6 as the Calder is a significant input to the Ribble. The average represents themean for the full monitoring period. The baseflow and stormflow averages representthe average for the bottom and top 10% of flows through the full monitoring period.Stormflow/Baseflow represents the ration of the stormflow to baseflow averages.

[Ti] (μg/l)

Average Baseflow Stormflow Storm/Baseflow

Calder sequence in downstream order1 1.45 0.41 4.89 11.92 2.69 1.7 7.11 4.23 1.73 1.1 3.99 3.64 2.02 0.92 7.52 8.25 2.16 1.85 4.49 2.4

Douglas sequence in downstream order1 1.55 1.65 2.67 1.62 1.73 2.05 2.42 1.23 2.41 1.96 6.49 3.34 1.51 0.72 4.59 6.45 1.86 1.19 3.86 3.26 3.53 3.34 4.43 1.37 6.48 6.08 7.02 1.2

Ribble sequence in downstream order1 0.87 0.5 1.42 2.82 2.9 0.97 13.05 13.53 1.31 0.5 1.9 3.84 1.26 0.72 2.39 3.35 1.32 0.66 2.09 3.26 2.16 1.85 4.49 2.47 1.74 1.66 3 1.88 4.97 5.63 3.34 0.6

Wyre sequence in downstream order1 1.62 0.55 2.46 4.52 1.81 0.6 2.93 4.93 1.74 0.8 2.87 3.64 1.9 0.46 3.64 7.95 2.17 1.45 6.12 4.26 2.07 0.91 3.89 4.37 3.27 2.04 5.52 2.7

Statistical summaryAvg 2.23 1.56 4.39 4.1Min 0.87 0.41 1.42 0.6Max 6.48 6.08 13.05 13.5


gyration and degree of deformation in the flow field duringultrafiltration. A nanometre corresponds in our case to a moleculargrouping of around 6 to 12 TiO2 molecules: later in the paper wediscuss the findings of the measured fractionation in relation to thecolloidal, sub-colloidal and the “truly dissolved” components.

4.2. Mixing relationships

Here, B and SRP are used as surrogates for effluent inputs. Overmany years, inter chemical determinand relationships have been usedto examine if river water can be explained in terms of mixing of watertypes. This is encompassed within an end-member mixing approach(EMMA: Christophersen et al., 1990). For lowland rivers, B has beenused as prime marker as it has relatively inert, it has low backgroundconcentration and its concentration is particularly high in effluent(Neal et al., 2010a).

Boron concentrations have been reducing in effluents since thelate 1990s (Neal et al., 2010a), but this does not affect the analysishere. This is because there are no effluent sources at Plynlimon andthe monitoring period for the Ribble and Wyre postdates the majorreductions in B in effluents and over the period of monitoring B levelsremain relatively stable.

While B is essentially chemically conserved in the water column,SRP is not and over 50% of the input from STWs can be lost within afew kilometres of the input (Neal et al., 2010b). This has been shownnot only from flux studies along river lengths (Jarvie et al., 2006), butalso within GIS studies of population density versus SRP levels inrivers (Davies and Neal, 2007). Just downstream of STW inputs,average SRP concentrations are linearly correlated with population,but further downstream the relationship becomesmore scattered andthe SRP concentrations are relatively low (Davies and Neal, 2007).

If [Ti] was determined from effluent or population/industrialsources and it remained chemically conserved within the watercolumn, then its concentration would be linearly correlated with B.However, if it was not conserved, or there was more than twochemically distinct water typesmixing, then departures from linearityor “kinks” in linearity might be expected. As SRP also occurs withineffluents and it can be lost (albeit through biological processes) fromthe water column, then it would be expected that [Ti] would be morestrongly correlated if Ti were also lost from the water column. Thus,here, mixing relationships were applied to test for Ti loss. For this,linear regression analysis was undertaken between [Ti] and [B] as wellas with [SRP]. In addition, the regressionwas undertaken between [Ti]and the [SRP]/[B] ratio, a B normalised loss term. Multiple linearregression for [Ti] with [B] and [SRP]/[B] was also undertaken — the Brepresent the effluent-input/dilution, and the SRP/B represents thenon-conservative decay term.

Within the study, average and average baseflow and stormflowconcentrations were examined. In the case of baseflow and stormflowthe average concentrations for the bottom and top 10% of flow,respectively, were used for each site.

5. Results

Table 2 provides summary statistics for [Ti] for the upper Severnand Table 3 statistics for the Ribble/Wyre. The average [Ti] was1.97 μg/l across the upper Severn and Ribble/Wyre sites, with a totalrange in concentration of 0 to 25.2 μg/l. The averages for individualsites varied 11 fold (0.55 to 6.48 μg/l).

5.1. Upper Severn

[Ti] was particularly low in rainfall with an average of 0.27 μg/l anda total range of 0 to 3.4 μg/l. [Ti] was over three times higher in cloudwater (average 0.7 μg/l with a range of 0 to 9.9 μg/l). The data provedto be highly skewed (much more so than the rivers) and median

values (0.10 and 0.40 μg/l, respectively) were around a third those ofthe average. In the case of the cloud water, the highest concentrationsoccurred for low volumes of catch, but there was no clear relationshipfor rainfall. On a flow weighted basis, that will correspond with theactual input of the rainfall (remembering that the catchment dampsdown highly the rainfall response, Neal and Kirchner, 2000), thecorresponding weighted averages are 0.16 μg/l for rainfall and0.48 μg/l for cloud water. These are more appropriate for describingthe normative values to compare with the stream. [Ti] in the upperSevern averaged 0.61 μg/l and there was little difference across themonitored rivers (range in average 0.55 to 0.69 μg/l) although therewas a high scatter to the data (range 0.1 to 12.5 μg/l) with occasionaloutliers. The lowest average value was for the moorland (the upperHafren). However, there was no clear pattern with regard topercentage forest coverage, extent of harvesting, and timing ofharvesting. For the upper Hore, it is worth noting that during fellinglarge amounts of sediment were disturbed as the waters where oftenmurky. Yet this seemed to have little effect on [Ti] and an analysis ofthe temporal patterns in the upper Severn revealed no clearseasonality or flow relationships when either the time series wasplotted directly or against month.

1

1

1 1 1 1 1 1 11 1 1

1

11

1 11

1 1 1 1 1

1

1 1 1 11 1 1

1

1 11 1 1 1

1 1 1 1 12

2

2 2 2 2 22 2

22 2

22

2 22 2

2 2 2 2 2

2

2 2 22

2 2 2

2

22

2 2 2

2

2

2 22

23

3

3 3 3 3 3 3 3 33 3

3

3 3

33 3 3 3 3 3 3

3

3 3 3 3 3 3 33 3 3

3 3 3 3 3 3

3

3 34

4

4 4 4 4 4 4 4 4 44

4

4

4

44

4

4 4 44

4

4

44 4 4 4 4 4

4

4 4 4 4 4 44

4 4 4 45

5

5 5 5 5 5 5 5 5 5 5

5

5

5

5 55

5 5 55

5

5

5 5 5 5 5 5 5 5 55

5 5 55

5 5

55

5

2 4 6 8 10 12 14 16 18 20 22 24 26 280

5

10

15

11

1 1 1

1

1 1 1 1 1 11

1

1

11 1

11 1 1 1

1

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 12 22 2 2

22 2 2 2 2 2 2 2

22 2

22 2 2 2 2

22 2 2 2 2 2 2 2 2 2 2

2 22

2 2 2 2 233

3 3 3

3

3 3 3 3 33

3 3

3

3

3

33 3 3

33

3

3 3 3 3 33 3 3 3 3 3 3

33 3

3 3 3 34 4 4 4 44

4 4 4 4 4 4

4

4

4

4

4

4 4 4 4 4 44

44 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 45

55 5 5 5 5 5 5 5 5 5 5 5

5 5 5 5 5 5 55

5

5

55

55 5 5 5 5 5

5 5 5 55

55 5 5 5

6

6

66

6

6

66 6 6

66

6 6 6 6 6 66 6 6 6

6

6

6 6

66

6 6 6 6 6

6 6 66

66

6 66 6

77

77

7

7 7

7

7

77

7

77

7

7

7 77

7 7 77

7

7 7

7

7 7 77

77

7

7

7 7 7

7

7

7 7 7

2 4 6 8 10 12 14 16 18 20 22 24 26 280

5

10

15

1 1 1 1 1 1 1 1 1 1 1 1 11

1 11 1 1

11 1 1 1 1 1 1 1 1 1 1 1

11

11

11

11 1 1 1 122 2

2

2 2 2

22

2

2

22

2

2

2

2 2 22

2

2

2 2 22

2 2 2

2

2 22

2 2

2

2 2 2 2 23 33 3 3 3

33 3

3

3 3 3 33

33 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3

3

3

3

3

33

3

3 3 3 334 4 4 4 4 4 4 4 4

4

4 4 4 44 4 4 4 4 4 4 4

44 4

4 4 4 4 4 4 4 4 4 4 44 4

4

4

4 4 4 455

5 5 5 5 5

5

55 5

5

5 55 5 5 5 5 5 5

5 5 5 5 5 5 5 5 5 5 5 55

5 55

5

5 5 557 7

7 7 7 7 7 7 77

7

7

7

7

7 7 7 7 7 7 7 7

7

7 7 7 7 7 7 7 7 77 7 7

77 7

77

7 7 7

7

8

8 8 8 88

88

8

8

8

8

8 8 8 8

8 8

8 88 8 8

8

8 8 8 8 88

8

88

8

88

88

88

8

88

8

6

6

6 6 6 6 6 6 6 6 6 6

6

6

6

6 66

6 6 66

6

6

6 6 6 6 6 6 6 6 66

6 6 66

6 6

66

6

2 4 6 8 10 12 14 16 18 20 22 24 26 280

5

10

15

1 1 1 1 1

1

1 1 11 1

1 11

1 11

11 1

11 1 1 1

1 1 1 1 1 11

1

11

11 1 1 1 1 1 1

12 2 2 2 2

2

2 2 22 2 2

22

2 22

2

22

2 2 22

2

2 2 2 2 2 22

2

22 2 2

2 2 22 2 2

2

3 3 3 3 3

3

3 3 3

33 3 3 3 3 3

33

3

33

3 33

33 3 3 3 3 3 3

3

33

33 3

33 3 3 3

34 4 4 4 4

4 4

4 4

4

44 4

4

4 44

4

4

4

4 4 4 4 44 4 4 4 4 4 4

4

4

4

4

4

4

4

44 4

4 45 5 5 5 55

5 5 5

55 5 5

5

55

5

5

5

5

55 5

5 5 5 5 5 5 5 5 5

5

5

5

5

5

5

5

5

5 5 5 56 6 6 6

6

66 6

6 66 6

6

6

6

66 6 6 6 6

66

6

6 6 6 6 6 6 6

6

6

6

6

6

6

6

6

6 6 6 677

7

77

7

7 7 7

7

7

77

7

77

77

77 7

7

77

7

7 7 7 7 7 7 7

7

7

7

77

7

77

7

77

7

2 4 6 8 10 12 14 16 18 20 22 24 26 280

5

10

15

Calder

Douglas

Ribble

Wyre

[Ti]

(µg

/l)

Month ( Month 1 = January 2008)

[Ti] in Ribble/Wyre and tributaries

Fig. 2. Time series for ‘dissolved’ (b0.45 μm) Ti concentrations ([Ti]) within the Calder, Douglas, Ribble andWyre basins. The numbers depict sampling within each basin/sub-basin indownstream order and they are referenced in relation to the site names in Fig. 1 and Tables 1 and 2.


5.2. Ribble/Wyre

[Ti] varied across the sites both in space and time although peakconcentrations often coincided on the same dates (Fig. 2). Forexample, in January 2009, there were clear peaks for many siteswithin the Ribble and Wyre basins and this corresponded with aperiod of snowmelt. We cannot say, however, that the peaks were orwere not universally present across all the sites as with the split insampling only a half of them were monitored in a given week.

[Ti] showed a broad relationship with urbanisation/industry sincethe highest average, baseflow and stormflow concentrations generallyoccurred at downstream sites where the main towns/industrial areasare. Further, on average, [B], [SRP] and [Ti] broadly increased together(Tables 2, 3). Despite this, the average, baseflow and stormflow [Ti]did not simply increase uniformly downstream through the urban andindustrial areas (Fig. 3). Indeed, in the case of the Calder in the Ribblebasin, there was very little difference in [Ti] along its length even

though [B] and [SRP] exhibited large changes. The largest increases in[Ti] occurred just downstream of inputs from major sewage works onthe Darwen and the downstream limit of the monitoring on theDouglas (average 4.97 and 6.48 μg/l, respectively, compared with anaverage for the lower Calder of 2.16 μg/l). Further, there wereintermittent spikes of [Ti] for the rural Loud that significantlyincreased the average, baseflow and stormflow concentrations.These intermittent spikes led to [Ti] being significantly higher forthe average and stormflow (from two fold to an order of magnitude)relative to rural inputs on the upper Ribble and Hodder (Table 3,Fig. 2).

On average, stormflow waters were enriched in Ti by a factor ofaround 4 times that of baseflow and the range in [Ti] during stormflowwas high across the sites (1.42 to 13.05 μg/l). The enrichment washighest for the Loud, while baseflow and stormflow averages werearound the same for the lower Douglas and Darwen, where the point-source inputs were highest.

image of Fig.�2

Site Number (Downstream sequence)

T

T

MM

M

T

TM

MM

T T

M

M

M

1 2 3 4 50

1

2

3

4

T

T

M MM

M

M

T

T

MM M

M

M

TT

MM

MM M

1 2 3 4 5 6 70

3

6

9

T

T

T

MM M M

M

TT

T

MM M M M

T TT

MM

M MM

1 2 3 4 5 6 7 80

1

2

3

4

5

6

7

TT T

M M M

M

TT

T

M

M M

MT

T

TM

MM

M

1 2 3 4 5 6 70

1

2

3

4

[Ti]

(µg

/l)

T = Tributary M = MainstemAverage: square, solid line

Baseflow: diamond, dotted lineStormflow: trainage, dashed line

Ribble

Calder

Douglas

Wyre

[Ti] in the Ribble and Wyre Basins/sub-Basins

Fig. 3. Average, baseflow and stormflow ‘dissolved’ (b0.45 μm) Ti concentrations ([Ti])within the Calder, Douglas, Ribble and Wyre basins plotted in a downstream sequence.The location numbers are referenced in relation to the site names in Tables 1 and 2, andtheir positions marked in Fig. 1. The solid lines join the main-stem samples, while theseparate points represent the tributary inputs. For the upper reaches of the Calder andDouglas, the main upper tributary are defined as main-stem to allow the variationsalong the main-stem from a clean to an impacted system (data-points joined) to beseparated these from tributary inputs further downstream. Solid, dotted and dashedlines represent the average, baseflow and stormflow values for each location.

Table 4Regression statistics for [Ti] with [B] (μg/l), [SRP] (μg/l) and the [SRP]/[B] ratio (μg/μg).

Linear regression

Gradient 2 std Inte

BoronAverage 0.037 0.013 0.93Baseflow 0.025 0.013 0.55Stormflow 0.057 0.018 1.05

SRPAverage 0.005 0.001 1.47Baseflow 0.004 0 0.84Stormflow 0.011 0.005 1.77

SRP/Boron ratioAverage 0.376 0.096 1.05Baseflow 0.314 0.079 0.56Stormflow 0.318 0.269 1.77

Multiple linear regression

Boron SRP/Boron

Gradient 2 std Gradient 2 std

Average 0.022 0.009 0.282 0.04Baseflow 0.013 0.009 0.272 0.03Stormflow 0.053 0.019 0.151 0.09


5.3. [Ti] relationships with [B], [SRP and the [SRP][B] ratio

There was a large range in both [B] and [SRP] across the sites(Tables 1, 2). For example, in the rural headwaters of the Severn,Wyreand Ribble, average [B] and [SRP] were around 8 and 5 μg/l,respectively, while for the most polluted site on the lower Douglas,the average concentrations were around 94 and 1220 μg/l,respectively.

Average, baseflow and stormflow [Ti] were correlated with [B],[SRP] and the [SRP]/[B] ratio although there was some clear scatter tothe relationships in some cases as indicated by the differences in theprobabilities and standard errors for the regression constants (Table 4,Fig. 4). The correlations were generally higher for the average andbaseflow cases and the weakest correlations were for [B] apart fromunder stormflow conditions. For [B], the correlation with [Ti] wasincreased significantly when an [SRP]/[B] component was alsoregressed. The nature of the relationship between these determinandsand location within the river system is illustrated in Fig. 5 for averageswithin the Douglas. For the main-stem of the Douglas, [B] increaseddownstream in line with increasing STW inputs. In contrast, [SRP] andthe ratio of [SRP]/[B] increased moderately downstream for the upperDouglas and then decreased in the mid reaches before increasingsharply at the downstream monitoring point. [Ti] tracked [SRP] andthe [SRP]/[B] ratio.

Within Fig. 5, data for two lowland tributaries (the Tawd and EllerBrook) were included to illustrate that despite them draining rural/agricultural land, the relatively high [B] indicate effluent sources. Atthese tributary sites [Ti] was also relatively high.

The mixing model was also applied to examine [B], [SRP] and [Ti]relationships for the individual monitoring points. However, whenthis was done, the correlations were poor because (apart from theDarwen and lower Douglas), there were no clear end-members ofdistinct [Ti] and the scatter in [Ti] was much higher than the mixingrelationship could pick up.

5.4. Ultrafiltration results

The results of the ultrafiltration baseline study show that for thecleaner upland site the greatest proportion of the [Ti], up to 79%, was

rcept 2 std R2 p

0.5 0.526 b0.0010.71 0.315 0.0010.58 0.565 0b0.001

0.26 0.761 b0.0010.19 0.912 b0.0010.53 0.394 b0.001

0.37 0.672 b0.0010.4 0.681 b0.0010.78 0.157 0.025

Intercept r2 p

error Intercept 2 std

1 0.62 0.33 0.819 b0.0018 0.18 0.45 0.753 b0.0019 0.82 0.64 0.597 b0.001

0 25 50 75 1000

2

4

6

Average

0 25 50 75 100 1250

2

4

6

8

Baseflow

0 20 40 60 800

5

10

15

Stormflow

0 1000 20000

2

4

6

Average

0 1000 20000

2

4

6

8

Baseflow

0 200 400 6000

5

10

15

Stormflow

0 10 200

2

4

6

Average

0 5 10 15 20 250

2

4

6

8

Baseflow

0 5 10 150

5

10

15

Stormflow

[Ti]

(µg

/l)

[B] (µg/l) [SRP] (µg/l) [SRP]/[B] (µg/µg)

[B] vs [Ti] [SRP] vs [Ti] [SRP]/[B] vs [Ti]

Fig. 4. Plots of average, baseflow and stormflow ‘dissolved’ (b0.45 μm) Ti concentrations ([Ti]) concentrations against boron, SRP and the boron/SRP ratio for the upper Severn, Ribbleand Wyre monitoring sites.


in the N1 kDa size range (Table 5) and this corresponds essentially tothe colloidal fraction (mass balance recovery 107%). In contrast, themore polluted sites have higher [Ti], but a much lower percentage ofthe N I kDa component, and down to 28% for the Douglas (massbalance recovery 92%). The amounts of the b1 kDa and the N1 kDacomponent increase linearly with [Ti] and the b1 kDa fraction exceedsthat of the N1 kDa, for [Ti] greater than around 2.5 μg/l (Fig. 6).

5.5. Particulate Ti

In this paper, the focus is on the filterable fraction and the work isnot directly linked to catchment fluxes of Ti in its entirety. ParticulateTi has not been measured in this study due to the lengthy analyticalprocedures required to decompose the solid phase, but somecomment on its potential significance to the total flux can be madebased on two measures. Firstly, within the analytical procedures,samples were acidified before filtration and left overnight to leach and

the acid available particulate Ti concentration [AAPTi] was determinedthe next day following subsequent filtration and subtraction of the [Ti]from the other measurement of Ti (cf method given in Neal et al.,1996, 1997a,b). Secondly, the SS could be compared with the crustalabundance. The results indicated that of the total Ti in the river, onaverage the AAPTi represented about 40% (range in average 25 to 55%)and around 60% at times of stormflow when the [SS] was also at itshighest. These values clearly will be underestimates for the totalparticulate concentrations, and the [AAPTi]/[SS] ratio was around ahalf that for the crustal abundance. Given that there will also havebeen a bed-load transport of Ti, the net sediment component of Ti willhave been higher than that for the filtered load.

6. Discussion

Our results indicate that for a rural site in mid-Wales, [Ti] inrainfall and cloud-water was relatively low compared with river

TT

MM

M

MM

1 2 3 4 5 6 70

30

60

90

120

[B]

(µg/

l)

TT

MM M

M

M

1 2 3 4 5 6 70

0.5

1

1.5

[SR

P]

(mg/

l)

T

T

M MM

M

M

1 2 3 4 5 6 701234567

[Ti]

(µg

/l)

Site Number (Downstream sequence)

M

MM

M

M

T

T

1 2 3 4 5 6 70

5

10

15

[SR

P]/

[B]

(µg/

µg)

[B], [SRP], [SRP]/[B], [Ti]along the Douglas and its tributaries

[SRP]/[B]

[Ti]

[B]

[SRP]

Fig. 5. Variations in average ‘dissolved’ (b0.45 μm) Ti concentrations ([Ti]), boron, SRPand SRP/boron ratio for sites on the Douglas. The site numbers are listed in Table 1 andthey are given in a downstream sequence. M and T refer to main-stem and tributary,respectively and site 1 (the Yarrow) is taken as representative of the upper Douglas. Thelines join the main-stem sequence.

DTW

EB DGGHB

0 1 2 3 4 5 6 7

Dis Ti (µg/l)

0

1

2

3

4

5

[Ti]

fra

ctio

nate

d (µ

g/l)

> 1 kDa < 1 kDa

Dis Ti in < & > 1 kDa fractions

< 1 kD fracti

on

> 1 kD fraction

D = Dunsop HB = Hynburn BrookTW = Tarnbrook Wyre EB = Eller Brook

DGG = Douglas Grimshaw Green

Fig. 6. Plots of ‘dissolved’ (b0.45 μm) Ti concentrations [Ti] versus the Ti concentrationin the less than and greater than 1 kDa size fraction.


water. The highest [Ti] in the inputs occurred within cloud water,particularly at low volumes of catch.While there was Ti enrichment inthe cloud water, in terms of depositional flux it contributes far lessthan rainfall. For example, for the upper Severn, the annual volumesupplied to the catchment from cloud water is approximately 33 mmmoorland 140 mm for forest compared with rainfall at around2459 mm (Wilkinson et al., 1997).

For the rivers, the upper Severn had the lowest [Ti] as these sitescorresponded to virtually no direct pollution input from agriculture,urbanisation and industrialisation other than acidic deposition andpollutant trace metal inputs from the atmosphere (Wilkinson et al.,1997; Neal et al., 1997a,b). However, even the atmospheric depositionterm does not seem to have been significant for [Ti] and even clear

Table 5Ultrafiltration results: included in the table is average boron concentration for a comparisonpresent as borate in a truly dissolved form, with N90% in the b1 kDa fraction.

Site B [Ti] [Ti]N1 kDa(μg/l) (μg/l) (μg/l)

Dunsop 8.6 1.36 1.07Tarnbrook Wyre 7.6 1.57 0.94Eller Brook 58.4 5.09 1.73Hyndburn Brook 59.8 2.62 1.47Douglas (GG) 103.0 6.46 1.81

felling of the forest did not lead to a significant increase in [Ti] despitethere being considerable land disturbance and sediment transport.

There were general positive relationships between average [Ti]and the markers of sewage effluent pollution across the range of rivermonitored in this study but clearly there was also some scatter to therelationship. For all the urban/industrial sites and their downstreammonitoring points, [B] and [SRP] levels were similar to those for highlyeffluent impacted British rivers (Neal, 2000), although there wassome reduction in [B] associated with reduced usage in households(Neal et al., 2010a). The occurrence of high [Ti] in the lower reaches ofthe Ribble and Wyre, especially just downstream of STWs, indicatessignificant anthropogenic inputs to the river. However, the strongpositive correlation between average [Ti] with both [SRP] and [SRP]/[B] also indicates that Ti may be retained within the river. This mayresult from aggregation and sedimentation processes and as observedpreviously, this effect may be temporary, with suspension of fine flocswhenwater velocities increase (Boncagni et al., 2009; Domingos et al.,2009; Keller et al., 2009). In the case of the Douglas, the loss of both Tiand SRP in the mid to lower reaches is remarkable but explainable bythe fact that just upstream of these monitoring points, the Douglasenters Worthington lakes. The lake will have increased the waterresidence time and provided the opportunity for additional loss be itby flocculation (Ti) or biological reactivity (SRP).

For individual sites there was no clear correlation with between[Ti] and [SRP] or [SRP]/[B]. This may well reflect the very differentprocesses, mechanisms and rates of retention of SRP in rivers (whichis controlled by biological uptake and physico-chemical sorptionprocesses). It may also reflect Ti retention, which is controlled by arange of electrostatic and steric interactions between particles and

linked to effluent inputs. N.B. for boron, there is no size exclusion unlike Ti: the boron is

[Ti] b1 kDa % [Ti] N1 kDa Ti mass balance recovery (%)(μg/l)

0.29 79 1070.63 60 803.36 34 601.15 56 854.65 28 92


their surface moieties (such as natural organic matter coatings orsurfactant coatings), sediments and organic and inorganic aquaticcolloids (Nowak and Bucheli, 2007). A sewage related source of Tiseems reasonable given that microparticulate and nanoparticulate Tihas been found in STWs and their effluents (Kiser et al., 2009). Indeedfrom our own studies at a STW serving an urban population of c. 300 Kp.e. in south central England, [Ti] concentrations have been observed:as high as 40.5 μg/l (average 30.5 μg/l) in the raw sewage and around1.6 to 3.6 μg/l in the final effluent (average 2.75 μg/l). Within the finaleffluents [B] and [SRP] averaged 192 and 208 μg/l, respectively and the[SRP]/[B] ratio averaged 1.08with corresponding averages for the [Ti]/[B] and [Ti]/[SRP] of 14.3 and 13.2, respectively.

The occurrence of “spikes” of [Ti] even in rural situations and thepropagation along the river network indicates the input of a more“diffuse” source of Ti. Colloidal transport from soils would be expectedunder wet conditions when surface runoff would be especially high.However, as with sediments, it might well be expected that diffusecolloidal concentrations would not simply match flows. This isbecause there are issues of supply-exhaustion of sediment stores.Under such circumstances, the highest concentrations might well beexpected on the rising limb of the hydrograph when supply-exhaustion is minimal and dilution potential is low. However,capturing this is difficult based on all but the most intense samplingregimes that capture individual storm events. This feature has beenseen for example in the case of particulate transfers of phosphorusfrom agricultural areas (Jarvie et al., 2002).

There are potential sources to the [Ti] from within the urban/industrial environment with runoff from contaminated land sites andfrom wash-off of hard-standing road and building surfaces. Suchrunoff might also be considered as “diffuse” and there is directevidence of wash-off of synthetic TiO2 nanoparticles from exteriorfacades (Kaegi et al., 2008). The diffuse rural/agricultural and urban/industrial signals may however differ, as there are issues such as rapidrunoff from hard surfaces and differing rates of leaching fromcontaminated land. Also, if reduction in [Ti] from the river results inflocs that are easily resuspended and may be physically disaggregatedduring high and turbulent flows, a complex dynamic for [Ti] is perhapsto be expected in our individual (impacted) rivers. Indeed, suchprocesses must link to the dynamics of the hydrology and as such ourfortnightly sampling programme cannot pick up such detail.

With regard to Ti the first results of the ultrafiltration studyindicate that for the urban/industrial system there is a dominance ofthe b1 kDa fraction. This means that the pollutant component of theTi must primarily exist as a truly dissolved component or asrelatively small macromolecules. The colloidal nature of Ti in naturalwaters is strongly inferred from earlier studies in the Earth andoceanographic sciences. Further, earlier studies on TiO2 solubilityseemed to indicate values far higher than concentrations observedin nature and it has been suggested that some of these earlierstudies used materials that were perhaps not representative of theenvironment (Van Baalen, 1993). We find the observationssurprising and it is not clear what the complexing agent might beto keep Ti in a truly dissolved form as for example, fluoridecomplexes do not seem important (Turner et al., 1981). However,recent studies (Gulley-Stahl et al., 2010) indicate that surfacecomplexation of organic ligands, such as catechol causes dissolutionof TiO2 at pHN5, while dental research indicates corrosion of Ti byorganic acids (Koike and Fujii, 2001; Mabilleau et al., 2006). Giventhe differences between our and earlier studies, there may well alsobe an inference for colloidal formation and instability. Catechol andits derivatives are a common class of chemicals from both naturaland anthropogenic sources, while sewage effluents contain a widevariety to organic compounds and biological activity may add to thecorrosion potential. However, new work is required to address thisissue and there is clearly a shortfall in understanding in this area. Atfirst sight, this might not seem relevant as complexation might

stabilise [Ti] in the river, whereas [Ti] removal is observeddownstream of the STWs However, with dilution in the river, thelowering of the concentration of Ti and the complexing ligandwould in turn shift the equilibrium towards titanium in non-complexed form and the ability for the freed Ti to produce colloidsand precipitate out of the water column. Alternatively, theultrafiltration results may point to a large anthropogenic componentof very fine size in [Ti]. With cross-flow ultrafiltration, the waterflow introduces some turbulence within the water and it may wellbe that colloidal flocs are physically dispersed unlike with standardfiltration. Within this paper we cannot speculate further on this andwe do not have sufficient information on reactive components andequilibrium constants to undertake a thermodynamic exercise on Ticomplexation and solubility. Nonetheless the study flags the need toestablish what forms these small sized components reside in and itcannot be automatically assumed that there is not a truly dissolvedcomponent. Indeed, our results may well point to complexationbeing important. In terms of colloidal material that might bepresent, there may be at least three types: (1) natural colloidsuninfluenced by human activity, (2) industrially manufactured(synthetic) nanoparticles and (3) anthropogenic colloids that areeither (a) directly introduced to the river of (b) pollutant inputsthat have evolved in the water column to generate colloids. Thesame may also be said for the truly dissolved phases.

7. Conclusion

Despite the complexity of the system, the paper presented hereindicates that there is an urban/industrial signal on the [Ti] based oncomparative for the Ribble/Wyre and the upper Severn. Much of thisurban signal comes from or truly dissolved components or very finecolloid that are directly inputted to the river. The nature of the physical,chemical and, perhaps, biological interactions are crypticdue toanumberof differing potential sources, hydrological factors and colloid stabilityrelated influences. Our studyprovides a clue that either colloidal and sub-colloidal Ti or urban/industrially linked Ti complexes are not stablewithin the water column and [Ti] decreases as water is progressivelytransported downstream of effluent sources. However, colloidal forma-tion, complexation, dissolution and colloidal/complexation instabilityhave a process that remains uncertain at present. Clearly there is a needto separate issues of natural colloids, nanoparticles and complexes fromtheir anthropogenic counterparts including industrially manufactured(synthetic), non-synthetic Ti colloids/nanoparticles and possibly anintermediary case where colloidal material forms directly in the watercolumn as effluents and chelating agents dilute. Such a step is far beyondwhat is capable within the current study. However, the work is ofimportance in providing new information on Ti in rivers of contrastingtypology as a backdrop on which to build and focus further work on Ticontaminationofwatercourses. The studyprovides a backgrounddataseton which to build on within the context of a basin wide observatory forlong-term environmental studies.

References

Alvarez PJJ, Colvin V, Lead J, Stone V. Research priorities to advance Eco-responsibleNanotechnology. ACS Nano 2009;3:1616–9.

Andersson K, Dahiqvist R, Turner D, Stolpe B, Larsson T, Ingri J, et al. Colloidal rare earthelements in a boreal river: changing sources and distributions during the springflood. Geochim Cosmochim Acta 2006;70:3261–74.

Ayers JC, Watson EB. Rutile solubility and mobility in supercritical aqueous fluids.Contrib Mineral Petrol 1993;114:321–30.

Boncagni NT, Warner E, Curran T, Ren JH, De Cortalezza MMF. Environ Sci Technol2009;43:7699–705.

Christophersen N, Neal C, Hooper RP, Vogt RD, Andersen S. Modelling stream waterchemistry as a mixture of soil water end-members — a step towards second-generation acidification models. J Hydrol 1990;116:307–20.

Davies HN, Neal C. Estimating nutrient concentrations from catchment characteristics.Hydrol Earth Syst Sci 2007;11:550–8.

Domingos RF, Tufenkji N, Wilkinson KJ. Aggegation of titanium dioxide nanoparticles:role of fluvic acids. Environ Sci Technol 2009;43:1282–6.


Federici G, Shaw BJ, Handy RD. Toxicity of titanium dioxide nanoparticles to rainbowtrout (Oncorhynchus mykiss): gill injury, oxidative stress, and other physiologicaleffects. Aquat Toxicol 2007;84:415–30.

Greenwood NN, Earnshaw A. Chemistry of the Elements. Amsterdam: Elsevier; 2005.1341 pp.

Gulley-Stahl H, Hogan PA, Schmidt WL, Wall SJ, Buhrlage A, Bullen HA. Surfacecomplexation of catechol to metal oxides: an ATR-FTIR adsorption and dissolutionstudy. Environ Sci Technol 2010;44:4116–21.

Handy RD, Owen R, Valsami-Jones E. The ecotoxicology of nanoparticles andnanaomaterials: current status, knowledge gaps, challenges, and future needs.Ecotoxicology 2008;17:315–25.

Jarvie HP, Neal C, Williams RJ, Neal M, Wickham HD, Hill LK, et al. Phosphorus sources,speciation and dynamics in a lowland eutrophic Chalk river: the River Kennet, UK.Sci Total Environ 2002;282/283:175–203.

Jarvie HP, Neal C, Jurgens MD, Sutton EJ, Neal M, Wickham HD, et al. Within-rivernutrient processing in Chalk streams: the Pang and Lambourn, UK. J Hydrol2006;330:101–25.

Ju-Nam Y, Lead JR. Manufactured nanoparticles: an overview of their chemistry,interactions and potential environmental implications. Sci Total Environ 2008;400:396–414.

Kaegi R, Ulrich A, Sinnet B, Vonbank R, Wichser A, Zuleeg S, et al. Synthetic TiO2nanoparticle emission from exterior facades into the aquatic environment. EnvironPollut 2008;156:233–9.

Keller AA, Wang HT, Zhou DX, Lenihan HS, Cherr G, Cardinale BJ, et al. Stability andaggregation of metal oxide nanoparticles in natural aqueous matrices. Environ SciTechnol 2009;44:1962–7.

Kennedy VC, Zelweger GW, Jones BF. Filter pore size effects on the analysis of Al, Fe, Mnand Ti in water. Water Resour Res 1974;10:785–90.

Kiser MA, Westerhoff P, Benn T, Wang Y, Perez-Rivera J, Hristovski K. Titaniumnanomaterial removal and release from wastewater treatment plants. Environ SciTechnol 2009;43:6757–63.

KoikeM, Fujii H. In vitro assessment of corrosive properties of titanium as a biomaterial.J Oral Rehabil 2001;28:540–8.

Lead JR, Wilkinson KJ. Aquatic colloids and nanoparticles: current knowledge andfuture trends. Environ Chem 2006;3:159–71.

Liu R, Lead JR. Partial validation of cross flow ultrafiltration by atomic force microscopy.Anal Chem 2006;78:8105–12.

Mabilleau G, Bourdon S, Joly-Guillou ML, Filmon R, Baslé MF, Chappard D. Influence offluoride, hydrogen peroxide and lactic acid on the corrosion resistance ofcommercially pure titanium. Acta Biomater 2006;2:121–9.

Neal C. The fractionation of the elements in river waters with respect to the continentalcrust: a UK perspective based on a river enrichment factor approach. Hydrol EarthSyst Sci 2000;4:499–509.

Neal C, Kirchner JW. Sodium and chloride levels in rainfall, mist, streamwater andgroundwater at the Plynlimon catchments, mid-Wales: inferences on hydrologicaland chemical controls. Hydrol Earth Syst Sci 2000;4:295–310.

Neal C, Robson AJ. A summary of river water quality data collected within the LandOcean Interaction Study: core data for Eastern UK rivers draining to the North Sea.Sci Total Environ 2000;251/252:585–665.

Neal C, Smith CJ, Jeffery HA, Jarvie HP, Robson AJ. Trace element concentrations inthe major rivers entering the Humber estuary. N E England J Hydrol 1996;182:37–64.

Neal C, Robson AJ, Jeffery HA, Harrow ML, Neal M, Smith CJ, et al. Trace element inter-relationships for the Humber rivers: inferences for hydrological and chemicalcontrols. Sci Total Environ 1997a;194/195:321–43.

Neal C, Wilkinson RJ, Neal M, Harrow ML, Wickham H, Hill L, et al. The hydrochemistryof the headwaters of the River Severn, Plynlimon, Wales. Hydrol Earth Syst Sci1997b;1:583–618.

Neal C, Reynolds B, Neal M, Hughes S, Wickham H, Hill L, et al. Soluble reactivephosphorus levels in rainfall, cloud water, throughfall, stemflow, soil waters,stream waters and groundwaters for the Upper River Severn area, Plynlimon, MidWales. Sci Total Environ 2003;314–316:99-120.

Neal C, Williams RJ, Bowes MJ, Harrass MC, Neal M, Rowland P, et al. Decreasing boronconcentrations in UK rivers: insights into reductions in detergent formulationssince the 1990s and within-catchment storage issues. Sci Total Environ 2010a;408:1315–30.

Neal C, Jarvie HP, Williams RJ, Love A, Neal M,Wickham H, et al. Declines in phosphorusconcentration in the upper River Thames (UK): links to sewage effluent cleanupand extended end member mixing analysis. Sci Total Environ 2010b;408:1315–30.

Nowak B, Bucheli TD. Occurrence, behaviour and effects of nanoparticles in theenvironment. Environ Pollut 2007;150:5-22.

Schmidt J, Vogelsberger W. Aqueous long-term solubility of titania nanoparticles andtitanium (IV) hydrolysis in a sodium chloride system studied by adsorptionstripping voltammetry. J Solution Chem 2009;38:1267–82.

Skrabal SA. Distribution of dissolved titanium in Chesapeake Bay and the Amazon RiverEstuary. Geochim Cosmochim Acta 1995;59:2449–58.

Skrabal SA, Terry CM. Distribution of dissolved titanium in pore waters of estuarine andcoastal marine sediments. Mar Chem 2002;77:109–22.

Skrabal SA, UllmanWJ, Luther III GW. Estuarine distributions of dissolved titanium. MarChem 1992;37:83-103.

Stolpe B, Hassellov M. Changes in size distribution of fresh water colloidal matter andassociated elements on mixing with seawater. Geochim Cosmochim Acta 2007;71:3292–301.

Stolpe B, Guo LD, Shiller AM, Hassellov M. Size and composition of colloidal organicmatter and trace elements in the Mississippi River, Pearl River and the northernGulf of Mexico, as characterised by flow-field fractionation. Mar Chem 2010;118:119–28.

Turner DR, Whitfield M, Dickson AG. The equilibrium speciation of dissolvedcomponents in freshwater and seawater at 25 °C and 1 atm pressure. GeochimCosmochim Acta 1981;45:855–81.

Valzeboer I, Hendriks AJ, Ragas AMJ, Van de Meent D. Aquatic ecotoxicity tests of somenanomaterials. Environ Toxicol Chem 2008;27:1942–7.

Van Baalen MR. Titanium mobility in metamorphic systems — a review. Chem Geol1993;110:233–49.

Wiggington NS, Haus KL, Holchella Jr MF. Aquatic environmental nanoparticles. JEnviron Monit 2007;9:1306–16.

Wilding A, Liu RX, Zhou JL. Dynamic behaviour of river colloidal and dissolved organicmatter through cross-flow ultrafiltration system. J Colloid Interface Sci 2005;287:152–8.

Wilkinson J, Reynolds B, Neal C, Hill S, Neal M, Harrow M. Major, minor and traceelement composition of mist and rainwater at Plynlimon, mid-Wales. Hydrol EarthSyst Sci 1997;1:557–70.

Titanium in UK rural, agricultural and urban/industrial rivers: Geogenic and anthropogenic colloidal/sub-colloidal sources and the significance of within-river retention

Documents