Journal of Hydrology 527 (2015) 1021–1033

Contents lists available at ScienceDirect

Journal of Hydrology

journal homepage: www.elsevier .com/ locate / jhydrol

Baseflow dynamics: Multi-tracer surveys to assess variable groundwatercontributions to montane streams under low flows

http://dx.doi.org/10.1016/j.jhydrol.2015.05.0190022-1694/� 2015 Elsevier B.V. All rights reserved.

⇑ Corresponding author. Tel.: +44 (0)1224 273696.E-mail address: m.blumstock@abdn.ac.uk (M. Blumstock).

1 Tel.: +44 (0)1224 273696.

M. Blumstock a,⇑, D. Tetzlaff a,1, I.A. Malcolm b, G. Nuetzmann c, C. Soulsby a,1

a Northern Rivers Institute, School of Geosciences, University of Aberdeen, AB24 3UF, United Kingdomb Marine Scotland Science, Freshwater Fisheries Laboratory, Faskally, Pitlochry PH16 5LB, United Kingdomc Leibniz Institute of Freshwater Ecology and Inland Fisheries, D-12587 Berlin, Germany

a r t i c l e i n f o

Article history:Received 12 December 2014Received in revised form 31 March 2015Accepted 8 May 2015Available online 1 June 2015This manuscript was handled by Peter K.Kitanidis, Editor-in-Chief, with theassistance of Massimo Rolle, AssociateEditor

Keywords:Base flowLow flowDroughtsGroundwaterTracersIsotopes

s u m m a r y

We monitored changing groundwater-surface water interactions during a drought with a 10 year returnperiod in a 3.2 km2 catchment in the Scottish Highlands. The montane catchment is underlain by graniteand metasediments and has extensive cover of diverse drift deposits (70%), which are up to 40 m deep.Flat valley bottom areas fringing the stream channel are characterised by deep peat soil (0.5–4 m deep)which covers about 20% of the catchment and receive drainage from upslope areas. The drought resultedin small declines in soil moisture and groundwater levels in the valley bottom wetlands, but marked,rapid declines on steeper upland slopes. These coincided with gradual decreases in discharge; however,the chemical and isotopic composition of reduced stream flows showed both temporal and spatialvariation. Synoptic hydrogeochemical surveys were carried out on four occasions as flows declined.Each survey repeated sampling of 26 sites along the 3 km long stream network. Samples were analysedfor major anions, cations and water isotopes. Initial surveys just after the last winter rain showedrelatively homogenous stream chemistry, consistent with dominant near-surface drainage from acidicriparian peat soils. Stream chemistry became increasingly enriched with weathering-derived solutes(e.g. alkalinity, Ca2+, Mg2+, etc.) as flows declined and groundwater dominance of flow increased.However, these changes showed marked spatial variability implying geochemical differences in the bed-rock geology and the distribution of storage in drift deposits. Temporal dynamics inferred heterogeneousmontane groundwater bodies contributed to flows differentially during the recession. Isotope data indi-cated that in places the stream was also influenced by evaporative losses from the surface of the peatsoils. The largest sources of groundwater appear to be located in the drift in the lower catchment wherethe most marked increase in weathering-derived ions occurred, and depleted, non-fractionated isotopesignatures implied deeper inflows.

� 2015 Elsevier B.V. All rights reserved.

1. Introduction

Groundwater is often the sole source of river water during lowflow conditions; an issue that is particularly critical in montaneheadwaters which sustain downstream water supplies and providemany other ecosystem services (Frisbee et al., 2011; Gleeson et al.,2012; Batlle-Aguilar et al., 2014). The geology in montane areas isusually complex, with a high degree of heterogeneity in bedrocktypes and fracture distributions. Additionally, in areas affected byglaciation, diverse drift deposits vary in size and aquifer properties.These can affect spatial patterns of groundwater recharge, storage,and contributions to stream flow (Soulsby et al., 2004). Such

groundwater stores can also have contrasting temporal dynamicsas small patchily distributed aquifers in drift deposits may drainrelatively quickly and become disconnected. Thus, the quantityand quality of groundwater contributions to stream flow may behighly variable and these are usually poorly understood. This lim-ited understanding is an increasing cause for concern given thatclimate change has far reaching implications for many montaneareas, including reduced snow influences, higher evaporation andincreased seasonality of precipitation (Orr and Carling, 2006;Murphy et al., 2009; Kay et al., 2014). Such factors have thepotential to affect groundwater recharge and storage and thus tocompromise downstream water use.

Investigation of montane groundwater has many challenges:high levels of heterogeneity in the subsurface are difficult to char-acterise by borehole installation, which is also expensive and often

1022 M. Blumstock et al. / Journal of Hydrology 527 (2015) 1021–1033

impractical in high altitude terrain (Gabrielli and McDonnell,2011). Other more spatially integrative techniques such as theuse of geophysics are also logistically problematic in mountains,especially if boreholes are not drilled to ‘‘ground truth’’ interpreta-tions (Parsekian et al., 2015). However, in previous studies envi-ronmental tracers have been used as tools for identifyinggroundwater contributions to stream flow and the dynamics ofhow these change under different flow conditions (Haria et al.,2013; Liu et al., 2013; Batlle-Aguilar et al., 2014; Bertrand et al.,2013; Shaw et al., 2014). This is an approach that can be particu-larly useful in complex montane areas where geological hetero-geneity would be difficult to characterise directly.

Under low flows, when stream water may be entirely suppliedby groundwater reservoirs, it will acquire a chemical compositionthat reflects the geochemistry of its storage origin (Zimmer et al.,2012). For example alkalinity, which integrates the cumulativeeffects of weathering, has been used to distinguish between contri-butions of acidic soil water and more alkaline groundwater inmontane streams (Haria and Shand, 2004; Rodgers et al., 2004;Birkel et al., 2011b). Sometimes alkalinity, or some otherweathering-derived tracers, can also be used to infer the residencetime of water, if mineral dissolution is kinetically controlled(Gonzales et al., 2009). In addition, spatial variation ofweathering-derived ions can be used to identify streamflow prove-nance in larger river systems with heterogeneous hydrogeologicalenvironments (Szramek et al., 2011). Thus, synoptic surveys ofstream water chemistry can provide unique insights into the geo-graphical sources of groundwater contributing to low flows, andhow they integrate at larger spatial scales (Soulsby et al., 2004;Froehlich et al., 2008; Heal, 2008; Zimmer et al., 2012; Inamdaret al., 2013; Wirmvem et al., 2014).

Other chemical species, not primarily derived from mineralweathering, can provide additional insights into the hydrologicalprocesses influencing groundwater recharge. Atmospheric aero-sols, especially Cl� which is assumed to be conservative, can indi-cate the timing of recharge and the effects of evaporation (Vuai andTokuyama, 2007; Froehlich et al., 2008). For example, in maritimeregions the dominant source of recharge may occur in winter whensalt-rich precipitation is highest and evaporation rates are low(Hrachowitz et al., 2009). Also, labile species heavily regulated bybiogeochemical interactions (e.g., SO4

2�, NO3�, DOC, DON) can pro-

vide insight into runoff sources from soils and the shallowsub-surface (Weiler and McDonnell, 2006).

The stable isotopes of water (18O and 2H) are also useful tracersof groundwater sources and their influence on stream flows(Darling et al., 2003). These behave conservatively inlow-temperature environments, due to negligible interactionsbetween 18O and 2H in organic and geologic materials. The pro-cesses dictating 18O and 2H composition are phase changes thataffect water above or near the ground surface (evaporation, con-densation, melting) as well as simple mixing at or below theground surface (Leibundgut et al., 2009). Thus, isotopes can beused for estimating groundwater recharge, identification of sourceareas and resolving water flow paths (Koeniger et al., 2009;Leibundgut et al., 2009).

This paper reports the results of a targeted series of synopticsurveys during a prolonged period (4 months) of low flows witha 10 year return period in the Scottish Highlands. The study wasbased in the Bruntland Burn, an intensively studied sub-basin ofthe Girnock experimental catchment. Previous field and modellingwork has identified dominant sources of runoff at thecatchment-scale and the associated landscape controls on theirdynamics and associated transit times (Soulsby et al., 2007;Tetzlaff et al., 2007; Birkel et al., 2011a). More recent work hasintegrated hydrometric and tracer based studies to examine catch-ment storage dynamics (Birkel et al., 2011b). Whilst these studies

have emphasised the importance of riparian wetlands as the dom-inant source of runoff (Tetzlaff et al., 2014), groundwater dischargeprovides both an important source of water to these wetlands (asseepage) throughout most of the year, as well as a direct flux intothe channel network to sustain the lowest flows (Birkel et al.,2011a, 2014). Groundwater also provides a large store of waterthat mixes with conservative isotope tracers to affect the attenua-tion and lag observed in stream waters compared to precipitation(Tetzlaff et al., 2014; Birkel et al., 2015). Nevertheless, little isknown about the heterogeneous nature of groundwater storageat the catchment scale and its influence on the spatial variationin stream water quantity and quality.

The aim of this paper was to use synoptic surveys along theriver channel network of the Bruntland Burn to infer the dynamicsof groundwater contributions as the catchment transitioned fromthe end of a wet spring to a dry summer with declining flows.With the surveys we sought to answer the following questions:

(i) Is there spatial variability in base flow hydrochemistry andisotopic signatures which could be used to infer differencesin groundwater sources and influence?

(ii) Did spatial patterns change over the course of the droughtperiod in a way that could infer the dynamics of groundwa-ter contributions to base flows?

(iii) Can these spatial and temporal patterns be used to betterunderstand groundwater influence on low flow generationat the catchment scale?

2. Site description

The Bruntland Burn is located in the Cairngorms National Park,NE of Scotland, covering an area of 3.2 km2 (Fig. 1a). More detaileddescriptions of the catchment are given elsewhere (Soulsby et al.,2007; Birkel and Tetzlaff, 2010; Birkel et al., 2011a,b). In brief,the main stem (MS) flows approximately 1.2 km in a NE directionfrom the confluence of three headwater (HW) tributaries beforedischarging into the larger Girnock Burn. The headwaters (HW1,0.64 km2; HW2, 0.39 km2; HW3, 0.94 km2) drain from the upperparts of the catchment. The catchment spans an altitude range of248–539 m.a.s.l. The highest point is located at the south westernedge of the catchment at the head of HW3.

The climate is temperate oceanic, with relatively cool summersand winters. Daily mean air and stream temperatures are 7.4 �Cand 6.3 �C, respectively (Hannah et al., 2008). Mean annual rainfallis around 1000 mm, usually with limited seasonality. Half of therain falls in frequent low intensity events of less than 10 mm;<25% of annual rain falls in events with daily totals greater than20 mm. Mean annual potential evapotranspiration is about400 mm (Birkel et al., 2011a) based on a simplified Penman–Monteith equation adjusted to aerodynamic and canopy roughnesscharacteristics of the study region (Dunn and Mackay, 1995). Meanannual discharge is 1.8 mm d�1. High and low flow indices,expressed as Q5 and Q95, are 5.8 mm d�1 and 0.4 mm d�1, respec-tively (for the period June 2011–January 2014).

Due to the glacial history of the landscape, the catchment ischaracterised by a flat, wide valley bottom and steep hillslopes.HW2 and 3 are dominated by low permeability schist, thoughsome calcium-rich meta-sediments are also present in HW3(Fig. 1b, Table 1). In contrast, HW1 and the MS are dominated bylow permeability granite. Around 70% of the catchment is coveredby glacial drift deposits. Recent geophysical surveys have shownthat these are typically �5 m deep on the steeper hillslopes andup to 40 m deep in the valley bottoms (Birkel et al., 2015). Giventhe poor aquifer properties of the bedrock and the extensive driftcover, these drifts have been identified as the likely largest sourcesof groundwater storage (Tetzlaff et al., 2014). They are dominated

Fig. 1. (a) Location map with elevation, weather station, flow recorder and synoptic sampling locations, (b) Bedrock geology classification and synoptic sampling locationsand (c) Soil classification and groundwater logger locations across the Bruntland Burn.

M. Blumstock et al. / Journal of Hydrology 527 (2015) 1021–1033 1023

by a medium-fine textured silty-sand matrix, with abundant clastsranging from pebbles to large boulders. Porosity is 10–20% andprovisional hydraulic conductivity estimates are in the order of10�6 to 10�4 m s�1 (Malcolm et al., 2004).

The steeper hillslopes are characterised by freely draining pod-zols which transition into rankers on the upper slopes where thedrift is absent (Fig. 1c). Under the podzols, the water table depthvaries between ca 1.5 m in drier periods to <0.2 m in the wettestperiods (Tetzlaff et al., 2014). The lower, flatter foot-slopes arepoorly draining. Organic soils (histosols) grading from peaty gleys(0.5 m deep) on the lower slopes to peat (typically 1–2 m deep) inthe flatter valley bottom areas predominate. These organic richsoils are characterised by high storage capacity and are generallysaturated throughout the year. The low local gradient and seepagefrom groundwater in upslope areas maintains a high water table(<0.2 m deep) throughout the year (Geris et al., 2015). Land useis primarily heather (Calluna vulgaris) dominated moorland onthe steeper slopes. In the riparian zone, where peat is deepest,Sphagnum and purple moor grass (Molinia caerulea) dominatedwetlands occur. Further away from the stream coniferous forestis found at higher altitude in some parts of the catchment. This ismainly natural Scots Pine (Pinus sylvestris) with some birch(Betula spp.), though some areas of coniferous plantation are alsopresent.

The three sub-catchments have distinct soil and vegetationcharacteristics (Table 1). HW1 is mainly south facing. It has anextensive raised peat bog dominated by Sphagnum species in itslow lying area which covers 17% of the catchment. The steeperslopes are heather dominated with some Scots Pine and birch.HW2 is steeper and dominated by heather vegetation and podzolicsoils, with peat present only in the lower reach (5% of the area). Thelandscape characteristics of HW3 are similar to the lower catch-ment fringing the main stem, being dominated by histosols (peatsand peaty gleys cover 21% of the sub-catchment) in the valley bot-toms with heather on the steeper slopes where podzols grade intothe rankers. Along the main stem, quasi-permanently saturatedpeats dominate the valley bottom, with podzols on most steepslopes. The southerly facing slopes on the northern edge are dom-inated by scree, though there is still a reasonable coverage of ScotsPine. The low tree cover reflects historic forest clearance and highdensities of red deer (Cervus elaphus) which average 21 animals perkm2 and create high grazing pressures which inhibit regeneration(Birkel et al., 2013).

3. Data and methods

Synoptic stream water sampling along the three headwatersand the main stem (Fig. 1a) was carried out between 18th of

Table 1Sub-catchment characteristics for the three sampled headwaters (HW), the mainstem and the Bruntland Burn outlet.

HW1

HW2

HW3

MainStem

BruntlandBurn

Area (km2) 0.64 0.39 0.94 1.2 3.17Slope (�) 15 15 14 13 13Soil type (%) Brown Ranker 45.1 70.3 40.1 2.8 30.7

Peaty Podzol 18.9 24.6 38.8 45.6 35.6Peat 15.7 0.9 2.8 14.4 9.6Peaty Gley 0.8 4.2 18.3 15.4 11.9Peaty Ranker 19.5 21.9 12.2

Geologytype (%)

Granite 61.1 1.6 3.9 83.6 45.3

Psammite &semipelite

38.7 94.1 74.1 14.5 46.9

Biotite Granite 0.2 4.3 0.2 0.6Quartzite andpsammite

12.8 0.4 4.0

Calcsilicate-rock &psammite

8.8 2.6

Metalimestone 0.1 0.02PorphyriticMicrogranite

1.5 0.6

Vegetationtype (%)

Moorland 36.9 53.9 78.7 58.9 59.7

Rock/Scree 24.8 43.5 20.9 16.1 22.7Blanket Bog 13.3 0.2 0.4 8.2 5.9Woodland 25.0 2.4 15.2 11.1Improved Land 1.2 0.5Grassland 0.4 0.2

1024 M. Blumstock et al. / Journal of Hydrology 527 (2015) 1021–1033

April and 4th of September 2013. Samples were collected atapproximately 100 m intervals, with the position located using aGPS to ensure accurate repeat sampling. The first out of four sur-veys was carried out during higher flow conditions on the reces-sion limb of a hydrograph on the 18th of April 2013 (sum ofdischarge on that day Q1,Sum = 4.6 mm) followed by three subse-quent surveys (10th June, 16th August, and 4th September 2013)with successively decreasing flows during a prolonged dry periodin summer (Fig. 2). Four sites were sampled in HW1, six alongHW2, ten along HW3 and six along the main stem. As part of rou-tine monitoring in the catchment, discharge, precipitation andgroundwater levels have been measured continuously since June2011 (Fig. 1). Stream level was measured at the catchment outletat 15 min intervals using an Odyssey capacitance logger and dis-charge was calculated using a rating equation. A Davis tippingbucket rain gauge (0.2 mm resolution) was located nearby, con-nected to an Odyssey data logger recording at 15 min intervals.Furthermore, a groundwater monitoring network was set up, con-tinuously recording water table elevation at 14 different locationsusing Odyssey capacitance loggers (length = 1.5–2 m; approx.0.8 cm resolution, Fig. 1c). Depth to groundwater table was relatedto the soil surface, where negative values indicate water tablebelow soil surface and positive values would indicate runoff gener-ation. Time periods where groundwater levels have been recordedwere used to estimate time periods where groundwater levels fellbelow the maximum recordable water level depth. This has beendone by calculating the rate of groundwater level rise per time(cm 15 min�1) for each groundwater logger for several rainfallevents (the same was done for recession period). These estimatedrates (individual rates for each groundwater logger) were thenapplied to periods of missing groundwater level measurements.

Samples were analysed for major ions after filtration using0.45 lm filters. Samples were stored in 250 mL HDPE bottles andrefrigerated until analysis at the Marine Scotland Science,Freshwater Fisheries Laboratory, in Pitlochry. The analysis of anion

(Cl�, NO3�, SO4

2�, NO2�, Br�) and major cation (Na+, NH4

+, K+, Mg2+,Ca2+) concentrations was performed via ion chromatography(Dionex ICS-1000) and inductively coupled-plasma mass spec-troscopy (Perkin Elmer Optima 5300DV), respectively, underUKAS accreditation. Alkalinity was determined by Gran titration.Samples for d 2H and d 18O were stored in 8 mL glass vials and anal-ysed with a Los Gatos DLT-100 laser spectrometer following a stan-dard measurement protocol. Isotope concentrations wereexpressed in d units (‰) after calibration using Vienna StandardMean Ocean Water (VSMOW). Analytical precision of d2H andd18O determinations was approximately ±0.4‰ and ±0.1%,respectively.

Multivariate statistical analysis was performed to identify therelative differences between the three HWs and MS based on thehydrochemical and isotopic characteristics of the 26 locations.Principal Components Analysis (PCA) was performed on data col-lected during April (high flow) and September 2013 (after pro-longed dry period), reflecting variable flow conditions. Theresulting principal components were used to identify similaritiesand differences and to identify groups of similar hydrochemicalcharacteristics on the basis of high covariance. The PCA plots pro-vide integrated information of the spatial variability of the streamchemistry data, as well as information on how these spatial pat-terns change temporally, by comparing the PCAs for the differentsampling dates.

4. Results

4.1. Hydrometric conditions

The first synoptic survey was conducted at the end of a wet per-iod on 18th April 2013. Both the 7 and 14 day antecedent precipi-tation was high (P7,Sum = 108.4 mm and P14,Sum = 113.5 mm), withmost precipitation occurring in the 7 days preceding sampling.Discharge was double the annual mean (Table 2). During the fol-lowing three surveys, summer drought caused the catchment tobecome increasingly dry as periods without rainfall increased.The longest time period between precipitation events was 18 days(Fig. 2a). The 10th June survey coincided with the start of a drierperiod (with antecedent precipitation of P7,Sum = 0.1 mm andP14,Sum = 15.1 mm, respectively) with discharge just 20% of thatobserved in April (Table 2, Fig. 2a). The 16th August samples werecollected shortly after a wetter spell in the middle of the dry periodstarting in late July (P7,Sum = 9.3 mm, P14,Sum = 16.8 mm) which wasreflected in the increased groundwater levels (Fig. 2a and Table 2).Nevertheless, flows rapidly receded again and by 16th of Augustdischarge (Q1,Sum = 0.5 mm) was about 30% lower than it had beenin June. The final survey on 4th September was carried out whenthe drought reached its maximum (P7,Sum = 0.4 mm,P14,Sum = 10.3 mm), and discharge was only 7% of that recorded inApril and well below the Q95.

Groundwater fluctuations largely reflected the precipitationand discharge patterns (Fig. 2). Within the riparian zone, watertables remained close to the soil surface and responded to any pre-cipitation event with continued seepage into the stream (wellsP1-P6 in Table 2; see Fig. 1c for locations). Even during the 10 yeardrought the water table remained within the upper 20 cm of thesurface at 6 out of 14 sites. Groundwater levels in the upper hill-slopes were much more dynamic and showed strong correlationwith precipitation and discharge variations (example shown forPG2 in Fig. 2b). During the wetter conditions at the beginning ofthe study period the water levels remained within 20 cm of theground surface, followed by a rapid drop in June and finally reach-ing a maximum depth of >60 cm below surface as drought pro-gressed. Changes in the water tables in the PP1-3 (Table 1) wells

Fig. 2. (a) Precipitation and discharge and (b) Groundwater level at location PG2 during sampling period. Groundwater levels are measured as depth below soil surface.Missing data have been interpolated from available data using rates of water table fluctuation from the same logger (black dashed lines). Blue (wet) and red (dry) dashed linesindicate different wetness states of the catchment. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Table 2Overview of antecedent conditions for each sampling date: cumulative precedingprecipitation and discharge during the sampling day (Q1,Sum, P1,Sum) and over a 7(Q7,Sum, P7,Sum) and 14 day (Q14,Sum, P14,Sum) period before sampling. Recorded streamtemperatures and the average groundwater level as depth below soil surface duringsampling day (groundwater logger locations P1-6, PG1-5 and PP1-3 are shown inFig. 1c) across the Bruntland Burn.

18th April‘13

10th June‘13

16th August‘13

4thSeptember ‘13

Q1,Sum (mm) 4.6 0.8 0.5 0.3P1,Sum (mm) 0 0 0 0Q7,Sum (mm) 31.8 6.6 4.6 2.5P7,Sum (mm) 108.4 0.14 9.3 0.4Q14,Sum (mm) 46.2 16.4 11.4 5.7P14,Sum (mm) 113.5 15.1 16.8 10.3TMean (�C) 00:00–

24:007.3 14 13.7 12

TMean (�C) 10:00–20:00

9 16 14.7 12.3

P1 (cm) �5.7 �5.3 �5.6 �5.9P2 (cm) �9.3 �14.8 �15.4 �18.9P3 (cm) � �35.2 �37.1 �45.1P4 (cm) � �14.1 �16.4 �21.8P5 (cm) 13.1 �2.8 1.1 �1.5P6 (cm) 3.2 2.4 �4.4 �14.7PG1 (cm) � �22.1 �30.3 �44.8PG2 (cm) �18.5 �29.6 �38.2 �60.6PG3 (cm) 3.1 �1.4 – –PG4 (cm) 3.4 �0.6 �11.0 �16.9PG5 (cm) – 1.1 �5.7 �16.2PP1 (cm) 0.5 �46.6 �53.3 �88.0PP2 (cm) �15.9 �85.2 �84.6 �103.6PP3 (cm) �22.7 �35.5 �61.0

M. Blumstock et al. / Journal of Hydrology 527 (2015) 1021–1033 1025

on the upper hillslopes were even more pronounced falling tobelow 1 m depth.

4.2. Spatial and temporal variability in major ions

Following the wet period in April 2013, all major ions showedrelatively homogeneous concentrations across the catchment.Table 3 summarizes the stream chemistry at the lowest point ofeach catchment headwater tributary and at the catchment outletat the end of the main stem. However, as the dry period progressedthe spatial variability of most weathering-derived ions becamemore pronounced, as was evident in the increased concentrations(Table 3). For example, Ca2+:Cl� equivalence ratios were typically>0.7 during low flows, compared to 0.16 in precipitation, indicatinga clear weathering origin. The ratio was always highest at site 26 atthe catchment outlet.

The spatio-temporal variability in catchment hydrochemistrywas summarised as a series of Piper diagrams in Fig. 3. Along themain stem and at the catchment outlet (Fig. 3d) there was a shiftfrom less variable, Na+ and Cl� dominated waters in April to moredynamic alkalinity-dominated waters in the summer dry periodsamples. Increased Ca2+ concentrations affected buffering with ageneral doubling in concentrations, though Na+ and Mg2+ concen-trations increased by around 70%. There was also a suggestion ofgreater Ca2+ and Mg2+ influence in June. Similar patterns wereobserved in the headwater catchments (Fig. 3a–c). Somewhat sur-prisingly, concentrations in the August sampling tended to belower than June despite the lower flows (Table 3).

Spatial variation in alkalinity was more effectively visualised foreach of the sampling periods by plotting downstream changes(Fig. 4). This again showed the relatively homogeneous alkalinityof stream waters in the April sampling period (Fig. 4a), with con-centrations close to 100 leq l�1 in all HW tributaries and alongthe main stem of the river. In June, under lower flows, greater spa-tial differentiation became more apparent. HW1 had the highestalkalinities of the tributary streams, which decreased slightlydownstream. Concentrations were lower in HW2 and 3, but werestill double those observed in April. Along the main stem of theriver, concentrations increased substantially (by around 25%)between sites 22 and 23 in the lower catchment. Here, concentra-tions were increased by a factor of four relative to April. Spatialpatterns were broadly similar in August, but concentrations werelower than those observed in June. Additionally, concentrationsnow increased with distance downstream in HW1. Finally, spatialvariation was preserved in September, and concentrationsincreased at most sites over previous months, though the increasein the lower catchment was more pronounced compared to June.

The individual base cations contributing to alkalinity showedsimilar temporal and spatial variability (Fig. 5). Sodium was gener-ally the dominant base cation. It exhibited limited spatial variationin the April sampling period, though HW1 had the highest levelsand concentrations increased downstream along the main stem.In September Na+ concentrations increased across the catchment,though HW1 remained highest, with a downstream increase, incontrast to HW2 and HW3, which were constant, apart from theirlower reaches which exhibited a slight increase. However, the lar-gest increases were evident in the lower catchment along the mainstem of the stream.

For divalent cations, Ca2+ concentrations were uniformly low inthe April samples, although the highest concentrations wereobserved in HW2, whilst lowest concentrations were observed inthe lower reaches of HW3. However, there was no downstream

Table 3Major ions (leq l�1) for the lowest sub-catchment points and the Bruntland Burn outlet (locations are shown in Fig. 1a).

Site Sampling location Na+ NH4+ K+ Mg2+ Ca2+ Alkalinity as HCO3

� Cl� NO2� Br� NO3

� SO42�

LOD 1 1 1 2 5 2 0.5 0.5 1 1

April ‘13HW1 4 226 0 17 77 119 103 266 0 0 6 64HW2 10 124 1 10 46 104 74 163 0 0 1 46HW3 20 186 0 17 72 94 82 236 0 0 0 51BB Outlet 26 216 0 18 77 115 117 250 0 0 0 59

June ‘13HW1 4 253 0 11 92 166 272 214 0 0 0 36HW2 10 206 0 10 74 161 199 194 0 0 0 58HW3 20 204 0 11 96 154 242 175 0 0 1 47BB Outlet 26 266 0 16 127 224 354 210 0 0 0 68

August ‘13HW1 4 220 0 8 73 140 197 181 0 1 0 63HW2 10 217 0 10 66 140 163 190 0 0 0 80HW3 20 212 0 9 78 123 196 164 0 0 0 62BB Outlet 26 301 0 17 103 180 300 212 0 0 0 89

September ‘13HW1 4 295 5 14 97 183 282 226 0 0 6 75HW2 10 255 0 13 84 165 250 200 1 1 0 67HW3 20 228 0 12 86 138 232 167 0 0 2 63BB Outlet 26 378 0 22 131 232 392 251 0 0 3 117

Fig. 3. Piper diagrams of major ion samples for the different sub-catchments: (a) HW1, (b) HW2, (c) HW3 and (d) MS. Lower ternary diagrams show major cation (left) andanion (right) composition as relative abundance in % based on charge equivalents. Upper diamond in each diagram is a composite plot, aggregating both cation and anions.Colours represent different sampling dates. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

1026 M. Blumstock et al. / Journal of Hydrology 527 (2015) 1021–1033

increase along the main stem. In contrast, HW1 had the highestconcentrations in September, though HW2 remained higher thanHW3. However, downstream of the confluence of the HWs,

concentrations increased along the main stem, particularlybetween sample sites 22 and 23. For Mg2+, concentrations werelow (accounting for 10–20% of the cationic composition), with little

Fig. 4. Spatial distribution of alkalinity concentrations for (a) April, (b) June, (c)August and (d) September 2013.

Fig. 5. Spatial distribution of base cation concentrations for (a) April and (b)September 2013.

M. Blumstock et al. / Journal of Hydrology 527 (2015) 1021–1033 1027

spatial variation in April, though levels were slightly higher inHW1. In September, concentrations were highest in HW1, withHW2 being the lowest. Again, in the lower catchment, a markedincrease between sample points 22 and 23 was evident, with aslight decline further downstream.

In the case of anions (Fig. 6), Cl� was uniformly high in April,but exhibited spatial differences between the headwaters underlower flows in the order HW1 > HW2 > HW3. Over summer, con-centrations declined in the June and August surveys, but spatialdifferences in the HW sites remained similar. In common withthe base cations, Cl� concentrations increased markedly down-stream of the confluence between sites 22 and 23. Sulphate wasgenerally the third most abundant anion throughout theBruntland system and equivalence ratios to Cl� were generallyabove the ratio of 0.3 characteristic of precipitation. In the wetterApril, sampling concentrations were highest in HW1 and HW2compared to HW3 and the main stem sites, though like most ions,there was limited spatial variation. Spatial patterns became morepronounced in the lower flow samples, as shown in September.Whilst SO4

2� concentrations remained highest in HW1 and HW2,they were lowest in HW3. However, like almost all other ions,SO4

2� showed a marked increase with proximity to the outlet inthe lower catchment.

Nitrate concentrations are generally below detection at mostsites. Though low concentrations were occasionally detected atsome sites. The system is oligotrophic, so N is generally retainedby terrestrial and aquatic vegetation, though the leakage detectedmay reflect inputs from the large red deer populations in the catch-ment which congregate near streams and wetlands in summer fordrinking water and cooling.

4.3. Stable isotope dynamics

Stable isotopes were used to gain additional information onwater sources and their dynamics. Processes regulating the relatived18O and d2H composition are phase changes that affect waterabove or near ground surface (in this case evaporative fractiona-tion) as well as simple mixing. Thus, isotopes can be used for dif-ferentiating groundwater sources and near surface flow paths.The isotopic composition of samples differed in the four surveys(Fig. 7). In April, all sites had depleted values of d18O and d2H withall samples plotting to the left of the Global Meteoric Water Line(GMWL) and Local Meteoric Water Line (LMWL), which wasderived from daily precipitation samples from the catchment out-let (2011–2014, dD = 7.5 ⁄ d18O + 4.0). By June most sites plotted tothe right of the GMWL indicating the effects of evaporative frac-tionation in some sources. This effect was less marked in theAugust samples, but was again evident in September.

Looking at the individual parts of the catchment, sites in HW1showed some effects of fractionation – indicated by a deviationfrom the meteoric water line – in all but the April survey, withlocations 1 and 2 being especially enriched in September(Fig. 7a). However, in HW2 (Fig. 7b) all samples plotted left ofthe LMWL, with relatively few water samples indicating fractiona-tion even in the driest periods. In HW3 (Fig. 7c), the June samplesplotted most noticeably to the right of the LMWL, indicating thestrongest fractionation signals. Occasional samples in August andSeptember were similarly affected, though these were restrictedto the lower sites (e.g., 18, 19 and 20). Along the main stem, thethree summer surveys showed the most marked fractionation inJune and September (Fig. 7d). The degree of fractionationdecreased downstream along the main stem between sites 21 to26, with 25 and 26 showing similar values compared to April.Although the effect was less marked in the August survey, the spa-tial patterns were similar.

4.4. Variance in stream water chemistries and groundwatercontributions

Major ions and isotopes were grouped within a PCA to providean integrated overview of hydrochemical characteristics between

Fig. 6. Spatial distribution of anion concentrations for (a) April and (b) September2013.

1028 M. Blumstock et al. / Journal of Hydrology 527 (2015) 1021–1033

the sampled HWs and the main stem (26 sampling locations,Fig. 8). The PCA helped to identify differences and similaritiesbetween the different sampling sites. During both the wettest(Fig. 8a) and driest (Fig. 8b) conditions 69% of the variance of the26 sampling locations were explained with two principal compo-nents for both analysed surveys. Surprisingly, despite the relativelyhomogenous stream chemistry during the higher flow in April, the

Fig. 7. Isotopic composition in relation to global meteoric waterline (dD = 8 ⁄ d18O + 10)and (d) MS.

HW streams and MS clustered separately in the PCA space, thoughsite 10 was an outlier it still plotted closest to other HW2 sites onthe 2nd principal component (Fig. 9a). Although positions shiftedfor the September survey, the sampling locations were broadlyseparated into three groups primarily in relation to principal com-ponent 1 (PC1): MS (23-26), HW1 (2-4) and HW2/3 (5-9, 11-18)(Fig. 9b). These groups generally formed non-overlapping clusters,indicating clear differentiation of hydrochemistry reflecting thedifferences in groundwater influence on low flow generationacross the catchment. A marked change in stream water concentra-tion occurred after the confluence of the three HWS as additionalgroundwater inflows shifted the PCA position to the left alongPC1 indicating higher concentrations of major ions and moredepleted isotope signatures.

5. Discussion

5.1. Spatial variability in baseflow dynamics

The spatial patterns of stream chemistry were strikingly differ-ent between the first survey, when the catchment was still wet,and the subsequent surveys as the dry period progressed. Thedominant controls for the wet, colder sampling period and thedry, warm period are summarised conceptually in Fig. 9. In the firstwetter survey, there was limited spatial variability most likely as aresult of stream flow being mainly generated by near surface flowpaths in the acidic riparian peat soils (Birkel et al., 2011b). Thesedelivered low alkalinity water to the stream network from the con-nected riparian wetland which covers up to 40% of the catchmentunder wet conditions. This dominance of near-surface flow pathsin wet conditions is characteristic of northern catchments with

and local meteoric waterline (dD = 7.5 ⁄ d18O + 4.0) for (a) HW1, (b) HW2, (c) HW3

Fig. 8. PCA of hydrochemical (Cl�, NO3�, SO4

2�, NO2�, Br�, Na+, NH4

+, K+, Mg2+, Ca2+) and isotopic (2H, 18O) parameters as biplot for (a) April and (b) September 2013. The first twoprincipal components are displayed on the primary axes, explaining 69% of the total variance. The investigated sampling locations are printed as numbers (see Fig. 1a).

Fig. 9. Conceptual scheme showing the origin of major ions along a hillslope transect in the Bruntland Burn during wet (a) and dry (b) conditions. Red arrows indicatehydrological processes affecting mass transport into the stream. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version ofthis article.)

M. Blumstock et al. / Journal of Hydrology 527 (2015) 1021–1033 1029

peaty soils (Tetzlaff et al., 2015). Thus, the influence of groundwa-ter as a proportion of stream flow was low at these times (Tetzlaffet al., 2014). This common source of water was consistent with thesimilarities of the isotope composition in this survey. As with stud-ies elsewhere, most weathering-related ions exhibited strongconcentration-flow relationships and were diluted at higher flowswhen groundwater contributed a lower proportion of stream flows(Froehlich et al., 2008; Goulsbra et al., 2014).

Despite this general uniformity of geochemistry, some subtledifferences were noticeable. Higher Na+ and Cl� concentrationswere evident in HW1. In addition to the direct input of atmo-spheric derived ions by precipitation, additional inputs throughdry deposition of aerosols on forest canopies and through occultdeposition of solutes in cloud droplets is likely to have occurred.These processes are common in the UK uplands where high windspeeds and frequent low cloud can enhance deposition of atmo-spheric solutes by up to 50% (Wilkinson et al., 1997). Forest coverin the Bruntland Burn is generally low; the exception is HW1 with25% forest (Table 1). Increases in these deposition processes, alongwith higher interception and evaporative loss under forests proba-bly contributed to higher concentrations in HW1 compared toHW2 and HW3 (Gustafsson and Franze, 2000; Batlle-Aguilaret al., 2014; Geris et al., 2015). In addition, recent work by Dick

et al. (2015) has shown that the south facing aspect of HW1 resultsin a higher input of solar radiation, higher temperatures and higherevaporation rates, which may contribute further to higherevapo-concentration of conservative solutes like Cl� (Sullivanet al., 2014). Another subtle difference was the high Ca2+ andSO4

2� concentrations in HW2 during wet and dry conditions. Thisheadwater is dominated by free-draining mineral soils and hasthe lowest proportion of quasi-permanently saturated peat soilcoverage (5%), and hence, may be more influenced by deepersources of water with higher concentrations of weathering prod-ucts and limited effects of redox processes in saturated peat soilswhich may affect SO4

2� concentrations (Calmels et al., 2011). Thiscoincided with the isotopic signature in HW2 showing limited evi-dence of evaporative fractionation compared to HW1, 3 and theMS. This is consistent with previous work by Birkel et al. (2011b)and Geris et al. (2015), who showed that averaged d2H signaturesof deep groundwater samples are more depleted than those ofthe stream and riparian wetlands, and that the fractionation inpeat soils is higher than in podzols.

More obvious, consistent spatial patterns emerged in the subse-quent summer surveys as the influence of surface water sourcesdecreased and relative groundwater contribution increased duringthe drier period. This occurred as the riparian wetlands became

1030 M. Blumstock et al. / Journal of Hydrology 527 (2015) 1021–1033

more disconnected from the stream network and deeper ground-water sources became increasingly dominant (Fig. 9b). Of all threeheadwater sites, the highest alkalinities and base cation concentra-tions were evident in HW1, whilst concentration changes in HW2and 3 were less marked. The importance of catchment characteris-tics (including geochemical differences in the drift and bedrock) inaltering hydrochemical signals was apparent. However, the stepchange in concentrations in the main stem of the stream betweensites 22 and 23 was the most striking feature that was also evidentin the patterns of individual base cations. This was consistent withincreasing influence of deeper groundwater inputs in the lowercatchment which have had a longer residence time and as a resulthigher solute loads (Rodgers et al., 2004; Asano et al., 2009;Zimmer et al., 2012). The probability of more mature, deepergroundwater sources near the stream outlet (25, 26) was alsoreflected in the isotopic signature showing lower fractionationeffects and more depleted signatures compared to the samplingsites upstream.

As the catchment transitioned from wet to dry conditions therelative abundance of Cl� and HCO3

� changed in allsub-catchments, while the proportion of SO4

2� to the anion balancewas always �20% throughout the sampling campaign (Fig. 3).Neither Cl� nor SO4

2� have weathering sources in the catchment.Cl� is almost entirely derived from the atmosphere. Whilst SO4

2�

is also derived from the atmosphere, it is generated by the miner-alisation of organic matter in peaty soils, too (Blodau et al., 2007).In the summer surveys, Cl� concentrations were highest in HW1and along the main stem of the river. Both HW1 and the northernpart of the lower catchment are south facing and have the greatestlevels of tree cover. Thus, occult and dry deposition, together withevapo-concentration due to the high levels of solar radiation inputsto south facing slopes, probably accounted for the high concentra-tions. In addition, the greater inputs of deeper groundwater evi-denced by the high alkalinity and cation concentrations along themain stem are most likely to reflect greater winter recharge whenCl� concentrations are highest (Hrachowitz et al., 2009).

Additional SO42� to atmospheric inputs can be mineralised from

organic soil horizons, but will be reduced in anoxic peat waters andgroundwater. Previous work has shown that deeper groundwatersin the Girnock catchment are highly reduced (Malcolm et al., 2004,2005). At the lowest flows when most groundwater was dis-charged directly through the stream bed, inputs of reduced Smay be rapidly oxidized in the stream channel, causing SO4

2� con-centrations to increase in the lower part of the catchment (Devitoand Hill, 1997). This increase is particularly striking in theSeptember survey, when SO4

2� to Cl� ratio is 0.46 at site 26 com-pared to 0.3 in precipitation. However, this inference is tentativeand more detailed sampling would be needed to elucidate the con-trols on the SO4

2� increases.An increased groundwater influence in the lower part of the

main stem was also consistent with the isotopic signatures becom-ing less fractionated in low flows, particularly in the June andSeptember surveys with many sites plotting away from LMWLindicating evaporation effects (Meredith et al., 2009). Effects offractionation were most marked in stream sections with largercoverage of peaty soils (HW3 and the upper part of MS at site22). This presumably reflects the evaporation of water inquasi-permanently saturated riparian wetlands which seeps intothe stream network (Geris et al., 2015). Especially in the lowercatchment (sites 25 and 26) a move toward more depleted, lessfractionated waters implied increased contributions of deepersources of groundwater. This will have been mainly recharged bywinter precipitation which off-sets the effects of fractionation fromnear surface sources at upstream sites. However, it is clear even atlowest flows that water sources are not well-mixed, with contribu-tions from a complex range of flow paths reflecting a

heterogeneous soil-aquifer system with different chemical charac-teristics (Shand et al., 2007; Katsuyama et al., 2009; Pierret et al.,2014).

5.2. Temporal variability in spatial patterns

The main temporal difference in the surveys was between theinitial April survey and the subsequent three drier surveys. InApril, the wet conditions and aftermath of a large Atlantic frontalsystem resulted in high sea salt loadings (Soulsby and Tetzlaff,2008). Thus, in April the catchment was wet with even the podzolson the steeper hillslopes being saturated after the winter with sat-uration overland flow in the valley bottom forming the dominantwater flow path and contributions to stream flow. The connectivityappeared to be highest, with the saturation area at its maximumextent, connecting different hydrological units and thus, compress-ing the spatial variability in chemistry (Fig. 9).

With the progress of drought conditions, increasingly heteroge-neous hydrochemical patterns emerged and the overall streamwater chemistry appeared to have higher concentrations of weath-ering derived ions in all sub-catchments. However, despite the pro-gressively lower flows at each successive sampling survey, it isclear that the August samples once again reflected greaternear-surface soil water contributions to stream flow followingrainfall in late July. However, the evolving low stream flow varia-tion implied spatio-temporal dynamics in the quantity and qualityof groundwater contributions due to lower connectivity betweendifferent hydrological units. This most likely reflects the hetero-geneity in montane groundwater sources having different storagecapacities, different geochemical composition and unknown pref-erential flow pathways that vary in residence times and fluxes tostreams (Haria and Shand, 2004). Furthermore, it seems likely tobe that water can move laterally and vertically through fissuresand bedrock before being forced back toward the surface alongpressure gradients (Payn et al., 2012). These characteristics are cur-rently only inferred indirectly and drilling and geophysical surveysare needed for greater understanding. For example, the presence ofcalc-silicate schists in HW3 might be expected to result in higherconcentrations of Ca2+ and other weathering derived minerals instream waters as the drought progressed (Touhari et al., 2014).However, concentrations were low on all sample occasions andlower than in the other HW streams. According to groundwaterlevel recordings in HW3 (P6, PG5), water tables stayed in the upper16 cm of the soil layer even during low flow conditions. The highproportion of peaty soil seems to act as a buffer zone betweengroundwater coming from the hillslope and the stream (Grabset al., 2012). This coincided with SO4

2� data that remainedunchanged during the whole sampling campaign, whereas HW1,2 and the MS showed increased concentrations with ongoingdrought.

5.3. Aggregation and scaling of baseflow dynamics and groundwatercontributions

Traditional assumptions in tracer hydrology approaches such asend member mixing and their incorporation in catchment-scalerainfall-runoff models assume well mixed sources of runoff in dif-ferent water stores (Brooks et al., 2009; Birkel et al., 2011a).Although such studies have provided invaluable information onbasin-wide processes, most analyses miss the effects of the highlyspatial heterogeneity of landscape units on subsurface water com-position. However, high density synoptic spatial surveys incorpo-rating major ions and isotopes, such as the ones reported here,highlight the geochemically diverse range of water sources andhow they regulate the chemical stream water composition at thecatchment scale.

M. Blumstock et al. / Journal of Hydrology 527 (2015) 1021–1033 1031

Whilst the April survey showed that stream flow at higher flowswas largely generated from a relatively well-mixed source of run-off in the saturated peaty soils of the extended riparian wetland, anunusual prolonged dry spell during the sampling campaign gavenew insights into the dynamic nature of baseflow. This dynamicand the associated heterogeneity is consistent with inputs fromdiverse shallower and deeper groundwater sources with differentchemical and isotopic composition, particularly in the HW tribu-taries. A more uniform groundwater inflow was evident in thelower catchment where extensive glacial drifts fill wide valley bot-toms and provide a large storage capacity. Assuming a catchmentaverage drift thickness of 10 m and a typical porosity of 15% forglacial drift, the storage capacity is approximately 1500 mm, whichis almost equivalent to twice the annual precipitation. Our datashowed that these larger aquifers appeared to become increasinglyimportant to the downstream chemistry and isotope compositionas the drought progressed. The presence of such larger groundwa-ter inputs has significant implications for catchment storage, flowpaths and residence time distribution as well as downstream flows,as shown in other experimental studies in different parts of theworld (Zimmer et al., 2012; Ward et al., 2014).

Generally groundwater level changes are small, particularly inthe saturated peat soils as they continuously receive water fromthe steeper hillslopes. Even during drier periods, the riparian wet-lands act as critical mixing zone altering the isotopic and hydro-chemical signal through precipitation, sorption, transformation ordegradation of substances (Devito and Hill, 1997). Due to the lowhydraulic conductivity of peat soils, together with the wet climaticconditions, longer and more constant water residence times can beassumed compared to catchments in drier and more seasonal cli-mates (Rinaldo et al., 2011).

The significance of such groundwater – surface water exchangefor the ecology and metabolism of streams has been shown else-where (Malcolm et al., 2005; Birkel et al., 2013). These extendedquasi-permanently riparian wetlands act as an interface betweenterrestrial and aquatic ecosystems, controlling hydrological andgeochemical fluxes, moderating surface water temperature andattenuating pollutants by dilution or biodegradation (Bertrandet al., 2013). A better understanding of the connection and massfluxes rates between deeper groundwater systems and surfacewater within a catchment adds fundamental information for themanagement of water resources, the implementation of restora-tion measures and the prediction of catchment response to climatechange (Gardner et al., 2011; Grathwohl et al., 2013; Yidana andKoffie, 2014).

6. Conclusions

To improve our understanding of the heterogeneous nature ofgroundwater storage at the catchment scale and its influence onthe spatial variation in stream water quantity and quality, we car-ried out four systematic synoptic surveys of the stream network ina 3.2 km2 montane catchment over a five month drought period ofincreasingly low flows. We used major ions and isotopes as inte-grated tools to infer the increased influence of groundwater tostream chemistry as the drought progressed. We found that whilstnear surface sources of stream flow, which predominated in wetterconditions, are relatively well-mixed, baseflows are variable andreflect a diverse range of sources. This most likely reflects hetero-geneous groundwater stores in the catchment drift, which havediffering physical and chemical characteristics. In some placesthe groundwater discharges across valley bottom wetlands andappears to be subject to evapo-concentration and fractionation.Elsewhere, particularly in the lower catchment, groundwaterappears to discharge through the stream bed and consists of

deeper, older groundwater sources that reflect winter rechargeand longer times for chemical interaction with minerogenic mate-rial. More extensive drilling programmes, in conjunction with geo-physical exploration, are needed to better characterise andunderstand upland groundwater reservoirs. This may help toimprove our understanding of baseflows and how to predict them.This is likely to become increasingly important under a changingclimate and with increasing demand for water resources.

Acknowledgements

This work is part of the Aqualink project and has been fundedby the Leibniz Association: Joint Initiative for Research andInnovation (PAKT). We would like to thank the NRI staff for theirhelp during field and laboratory work, especially Audrey Innesand Jonathan Dick. The water samples were analysed for majorions at the Marine Scotland Science, Freshwater FisheriesLaboratory, in Pitlochry. We also would like to thank GuillaumeBertrand and two anonymous reviewers for the constructive com-ments which helped improve this manuscript.

References

Asano, Y., Uchida, T., Mimasu, Y., Ohte, N., 2009. Spatial patterns of stream soluteconcentrations in a steep mountainous catchment with a homogeneouslandscape. Water Resour. Res. 45 (10), 1–9. http://dx.doi.org/10.1029/2008WR007466.

Batlle-Aguilar, J., Harrington, G.A., Leblanc, M., Welch, C., Cook, P.G., 2014.Chemistry of groundwater discharge inferred from longitudinal riversampling. Water Resour. Res., 1550–1568 http://dx.doi.org/10.1002/2013WR013591.

Bertrand, G., Siergieiev, D., Ala-Aho, P., Rossi, P.M., 2013. Environmental tracers andindicators bringing together groundwater, surface water and groundwater-dependent ecosystems: importance of scale in choosing relevant tools. Environ.Earth Sci. 72 (3), 813–827. http://dx.doi.org/10.1007/s12665-013-3005-8.

Birkel, C., Tetzlaff, D., 2010. Towards a simple dynamic process conceptualization inrainfall–runoff models using multi-criteria calibration and tracers in temperate,upland catchments. Hydrol. Process. 275 (October 2009), 260–275. http://dx.doi.org/10.1002/hy.

Birkel, C., Soulsby, C., Tetzlaff, D., 2011a. Modelling catchment-scale water storagedynamics: reconciling dynamic storage with tracer-inferred passive storage.Hydrol. Process. 25 (25), 3924–3936. http://dx.doi.org/10.1002/hyp.8201.

Birkel, C., Tetzlaff, D., Dunn, S.M., Soulsby, C., 2011b. Using time domain andgeographic source tracers to conceptualize streamflow generation processes inlumped rainfall-runoff models. Water Resour. Res. 47 (2), 1–15. http://dx.doi.org/10.1029/2010WR009547.

Birkel, C., Soulsby, C., Malcolm, I.A., Tetzlaff, D., 2013. Modelling the dynamics ofmetabolism in montane streams using continuous dissolved oxygenmeasurements. Water Resour. Res. 49, 5260–5275.

Birkel, C., Soulsby, C., Tetzlaff, D., 2014. Developing a consistent process-basedconceptualization of catchment functioning using measurements of internalstate variables. Water Resour. Res. 50, 3481–3501. http://dx.doi.org/10.1002/2013WR014925.

Birkel, C., Soulsby, C., Tetzlaff, D., 2015. Conceptual modelling to investigate theinfluence of tracer dynamics, hydrological connectivity and catchment storageon non-stationary water age estimates. Hydrol. Process. http://dx.doi.org/10.1002/hyp.10414.

Blodau, C., Mayer, B., Peiffer, S., Moore, T.R., 2007. Support for an anaerobic sulfurcycle in two Canadian peatland soils. J. Geophys. Res. 112 (G2), G02004. http://dx.doi.org/10.1029/2006JG000364.

Brooks, R.J., Barnard, H.R., Coulombe, R., McDonnell, J.J., 2009. Ecohydrologicseparation of water between trees and streams in a Mediterranean climate. Nat.Geosci. 3 (2), 100–104. http://dx.doi.org/10.1038/ngeo722.

Calmels, D., Galy, A., Hovius, N., Bickle, M., West, A.J., Chen, M.-C., Chapman, H.,2011. Contribution of deep groundwater to the weathering budget in a rapidlyeroding mountain belt, Taiwan. Earth Planet. Sci. Lett. 303 (1–2), 48–58. http://dx.doi.org/10.1016/j.epsl.2010.12.032.

Darling, W.G., Bath, A.H., Talbot, J.C., 2003. The O & H stable isotopic composition offresh waters in the British Isles. 2. Surface waters and groundwater. Hydrol.Earth Syst. Sci. 7 (2), 183–195.

Devito, K.J., Hill, A.R., 1997. Sulphate dynamics in relation to groundwater – surfacewater interactions in forested wetlands of the Canadian Shield. Hydrol. Process.11, 485–500.

Dick, J., Tetzlaff, D., Soulsby, C., 2015. Small scale spatial and temporal heterogeneityof water temperatures in a northern headwater catchment. Hydrol. Process.http://dx.doi.org/10.1002/hyp.10414.

Dunn, S.M., Mackay, R., 1995. Spatial variation in evapotranspiration and theinfluence of land use on catchment hydrology. J. Hydrol. 171 (1–2), 49–73.http://dx.doi.org/10.1016/0022-1694(95)02733-6.

1032 M. Blumstock et al. / Journal of Hydrology 527 (2015) 1021–1033

Frisbee, M.D., Phillips, F.M., Campbell, A.R., Liu, F., Sanchez, S.A., 2011. Streamflowgeneration in a large, alpine watershed in the southern Rocky Mountains ofColorado: Is streamflow generation simply the aggregation of hillslope runoffresponses? Water Resour. Res. 47 (6), W06512. http://dx.doi.org/10.1029/2010WR009391.

Froehlich, H.L., Breuer, L., Frede, H., Huisman, J.A., Vach, K.B., 2008. Water sourcecharacterization through spatiotemporal patterns of major, minor and traceelement stream concentrations in a complex, mesoscale German. Hydrol.Process. 2043 (2007), 2028–2043. http://dx.doi.org/10.1002/hyp.

Gabrielli, C., McDonnell, J.J., 2011. An inexpensive and portable drill rig for bedrockgroundwater studies in headwater catchments. Hydrol. Process. 26 (4), 622–632. http://dx.doi.org/10.1002/hyp.8212.

Gardner, W.P., Harrington, G.A., Solomon, D.K., Cook, P.G., 2011. Using terrigenic 4He to identify and quantify regional groundwater discharge to streams. WaterResour. Res. 47 (6), W06523. http://dx.doi.org/10.1029/2010WR010276.

Geris, J., Tetzlaff, D., McDonnell, J., Soulsby, C., 2015. The relative role of soil typeand tree cover on water storage and transmission in northern headwatercatchments. Hydrol. Process. http://dx.doi.org/10.1002/hyp.10289.

Gleeson, T., Wada, Y., Bierkens, M.F.P., van Beek, L.P.H., 2012. Water balance ofglobal aquifers revealed by groundwater footprint. Nature 488 (7410), 197–200.http://dx.doi.org/10.1038/nature11295.

Gonzales, A.L., Nonner, J., Heijkers, J., Uhlenbrook, S., 2009. Comparison of differentbase flow separation methods in a lowland catchment. Hydrol. Earth Syst. Sci.,2055–2068

Goulsbra, C., Evans, M., Lindsay, J., 2014. Temporary streams in a peatlandcatchment: pattern, timing, and controls on stream network expansion andcontraction. Earth Surf. Proc. Land. 39 (6), 790–803. http://dx.doi.org/10.1002/esp.3533.

Grabs, T., Bishop, K., Laudon, H., Lyon, S.W., Seibert, J., 2012. Riparian zonehydrology and soil water total organic carbon (TOC): implications for spatialvariability and upscaling of lateral riparian TOC exports. Biogeosciences, 3901–3916. http://dx.doi.org/10.5194/bg-9-3901-2012.

Grathwohl, P., Rügner, H., Wöhling, T., Osenbrück, K., Schwientek, M., Gayler, S.,Wollschlaeger, U., Benny Selle, B., Pause, M., Delfs, J.-O., Grzeschik, M., Weller,U., Ivanov, M., Cirpka, O.A., Maier, U., Kuch, B., Nowak, W., Wulfmeyer, V.,Warrach-Sagi, K., Streck, T., Attinger, S., Bilke, L., Dietrich, P., Fleckenstein, J.H.,Kalbacher, T., Kolditz, O., Rink, K., Samaniego, L., Vogel, H.-J., Werban, U.,Teutsch, G., 2013. Catchments as reactors: a comprehensive approach for waterfluxes and solute turnover. Environ. Earth Sci. 69 (2), 317–333. http://dx.doi.org/10.1007/s12665-013-2281-7.

Gustafsson, M.E.R., Franze, L.G., 2000. Inland transport of marine aerosols insouthern Sweden. Atmos. Environ. 34, 313–325.

Hannah, D.M., Malcolm, I.A., Soulsby, C., Youngson, A.F., 2008. A comparison offorest and moorland stream microclimate, heat exchanges and thermaldynamics. Hydrol. Process. 940, 919–940. http://dx.doi.org/10.1002/hyp.

Haria, A.H., Shand, P., 2004. Evidence for deep sub-surface flow routing in forestedupland Wales: implications for contaminant transport and stream flowgeneration. Hydrol. Earth Syst. Sci., 334–344

Haria, A.H., Shand, P., Soulsby, C., Noorduijn, S., 2013. Spatial delineation ofgroundwater-surface water interactions through intensive in-stream profiling.Hydrol. Process. 27 (4), 628–634. http://dx.doi.org/10.1002/hyp.9551.

Heal, K.V., 2008. Solutes. In: Holden, J. (Ed.), An Introduction to Physical Geographyand the Environment, second ed. Pearson Education, Harlow, pp. 412–440.

Hrachowitz, M., Soulsby, C., Tetzlaff, D., Dawson, J.J.C., Dunn, S.M., Malcolm, I.A.,2009. Using long-term data sets to understand transit times in contrastingheadwater catchments. J. Hydrol. 367 (3–4), 237–248. http://dx.doi.org/10.1016/j.jhydrol.2009.01.001.

Inamdar, S., Dhillon, G., Singh, S., Dutta, S., Levia, D., Scott, D., Mitchell, M., Van Stan,J., McHale, P., 2013. Temporal variation in end-member chemistry and itsinfluence on runoff mixing patterns in a forested, Piedmont catchment. WaterResour. Res. 49 (4), 1828–1844. http://dx.doi.org/10.1002/wrcr.20158.

Katsuyama, M., Kabeya, N., Ohte, N., 2009. Elucidation of the relationship betweengeographic and time sources of stream water using a tracer approach in aheadwater catchment. Water Resour. Res. 45 (6). http://dx.doi.org/10.1029/2008WR007458.

Kay, A.L., Crooks, S.M., Davies, H.N., Reynard, N.S., 2014. Probabilistic impacts ofclimate change on flood frequency using response surfaces II: Scotland. Reg.Environ. Change 14 (3), 1243–1255. http://dx.doi.org/10.1007/s10113-013-0564-x.

Koeniger, P., Leibundgut, C., Stichler, W., 2009. Spatial and temporalcharacterisation of stable isotopes in river water as indicators of groundwatercontribution and confirmation of modelling results; a study of the Weser river,Germany. Isot. Environ. Health Stud. 45 (4), 289–302. http://dx.doi.org/10.1080/10256010903356953.

Leibundgut, C., Maloszewski, P., Kuells, C., 2009. Tracers in Hydrology. Wiley-Blackwell, Sussex, pp. 1–415.

Liu, F., Hunsaker, C., Bales, R.C., 2013. Controls of streamflow generation in smallcatchments across the snow-rain transition in the Southern Sierra Nevada,California. Hydrol. Process. 27 (14), 1959–1972. http://dx.doi.org/10.1002/hyp.9304.

Malcolm, I.A., Soulsby, C., Youngson, A.F., Hannah, D.M., McLaren, I.S., Thorne, A.,2004. Hydrological influences on hyporheic water quality: implications forsalmon egg survival. Hydrol. Process. 18 (9), 1543–1560. http://dx.doi.org/10.1002/hyp.1405.

Malcolm, I.A., Soulsby, C., Youngson, A.F., Hannah, D.M., 2005. Catchment-scalecontrols on groundwater-surface water interactions in the hyporheic zone:

implications for salmon embryo survival. River Res. Appl. 21 (9), 977–989.http://dx.doi.org/10.1002/rra.861.

Meredith, K.T., Hollins, S.E., Hughes, C.E., Cendón, D.I., Hankin, S., Stone, D.J.M., 2009.Temporal variation in stable isotopes (18O and 2H) and major ionconcentrations within the Darling River between Bourke and Wilcannia dueto variable flows, saline groundwater influx and evaporation. J. Hydrol. 378 (3–4), 313–324. http://dx.doi.org/10.1016/j.jhydrol.2009.09.036.

Murphy, J.M., Sexton, D.M.H., Jenkins, G.J., Boorman, P.M., Booth, B.B.B., Brown, C.C.,Clark, R.T., Collins, M., Harris, G.R., Kendon, E.J., Betts, R.A., Brown, S.J., Howard,T.P., Humphrey, K.A., McCarthy, M.P., McDonald, R.E., Stephens, A., Wallace, C.,Warren, R., Wilby, R., Wood, R.A., 2009. UK Climate Projections Science Report:Climate change projections. Met Office Hadley Centre, Exeter, pp. 21–35. http://dx.doi.org/10.1787/9789264086876-5-en.

Orr, H.G., Carling, P.A., 2006. Hydro-climatic and land use changes in the River Lunecatchment, North West England, implications for catchment management.River Res. Appl. 22 (2), 239–255. http://dx.doi.org/10.1002/rra.908.

Parsekian, A.D., Sing, K., Minsley, B.J., Holbrook, W.S., Slater, L., 2015. Multi-scalegeophysical imaging of the critical zone. Rev. Geophys. http://dx.doi.org/10.1002/2014RG000465l.

Payn, R.A., Gooseff, M.N., McGlynn, B.L., Bencala, K.E., Wondzell, S.M., 2012.Exploring changes in the spatial distribution of stream baseflow generationduring a seasonal recession. Water Resour. Res. 48 (4). http://dx.doi.org/10.1029/2011WR011552.

Pierret, M.C., Stille, P., Prunier, J., Viville, D., Chabaux, F., 2014. Chemical and U-Srisotopic variations in stream and source waters of the Strengbach watershed(Vosges mountains, France). Hydrol. Earth Syst. Sci. 18 (10), 3969–3985. http://dx.doi.org/10.5194/hess-18-3969-2014.

Rinaldo, A., Beven, K.J., Bertuzzo, E., Nicotina, L., Davies, J., Fiori, A., Russo, D., Botter,G., 2011. Catchment travel time distributions and water flow in soils. WaterResour. Res. 47 (7). http://dx.doi.org/10.1029/2011WR010478.

Rodgers, P., Soulsby, C., Petry, J., Malcolm, I., Gibbins, C., Dunn, S., 2004. Groundwater–surface-water interactions in a braided river: a tracer-based assessment. Hydrol.Process. 18 (7), 1315–1332. http://dx.doi.org/10.1002/hyp.1404.

Shand, P., Darbyshire, D.P.F., Gooddy, D., Haria, A.H., 2007. 87Sr/86Sr as an indicatorof flowpaths and weathering rates in the Plynlimon experimental catchments,Wales, UK. Chem. Geol. 236 (3–4), 247–265. http://dx.doi.org/10.1016/j.chemgeo.2006.09.012.

Shaw, G.D., Conklin, M.H., Nimz, G.J., Liu, F., 2014. Groundwater and surface waterflow to the Merced River, Yosemite Valley, California: 36Cl and Cl- evidence.Water Resour. Res., 1943–1959 http://dx.doi.org/10.1002/2013WR014222.

Soulsby, C., Tetzlaff, D., 2008. Towards simple approaches for mean residence timeestimation in ungauged basins using tracers and soil distributions. J. Hydrol.363 (1–4), 60–74. http://dx.doi.org/10.1016/j.jhydrol.2008.10.001.

Soulsby, C., Rodgers, P.J., Petry, J., Hannah, D.M., Malcolm, I.A., Dunn, S.M., 2004.Using tracers to upscale flow path understanding in mesoscale mountainouscatchments: two examples from Scotland. J. Hydrol. 291 (3–4), 174–196. http://dx.doi.org/10.1016/j.jhydrol.2003.12.042.

Soulsby, C., Tetzlaff, D., van den Bedem, N., Malcolm, I.A., Bacon, P.J., Youngson, A.F.,2007. Inferring groundwater influences on surface water in montanecatchments from hydrochemical surveys of springs and streamwaters. J.Hydrol. 333 (2–4), 199–213. http://dx.doi.org/10.1016/j.jhydrol.2006.08.016.

Sullivan, P.L., Price, R.M., Miralles-Wilhelm, F., Ross, M.S., Scinto, L.J., Dreschel, T.W.,Sklar, F.H., Cline, E., 2014. The role of recharge and evapotranspiration ashydraulic drivers of ion concentrations in shallow groundwater on Evergladestree islands, Florida (USA). Hydrol. Process. 28 (2), 293–304. http://dx.doi.org/10.1002/hyp.9575.

Szramek, K., Walter, L.M., Kanduc, T., Ogrinc, N., 2011. Dolomite versus calciteweathering in hydrogeochemically diverse watersheds established on beddedcarbonates (Sava and Soca Rivers, Slovenia). Aquat. Geochem. 17 (4–5), 357–396. http://dx.doi.org/10.1007/s10498-011-9125-4.

Tetzlaff, D., Soulsby, C., Waldron, S., Malcolm, I.A., Bacon, P.J., Dunn, S.M., Lilly, A.,Youngson, A.F., 2007. Conceptualization of runoff processes using ageographical information system and tracers in a nested mesoscalecatchment. Hydrol. Process. 1307 (October 2006), 1289–1307. http://dx.doi.org/10.1002/hy.

Tetzlaff, D., Birkel, C., Dick, J., Geris, J., Soulsby, C., 2014. Storage dynamics inhydropedological units control hillslope connectivity, runoff generation, andthe evolution of catchment transit time distributions. Water Resour. Res., 969–985 http://dx.doi.org/10.1002/2013WR014147.

Tetzlaff, D., Buttle, J., Carey, S., McGuire, K.J., Laudon, H., Soulsby, C., 2015. Tracer-based assessment of flow paths, storage and runoff generation in northerncatchments: a review. Hydrol. Process. http://dx.doi.org/10.1002/hyp.10412.

Touhari, F., Meddi, M., Mehaiguene, M., Razack, M., 2014. Hydrogeochemicalassessment of the Upper Cheliff groundwater (North West Algeria). Environ.Earth Sci. http://dx.doi.org/10.1007/s12665-014-3598-6.

Vuai, S., Tokuyama, A., 2007. Solute generation and CO2 consumption duringsilicate weathering under subtropical, humid climate, northern Okinawa Island,Japan. Chem. Geol. 236 (3–4), 199–216. http://dx.doi.org/10.1016/j.chemgeo.2006.09.009.

Ward, A.S., Gooseff, M.N., Fitzgerald, M., Voltz, T.J., Singha, K., 2014. Spatiallydistributed characterization of hyporheic solute transport during baseflowrecession in a headwater mountain stream using electrical geophysical imaging.J. Hydrol. http://dx.doi.org/10.1016/j.jhydrol.2014.05.036.

Weiler, M., McDonnell, J.J., 2006. Testing nutrient flushing hypotheses at thehillslope scale: a virtual experiment approach. J. Hydrol. 319, 339–356. http://dx.doi.org/10.1016/j.jhydrol.2005.06.040.

M. Blumstock et al. / Journal of Hydrology 527 (2015) 1021–1033 1033

Wilkinson, J., Reynolds, B., Neal, C., Hill, S., Neal, M., Harrow, M., 1997. Major, minorand tracer element composition of cloudwater and rainwater at Plynlimon.Hydrol. Earth Syst. Sci. 1, 557–569.

Wirmvem, M.J., Ohba, T., Fantong, W.Y., Ayonghe, S.N., Hogarh, J.N., Suila, J.Y.,Nkengmatia, A., Asaah, E., Ooki, S., Tanyileke, G., Hell, J.V., 2014. Origin of majorions in monthly rainfall events at the Bamenda Highlands, North WestCameroon. J. Environ. Sci. 26 (4), 801–809. http://dx.doi.org/10.1016/S1001-0742(13)60502-1.

Yidana, S.M., Koffie, E., 2014. The groundwater recharge regime of some slightlymetamorphosed neoproterozoic sedimentary rocks: an application of naturalenvironmental tracers. Hydrol. Process. 28 (7), 3104–3117. http://dx.doi.org/10.1002/hyp.9859.

Zimmer, M.A., Bailey, S.W., McGuire, K.J., Bullen, T.D., 2012. Fine scale variations ofsurface water chemistry in an ephemeral to perennial drainage network.Hydrol. Process. http://dx.doi.org/10.1002/hyp.