A multi-parameter hydrochemical characterization of proglacial runo ff , Cordillera Blanca , Peru

Introduction Conclusions References


Introduction
Rapid glacier retreat in the tropical Andes is having significant consequences for mountain glaciers and the people who rely on glacially-fed water supplies (Bradley, 2006;Vuille et al., 2008).Among other effects, climate change and glacier recession threaten to decrease dry season discharge in this regions, representative of many global sites where highland water ecosystems reach downstream demand (Barnett et al., 2005;Weingartner et al., 2007).In the seasonally arid climate of the tropical Andes, glacier meltwater buffers discharge throughout the year and provides a net increase in overall discharge (Mark et al., 2007) at the expense of the negative glacier mass balance.Therefore during the dry season or droughts, glacier meltwater is an important water resource.In addition to water resource quantity, water qualityis also an important issue.In this context, proglacial areas downstream from glaciers are significant as the fractional glacier melt contribution diminishes further downstream (i.e.Kaser et al., 2010).Thus we focus here on how glacier melt fed streams are impacted by water-rock chemistry throughout a proglacial valley.
The Cordillera Blanca, located in the central Andes of Peru, contains about 70 % of the world's tropical glacier area (Vuille et al., 2008).Most of the Peruvian population lives along the Pacific coastal plains and western slopes of the Andes where the land is arid and heavily reliant on water runoff from the mountains (Vergara et al., 2007).Recent global climate changes are locally manifest in the Cordillera Blanca as proglacial watersheds are undergoing hydrological transformations, affecting human vulnerability across multiple shifting vectors (Mark et al., 2010;Bury et al., 2010).Thus, understanding the current physical and chemical hydrology of the region is imperative for making future predictions about climate impacts.
The impact of melting glaciers on the hydrological cycle has been quantified on a regional scale, but more detailed, valley-specific analyses are required to characterize the geological controls and variability of water quality and stream discharge accurately (Mark et al., 2005).An understanding of these geological and hydrological controls is beneficial for the utilization of mixing models which can be used to determine relative contributions from end-members such as precipitation, groundwater, and glacier melt.
Previous work has successfully utilized major ion and isotope hydrochemistry over multiple years to identify end-member contributions from tributary streams to the Rio Santa (Mark et al., 2005;Mark and McKenzie, 2007).During the dry season, groundwater is one of the largest sources of water to the hydrologic system in the tributary valleys (Baraer et al., 2009), yet many existing models oversimplify the role of groundwater in glacial environments (Hood et al., 2006).This case study focuses on the hydrochemistry of waters in a proglacial valley during the 2009 dry season, when precipitation is minimal and the relative contributions of glacier melt water and groundwater to streams is predominate (Mark and Seltzer, 2003).The proglacial zone is a mosaic of moraines and other deposits related to glacier advance, retreat, and hydrology (Tranter, 2005).This area has the potential for high geochemical activity, because it contains a variety of comminuted glacial debris, is subject to reworking by glaciofluvial activity, and can be colonized by vegetation.Waters interacting with sediment in this area mainly acquire new solutes via sulfide oxidation and carbonate dissolution (Anderson et al., 2000;Cooper et al., 2002).Furthermore, the concentration of solutes in the proglacial zone is normally higher than solute concentrations of glacial runoff (Tranter, 2005).Natural springs and low-gradient, low permeability plains called pampas are the main sources of groundwater in the proglacial zone of the Quilcayhuanca drainage basin.
A variety of hydrochemical parameters (ions, nutrients and stable isotopes) are examined to explore the fate of glacier meltwater downstream from receding glaciers in the Quilcayhuanca basin of the Cordillera Blanca, Peru.The primary objective of the paper is to understand how the geology and hydrology contribute to changes in the hydrochemistry and stable isotopes of surface water progressively downstream from the glaciers.Hydrochemical and isotopic variation between groups of samples from similar locations (e.g.principal channel surface water, tributaries, and groundwater) can be used in simple mass-balance mixing models.A two end-member mixing model is used to determine relative contributions of groundwater to the Quilcayhuanca basin, and these results are then scaled to the large Cordillera Blanca drainage basin.

Study area
The Quilcayhuanca basin is part of the Cordillera Blanca, which spans over 120 km between 8.5-10 • S latitude in the central Peruvian Andes (Fig. 1).The range strikes northwest-southeast and separates stream runoff between the Pacific and Atlantic oceans.Between 1970 and 2003 the surface area of glaciers in the Cordillera Blanca decreased from 723 to 597 km 2 , a 22 % loss (Racoviteanu et al., 2008).Future climate scenarios predict that these glaciers could disappear completely by 2200 (Pouyaud, 2004;Pouyaud et al., 2005).Water originating from the Quilcayhuanca basin drains into the Rio Santa near the city of Huaraz (population ∼120 000).The Rio Santa drains the Cajellon de Huaylas watershed (4900 km 2 ) which captures runoff from the western side of the Cordillera Blanca and eastern side of the non-glacierized Cordillera Negra.Originating at Laguna Conococha, the Rio Santa travels over 300 km to the Pacific Coast, descending from 4300 m and draining a watershed of 12 200 km 2 .The Rio Santa has the second largest discharge of rivers draining to the Pacific coast of Peru and also has the most regular monthly flow (Mark and Seltzer, 2003).Rio Santa discharge has a strong seasonal pattern, reflecting that approximately 80 % of total annual precipitation falls from roughly October to May (Mark et al., 2010).Contributions from glacier melt are thus most important during the dry season which is from roughly June to September.
The Quilcayhuanca basin has a drainage area of 90 km 2 and is approximately 20 % glacierized based on 2009 Advanced Spaceborne Thermal Emission and Reflectance Radiometer (ASTER) satellite imagery.In 2009 glaciers in the basin covered an area of 18 km 2 .Glacier outlines, name, type (in some cases), and size are available from the Global Land Ice Measurements from Space (GLIMS) glacier database (Armstrong et al., 2005).The valley has a distinct "Y" shape (Fig. 2).In later discussions and tables the northern right-lateral branch is referred to as Quil-R while the southern left-lateral branch is referred to as Quil-L.The main section of the valley at and below where the two upper branches merge is referred to as Quil-M.the high, steep walls of the mid to lower-valley.This intrusive body is characterized as granodiorite and tonalite which is approximately 8.2 ± 0.2 million years old (McNulty et al., 1998).Quaternary moraines, glaciolacustrine deposits, glaciofluvial deposits, and pampas compose the remainder of the valley (15 % of basin area).
Pampas are defined as low-gradient areas that formed from paludified moraine dammed lakes.The pampas are composed mostly of low-permeability, organic-rich, unconsolidated material and buried, higher permeability colluvial deposits (Mark and McKenzie, 2007).Glacial deposits, fluvial deposits, and pampas are important geological components controlling hydrochemistry as they are low gradient systems compared to the steep valley sides and thus they should theoretically contain water with longer residence times.Cuchillacocha Lake and Tuplacocha Lake, with areas of 0.14 km 2 and 0.46 km 2 respectively, are located immediately in front of glaciers in Quil-R.There are no major lakes in Quil-L.

Methodology
The study is based on the analysis of 25 water samples for major ion chemistry, nutrients, and the stable isotopes of water.Samples were divided into three hydrologic groups to look at spatial patterns, such as elevation: (1) Quil Streams, represented in figures by blue-filled circles, is the group of sampling sites (n = 14) from principal stream channels in each of the three valley sections (referred to as Quil-L, Quil-R, or Quil-M in Tables); (2) Tributaries Group, represented by green-filled squares, is composed of tributary samples (n = 8) in each of the three sections of the valley (referred to as Tribs in Tables ); and (3) Groundwater Group, represented by red-filled triangles, is composed of groundwater samples (n = 3) taken from springs (referred to as GW in Tables).
Tributaries were sampled in both upper sections of the basin.In Quil-R samples were taken from two tributaries (sites #2 and #3) thought to be draining a high elevation groundwater storage area called Jatun.In Quil-L samples were taken from tributaries on either side of the valley.In general, waters in the upper portion of the basin flow over bedrock, through moraines, and though fluvial deposits.Once the waters enter Quil-M they are mostly flowing across pampas.Waters draining from the Quilcayhuanca basin flow down a steeper gradient through the city of Huaraz.However, before the waters from Quilcayhuanca reach Huaraz, they merge with streams from two adjacent valleys.
The most downstream water sample, site Quilcay (#25), was taken at a point just above the city of Huaraz, approximately 2 km before the stream enters the Rio Santa.

Sample collection and analysis
Water samples were collected in the Quilcayhuanca basin during a three day period of the dry  Concentrations of nutrients (total P, total N, and Si) were measured using a Skalar San ++ Automated Wet Chemistry Analyzer in the School of Earth Sciences at The Ohio State University.Total P is a combination of orthophosphates, polyphosphates, and organic phosphorus.Total N includes organic nitrogen, ammonia, and nitrate plus nitrite.Duplicate samples analyzed for total N, P and Si had average precisions of 11 %, 3 % and 1 %, respectively.Silica values were reported as ppb Si as SiO 2 .
Stable isotopes of water were measured at Byrd Polar Research Center at The Ohio State University using mass spectroscopy (Finnigan MAT Delta Plus coupled to a HDO water equilibrator).Stable isotopes are reported using the δ-notation reported relative to the Vienna-Standard Mean Ocean Water (VSMOW) standard, with accuracy of ±0.1 ‰ for δ 18 O and ±1 ‰ for δ 2 H. Based on duplicates, δ 18 O measurements had a precision of 0.2 % and δ 2 H measurements had a precision of 2.4 %.A t-test was used to determine if sample groups (Quil Streams, Tributaries, and Groundwater) were statistically different from one another in terms of the various hydrochemical components being examined in this study.For this study a probability below the 5 % significance level (p < 0.05) refutes the null hypothesis that there is no difference between the two groups and confirms the alternative hypothesis that the two groups are different from one another.

Mixing model
A two component chemical mixing model was used to estimate relative contributions of surface water and groundwater at two sites in Quilcayhuanca valley.The first site, Quil Bel Conf (#19), was chosen because it sits above the pampa of the lower valley which is hypothesized to be a potential storage site for groundwater.The second site, Park Entrance (#24), is located at the lower end of the pampa (Fig. 2).Thus the relative percentages of groundwater and surface water at this site should represent the net mixture of groundwater and surface water exiting the valley as surface water.Averaged values of different parameters for the groups Quil Streams and Groundwater were used as the two end-members.The group Quil Streams is the union of the three subgroups Quil-L, Quil-R, and Quil-M.Precipitation was not used as an end member because precipitation events are rare, short duration events during the dry season and no precipitation samples were collected from Quilcayhuanca (Mark et al., 2005).
Similar to Baraer et al. (2009) selected tracers had to be conservative, meaning that values measured in the mixed stream water had to be within the range of source concentrations.Average values for each source were calculated from sites above the mixing point.The tracers that met these selection criteria were Ca 2+ , Mg 2+ , SO 2− 4 , Si, and δ 18 O.The mixing equation is: where f 1 is the unknown fraction of water at the site coming from surface water, (1 − f 1 ) is the unknown fraction of water at the site coming from groundwater, C out is the average measured concentration of components at the site in question, C Qms is the average measured concentration of components for the group Quil Streams, and C GW is the average measured concentration of components for the group Groundwater.

Field measurements
Discharge of the principal valley streams increases progressively down-valley.The discharge from the Cuchillacocha Lake outflow (#1) in Quil-R was measured to be 0.1 m 3 s −1 , while the highest sampling site in Quil-L, Cayesh Hi (#7), had a discharge of about 0.2 m 3 s −1 .The discharge was measured to be 0.8 m 3 s −1 at the confluence (#19) of Quil-L and Quil-R.Near the end of the valley at the site Park Entrance (#24) the discharge was measured to be 1.2 m 3 s −1 .The average pH for all sites in Quilcayhuanca was 3.6 and the overall range of pH measurements was from 2.8 to 7.3 (Table 1).This range shows a stark contrast compared to the normal pH range of 7-10 seen in most glacial environments (Tranter, Overall the pH of Quil Streams increases with decreasing elevation.This trend is attributed to the increasing net contribution of groundwater down-valley.Furthermore, Quil Streams sites in the Quil-M section of the valley should have a higher pH relative to main channel sites in Quil-L and Quil-R because tributaries feeding the principal channel are not exposed to the pyrite (Fe 2 S) bearing Chicama formation, which is the main driving force for production of H + ions in the valley.Groundwater sites can be thought of as a separate grouping since samples have statistically different (t-test: p < 0.0001) pH compared to Quil Streams.The groundwater sites do not show a trend with elevation.The pH of tributary streams also do not show a trend with elevation and have a much greater range than Quil Streams and groundwater.
The overall average specific conductance, a measure of total ionic activity in water, for all sites was 251 µS cm −1 although a large range was observed (26-495 µS cm −1 ).Quil Streams (n = 14) had an average specific conductance of 318 µS cm −1 and ranged from 226-495 µS cm .Specific conductance of Quil Streams is statistically different (t-test: p < 0.0006) from groundwater possibly due to the fact that igneous and metamorphic rocks contain silicates and alumino-silicate minerals that are slow to react with groundwater that has a nearly neutral pH.
The specific conductance values from all groups were plotted versus elevation (Fig. 4a).The solid blue significant trend line fit only through Quil Streams, but excluding the outlier site Quilcay (#25), shows that specific conductance decreases with decreasing elevation.This trend can be interpreted based on stream discharge.In glacial environments, as discharge increases, specific conductance decreases (Anderson et al., 2003;Tranter, 2005) because glacier meltwater is usually more dilute than surface water in the proglacial zone.Another interpretation is that surface waters are interacting and mixing more with groundwater in the pampas of Quil-M, which coincides with the pH decrease down valley due to increasing contributions from groundwater in the lower portions of the valley.It is also noteworthy that the tributaries Jatun Upper Conf (#2), Jatun Mid (#3), and North Waterfall (#21) are all similar to groundwater in terms of pH and specific conductivity (circled on Fig. 4a).This suggests that either the waters are originating from a groundwater source in the upper portion of Quil-R or they are interacting with a different lithology, presumably the intrusive formation which should not yield acidic waters.
During the 24 h sampling period the values for specific conductance at the site Quil Bel Conf ranged from 349-466 µS cm −1 (Fig. 5).From the local time of the first sample (09:30 p.m.), the specific conductance gradually decreased to its low value at 07:30 a.m.After 07:30 a.m. the specific conductance gradually increased to its high value at 14:30 (02:30 p.m.) and then gradually decreased again.With knowledge of the specific conductance variation between surface water and groundwater in the valley one explanation for this diurnal variation might be that surface water, originating mostly from glacier melt, contributes a greater percentage during the warmer daylight hours while groundwater has an increased role during the night (∼07:00 p.m.-07:00 a.m.).Anderson et al. (2003) showed that discharge of a river in a glacial valley (Kennicott River, Alaska) is inversely proportional to electrical conductivity.However, those measurements were made much closer to the glacier terminus (0.5 km) and in alkaline waters.The sampling site in this study should have a greater solute concentration because streams draining to this site have a much lower pH, presumably favoring mineral dissolution, and a greater distance (∼4 km from the terminus) to interact with minerals in the proglacial zone.
Glacial runoff is usually a dilute Ca 2+ -HCO − 3 -SO 2− 4 solution with variable Na + -Cl − (Tranter, 2005) and is usually more dilute than global mean river water (Anderson et al., 1997).Compared to average chemical compositions of some of the major rivers of the world (Faure, 1998), Ca 2+ and Mg 2+ concentrations from this study are above global averages while the other major ions (Na + , K + , Cl − , and HCO − 3 ) are below these global river averages.Furthermore, ionic concentrations from this study can be compared with concentrations of major ions in glacial runoff from different regions of the world (Brown, 2002).Concentrations of all the major ions, except Mg 2+ and SO 2− 4 , are similar when compared with ranges from other glacial environments while Mg 2+ and SO 2− 4 are anomalously high in comparison.Unusually high Mg 2+ concentrations may be the result of weathering of common minerals found in granodiorite and tonalite, such as amphibole, biotite, and possibly pyroxene.Very high SO 2− 4 concentrations are the result of pyrite oxidation in the shales of the metasedimentary Chicama formation in the upper section of the valley.
Pyrite oxidation, described by the equation below (Faure, 1998;Fortner et al., 2011), is likely the driving force of this unusually acidic natural system.The oxidation of pyrite can also take place in the absence of O 2 with Fe 3+ serving as an electron acceptor: sulfide oxidation is a dominant reaction in subglacial and proglacial environments.These reactions provide protons to solution, lowering the pH and allowing for more carbonate dissolution.The oxidation of pyrite preferentially dissolves carbonates, rather than silicates, because of a faster reaction rate (Tranter, 2005) and the processes can be accelerated by bacterial activity (Faure, 1998).
In natural settings, pyrite oxidation usually occurs in debris-rich environments where bedrock is crushed and scoured from past glacial contact (Tranter, 2005).Accordingly in this study, the Chicama formation in the upper valley (Fig. 3 with Chicama fm.contact) is exposed and steeply sloped debris piles abound, notably next to the two lakes and in Quil-L.Thus in the two lakes and Quil-L scoured rocks from the Chicama formation likely cause talus deposits that intersect with already acidic waters, further continuing the oxidation of pyrite and addition of protons to the surface waters.
Samples can also be compared by group.For example, there is little variation between Quil Streams and the Tributaries group.However, there are major differences between these two groups and Groundwater.Groundwater samples were, on average, lower in relative percentages of Mg 2+ , H + , and SO 2− 4 and higher in relative percentages of Na + , K + , and HCO − 3 .Increased percentages of Na + and K + in groundwater could be explained by cation exchange processes (Hounslow, 1995) while increased percentages of HCO − 3 could be explained by increased interactions with trace carbonates along groundwater flow paths or decreased interaction with pyrite and other sulfide minerals which lower the pH and reduce alkalinity.

Nutrients
Three nutrients, total P, total N, and Si, were considered (Table 4).The average total P for all samples from Quilcayhuanca was 971 ppb P as PO Groundwater sites had an average value of 443 ppb P as PO 3− 4 and ranged from 9-1297 ppb P as PO 3− 4 .The two springs in the upper portion of the valley had values close to zero while the site Quil Spring (#23) had the maximum value for the group.Total P shows no trend with elevation (Fig. 4b), and all groundwater sites (with the exception of Quil Spring and sites possibly originating from groundwater such as Jatun Upper Conf (#2), Jatun Mid (#3), and North Waterfall (#21)) have very low total P values.This observation suggests that total P could be used to distinguish between Groundwater and Quil Streams which are statistically different (t-test: p < 0.003).
Rocks which are comminuted by intense physical erosion release phosphorus.Average crustal rocks contain 1050 ppm of phosphorus, primarily from sparingly soluble minerals such as apatite and from calcium, aluminum, and ferrous phosphates (O'Neil, 1985).In glacial environments, between 1-23 µg P g −1 is present as readily extractable phosphorus on the surfaces of glacial flour (Hodson et al., 2004) and P fluxes from glacial environments are likely to respond directly to increased meltwater runoff (Hodson, 2009).In this study total P showed no trend with elevation (Fig. 4b), contrary to an expected trend of decreasing phosphorus with elevation.Scoured phosphate minerals should be most readily available near the glacier termini and thus measured total P concentrations should have their highest values at the highest sampling elevations.Additionally in older sediments of the lower valley, more of these scoured minerals should have been removed compared to the most recently glacierized terrain in the upper valley as plants in the lower valley should utilize P and remove it from aquatic systems.3 and a range of 205-1957 ppb N as NO − 3 .Two of the groundwater measurements are similar, but the third measurement, J Spring (#12), is very high and results in an unrepresentative average.No trend is observed between total N and elevation (Fig. 4c).Total N would not be a good method to distinguish between Quil Streams and Groundwater since they are not statistically different (t-test: p < 0.072).
Snow and ice melt provide limited quantities of nitrogen, mostly as NO − 3 and NH + 4 (Tranter, 2005) and it is possible that NH + 4 is sourced from mica and feldspar dissolution (Holloway et al., 1998) and some may originate from oxidation of organic matter.Cattle, horses and sheep which graze in the valley could also contribute to measured total N.
The average Si for all samples was measured to be 9.5 ppm Si as SiO 2 with a range of 0.5-1.3ppm Si as SiO 2 .The Quil Streams group had an average value of 9.3 ppm Si as SiO 2 and a range of 4.4-1.1 ppm Si as SiO 2 .Tributaries had an average value of 9.0 ppm Si as SiO 2 and 4.1-11 ppm Si as SiO 2 .Groundwater samples had an average value of 11.9 ppm Si as SiO 2 and a range of 9.9-13.5 ppm Si as SiO 2 .The samples J Spring (#12) and Quil Spring (#23) have the highest Si concentrations of all the samples from Quilcayhuanca while Cay Spring (#15) is close to the overall average.Silica could potentially be used to distinguish between groundwater and surface water samples since Groundwater values are significantly different (t-test: p < 0.024) from Quil Streams.Si concentrations increase down valley for Quil Streams (Fig. 4d).However, this trend is not quite statistically significant at the 5 % significance level.
Silica in most low-temperature natural waters is derived from silicate weathering (Hounslow, 1995).In proglacial environments, carbonate dissolution and sulfide oxidation reactions dominate initially after glacier retreat, but these minerals become exhausted as silicate weathering increases.plants further increases the rate of chemical weathering of silicates (Anderson et al., 2000).These observations coincide well with the measured concentrations of Si and how they vary with elevation in this study.Presumably the oldest glacial deposits are in the lower portions of the valley and these deposits have been interacting with naturally acidic waters for quite some time.As a result, carbonate and sulfide minerals in these lower, older sediment portions of the valley have probably been exhausted leading to increased silicate weathering in these areas.

Isotopes
The stable isotopes δ 18 O and δ 2 H were measured and deuterium excess was calculated for the samples from Quilcayhuanca basin (  (Schwartz, 2003) and thus so does conductivity.A plot of δ 18 O versus elevation shows an elevation effect of −0.155 ‰ per 100 m of elevation rise (Fig. 6a).This elevation effect is approximately two times as large as the elevation effect observed for the nonglacierized Cordillera Negra (−0.07 ‰ per 100 m) by Mark et al. (2007Mark et al. ( ) from 2004Mark et al. ( -2006.Although it is difficult to quantify the amount of glacier meltwater in the surface waters of Quilcayhuanca, this comparison shows that meltwater is a significant component during the dry season.δ 2 H shows a potential local meteoric water line for Quilcayhuanca valley, assuming that stream samples can act as a proxy for precipitation (Fig. 6b; McKenzie et al., 2001).The local meteoric water line falls below the global meteoric water line (MWL), similar to observations by Mark et al. (2007).

Mixing model
The two-component mixing model was used to calculate that the Quil Bel Conf (#19) site is comprised of 76 % surface water and 24 % groundwater with a standard deviation of ±16 % (Table 8).The Park Entrance site (#24) is estimated to be 66 % surface water and 34 % groundwater with a standard deviation of ±12 %.These two sites are not statistically different (t-test: p < 0.27) in terms of their relative contribution from surface water and groundwater.In general, however, it appears that groundwater is a slightly larger component at the lower site after the streams cross the potential pampa groundwater storage sites.A similar, but more complex, model was applied to the 7 % glacierized Querococha basin of the Cordillera Blanca by Baraer et al. (2009) where groundwater is the dominant contributor to Querococha surface waters during the dry season.The authors noted that the relative contribution from groundwater is variable, ranging from 18 to 74 %, but that proglacial groundwater contributions are a key component of the dry season hydrologic system in this valley and likely the rest of the Cordillera Blanca.Our results confirm the importance of groundwater in proglacial environments, and indicate that it should be accounted for when quantifying water resources, particularly during the dry season (Baraer et al., 2009).

Upscaling to the Rio Santa
Of interest is how the results from this study can be extrapolated to the larger Cordillera Blanca watershed.Interestingly, groundwater sites from Quilcayhuanca resemble the Rio Santa more closely than Quilcayhuanca surface waters.Furthermore, the average composition of groundwater, the Rio Santa, and the Cordillera Negra all fall on a mixing line.Assuming Quilcayhuanca water is representative of broader groundwater compositions, groundwater contributes approximately 60 % to the Rio Santa and surface waters from the Cordillera Negra contributing approximately 40 %.Although groundwater sites from Quilcayhuanca fall on a mixing line with the Rio Santa and the Cordillera Negra, groundwater from Quilcayhuanca alone obviously does not contribute 60 % of the dry season discharge to the Rio Santa but is representative of groundwater chemical compositions.
The Cordillera Blanca itself might contribute 60 % of dry season discharge to the Rio Santa, similar to what Mark et al. (2005) observed, if groundwater in other valleys is similar in ionic composition to the groundwater in Quilcayhuanca or if groundwater in Quilcayhuanca is similar to surface waters and groundwater in the other valleys.The ionic composition of groundwater in Quilcayhuanca measured in this study is very similar to the average ionic composition of the major tributaries to the Rio Santa measured by Mark et al. (2005) whose authors estimated that the Cordillera Blanca contributes about 66 % of dry season discharge while the Cordillera Negra contributes about 33 %.

The average δ 18
O and δ 2 H of Quilcayhuanca, the Rio Santa, and the Cordillera Negra were plotted with horizontal and vertical bars representing one standard deviation to represent mixing of these groups (Fig. 7).The waters from Quilcayhuanca have the most negative values of δ 18  O and δ 2 H resulting from glacier melt and runoff derived from higher elevation precipitation.Based on average isotopic values of the Rio Santa and Cordillera Negra (  (Clark and Fritz, 1997).However, in the Cordillera Blanca most precipitation falls in the wet season and is usually more depleted than the most negative value measured in Quilcayhuanca in this study.Thus shallow groundwater, at least in Quilcayhuanca, is not a mirror image of wet season precipitation and therefore dry season precipitation should not be a major factor in determining the isotopic composition of groundwater since precipitation during this time is usually at a minimum.

Conclusions
Rivers draining the Cordillera Blanca provide water for downstream communities, and during the dry-season water resources are particularly stressed with ongoing climate change (Bury et al., 2010).Water resources are defined by the volume of available water and the water's quality for a given activity.The water discharging from the Quilcayhuanca basin is not potable for domestic use due to the low pH and high dissolved SO 4 and metal load (Fortner et al., 2011).With ongoing glacier retreat, additional fresh rock faces of the Chicama Formation will be exposed up-valley, leading to sustained, if not worsening, water quality.
The results of the study show how the chemistry of melt-water changes downstream from glaciers in the Cordillera Blanca, and that not only the importance of groundwater in maintaining dry season discharge, but that groundwater is generally isotopically distinct from that glacier meltwater.The key findings from this research are: which are consistent with the mapped lithology of granodiorite/tonalite and the metasedimentary Chicama formation.
2. Hydrochemical parameters show trends with elevation and distance from the glaciers.pH decreases with elevation but specific conductance increases.Both of these trends support the influx of groundwater in lower portions of the valley.
Total P and total N of the Quil Streams sites show no trend with elevation.Si concentrations increase with decreasing elevation because sulfide and carbonate minerals have been more exhausted at lower elevations.An elevation effect of −0.155 ‰ per 100 m rise in elevation was observed for δ 18 O which is nearly twice the published elevation effect of the nonglacierized Cordillera Negra.This signifies that glacier melt is a large reason why Quilcayhuanca δ 18  O stream values are much more negative than would normally be expected for the given elevation range.
3. Surface water and groundwater in Quilcayhuanca are geochemically different.pH is lower for surface waters as a result of increased interaction with abundant pyrite.
Specific conductance is lower for groundwater samples due to their near neutral pH and decreased dissolution potential.Groundwater in this basin had considerably higher relative percentages of Na + and K + as a result of cation exchange processes and dissolution of trace carbonates.Surface waters had considerably higher relative percentages of Mg   O versus δ 2 H.The dark blue line is fit through all Quilcayhuanca samples and represents a proxy for a meteoric derived local meteoric water line.
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |

2485
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |

Discussion
Paper | Discussion Paper | Discussion Paper | Discussion Paper |

Figure 2
Figure 2 is a geologic map of the Quilcayhuanca basin and Fig. 3 is a hillshade of a Light Detection and Ranging (LiDAR) digital elevation model (DEM), focused on the upper portion of the valley.The purple line on Fig. 3 represents the contact between two different geologic formations.In both Figs. 2 and 3 sampling locations are plotted with symbols appropriate to their grouping and their site number which are explained below.The valley geology is dominated by metasedimentary and intrusive igneous rocks.The Chicama formation (14 % of basin area) dominates the upper portion of the valley.It contains metamorphic sedimentary rocks of Jurassic age, characterized by weathered shale, argillite, sandstone, and pyrite.Intrusive rocks (43 % of basin area) dominate 2487

− 3 )
was calculated as the difference in the solution charge balance.2489 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |

2493
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |

Discussion
Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | 0-2197 ppb P as PO 3− 4 .Quil Streams had an average value of 1211 ppb P as PO 3the tributary average is somewhat unrepresentative as four of the samples had values close to or equal to zero, while the other 4 samples had values greater than 1000 ppb P as PO 3− 4 .
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | The average total N for all samples from Quilcayhuanca was 391 ppb N as NO − 3 with a range of 161-1957 ppb N as NO − 3 .Quil Streams averaged 381 ppb N as NO − 3 and had a range of 161-708 ppb N as NO − 3 .Tributaries had an average of 258 ppb N as NO − 3 and a relatively narrow range of 201-336 ppb N as NO − 3 .Groundwater samples had an average of 836 ppb N as NO − Colonization of the proglacial zone by 2497 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | The average δ 2 H for the entire valley was −126 ‰ with a range of −134 to −110 ‰.For Quil Streams the average was −126 ‰ with a range of −134 to −120 ‰.Tributaries had an average value of −126 ‰ with a range of −134 to −110 ‰.Groundwater samples had an average of −122 ‰ with a range of −134 to −115 ‰.A plot of δ 18 O versus

2499
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |

2501
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | 1. Pyrite oxidation is the dominant mineral weathering reaction in the proglacial zone of Quilcayhuanca.This reaction adds protons to solution and lowers the pH to below 4 at many sites.As a result of such a low pH there is almost no alkalinity and igneous rocks are readily weathered, resulting in high solute concentrations.The sampling group Quil Streams is characterized by

Fig. 1 .Fig. 2 .
Fig. 1.Map of study site.Map of the Cordillera Blanca (modified from Mark and McKenzie, 2007).The red outline shows the approximate location of the Quilcayhuanca drainage basin relative to the Cordillera Blanca and the Rio Santa.2515

Fig. 5 .
Fig. 5. Specific conductance versus Local Time (24 h) at the Quil Bel Conf site.Specific conductance is theorized to correlate with glacier meltwater contribution.The maximum specific conductance value corresponds to peak glacial meltwater contribution (14:30) while the lowest specific conductance value corresponds to a minimum meltwater contribution (07:30).
Department of Earth and Planetary Sciences, McGill University (Dionex DX-100 with Dionex AS14 column, guard and suppressor).Dissolved concentrations of major cations were measured by atomic absorption spectrometry on a Perkin Elmer AAnalyst 100 at the Trace Element Analytical Laboratory (TEAL) at McGill University.
Table6).The average δ 18 O for the entire valley was −16.7 ‰ with a range of −17.8 to −14.4 ‰.For Quil Streams the average was −16.9 ‰ with a range of −17.8 to −16.3 ‰.Tributaries had an average value of −16.6 ‰ with a range of −17.6 to −14.4 ‰.Groundwater samples had an average value of −16.0 ‰ with a range of −17.5 to −15.3 ‰.J Spring (#12) and Quil Spring (#23) have identical values while Cay Spring (#15) is low (depleted).One possible explanation is that Cay Spring has a shorter residence time or flow path than the other two springs indicating that the sample has undergone less evaporation than the other groundwater.As a result, the surface water that initially entered this spring has undergone less fractionation than the other two sites.This theory is supported by measurements of conductivity.Cay Spring has the lowest conductivity of the three groundwater sites.With a longer residence time, the quantity of mass dissolved in groundwater increases

Table 7
) and the assumption that the site Quilcay(#25; ∼2 km from the Rio Santa) is representative of the Cordillera Blanca, this mixing model suggests that between 27 % (δ 18 O) and 38 % (δ 2 H) of the water in the Rio Santa is derived Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | from the Cordillera Blanca.This calculation is a crude estimate of relative contributions to the Rio Santa since discharge and precipitation are not included and one valley is used to represent the entire Cordillera Blanca.Nevertheless, it is a useful exercise and provides the foundation for further more detailed studies.Isotopically, it is of interest that the groundwater δ 18 O values are statistically different (t-test: p < 0.037) from surface waters samples and on average have more positive values.Normally δ 18 O of shallow groundwater and δ 18 O of local precipitation are approximately the same

Table 1 .
Quilcayhuanca Field Measurements by Group.
ter.Groundwater usually has almost no total P and higher concentrations of Si, relative to surface waters.δ18Oandδ 2 H of groundwater are, on average, more positive than δ18O and δ 2 H of surface waters.These differences provide ways to