Rock glacier outflows may adversely affect lakes: lessons from the past and present of two neighboring water bodies in a crystalline-rock watershed.

Despite the fact that rock glaciers are one of the most common geomorphological expressions of mountain permafrost, the impacts of their solute fluxes on lakes still remain largely obscure. We examined water and sediment chemistry, and biota of two neighboring water bodies with and without a rock glacier in their catchments in the European Alps. Paleolimnological techniques were applied to track long-term temporal trends in the ecotoxicological state of the water bodies and to establish their baseline conditions. We show that the active rock glacier in the mineralized catchment of Lake Rasass (RAS) represents a potent source of acid rock drainage that results in enormous concentrations of metals in water, sediment, and biota of RAS. The incidence of morphological abnormalities in the RAS population of Pseudodiamesa nivosa, a chironomid midge, is as high as that recorded in chironomid populations inhabiting sites heavily contaminated by trace metals of anthropogenic origin. The incidence of morphological deformities in P. nivosa of ∼70% persisted in RAS during the last 2.5 millennia and was ∼40% in the early Holocene. The formation of RAS at the toe of the rock glacier most probably began at the onset of acidic drainage in the freshly deglaciated area. The present adverse conditions are not unprecedented in the lake's history and cannot be associated exclusively with enhanced thawing of the rock glacier in recent years.

Introduction

Perennin class="Chemical">ally frozenpan> and glacierized high-n class="Species">alpine areas react very sensitively to changes in air temperature and even small differences in temperature determine their frozen status. As a consequence, glaciers in the European Alps, for example, have lost 30–40% of surface area since the end of the Little Ice Age (∼middle of 19th century),[1] and the lower permafrost limit is estimated to have risen by ∼1 m per year.[2] Discharge from melting glaciers and thawing permafrost can be highly enriched in solutes due to prolonged interaction with fine-grained rocks and freshly exposed mineral surfaces.[3−5] Solute outflows from catchments characterized by cryospheric features can cause serious changes in water chemistry downstream. A survey of remote high altitude lakes in the European Alps[6] has provided strong evidence of a substantial increase in base cation and sulfate concentrations in lake water over the past few decades, especially in lakes from glaciated areas. The increase of solutes in lake water was attributed to an enhanced weathering resulting from climate warming.

Unn class="Chemical">til now, very little research hn class="Chemical">as examined how changes in the mountain cryosphere affect downstream water quality, including the threats to ecosystem health and human use.[7] Recent studies that investigated water quality in alpine watersheds draining mineralized areas with sulfide-bearing lithologies demonstrate that retreating glaciers[8] and degrading permafrost[9] may adversely affect downstream ecosystems receiving meltwater runoff. The exposure of fresh surfaces of sulfide-rich rocks to air and oxygenated water by retreating glaciers or degrading permafrost increases the oxidation of sulfide minerals, which is responsible for the generation of natural acid rock drainage (ARD). ARD typically produces acid-sulfate waters enriched in Fe, Al, Mn, and various trace elements such as Cu, Ni, Pb, Zn, and others.[10,11] Concentration of the dissolved minor and trace elements in natural ARD systems generated by retreating glaciers or degrading permafrost can be comparable to ARD associated with mining activity.[8,9] Some of the mobilized trace elements associated with ARD pose a potential hazard to environmental quality and human health.

n class="Chemical">Although ice loss inpan> permafrost can n class="Chemical">be orders of magnitude slower than in a glacier, the total permafrost area usually exceeds the glacier-covered area in mountain environments and can contribute substantially to the hydrology of alpine catchments.[12] For example, the perennially frozen area in the European Alps is approximately three times larger than the glacier-covered area.[13] Despite the fact that rock glaciers are one of the most common geomorphological expressions of mountain permafrost,[14,15] still little is known about the geochemical content, especially about the trace-element composition of supra- and subpermafrost meltwater that drains rock glaciers.[16] Until recently, most of the research conducted on rock glaciers primarily focused on their movement, origin, internal structure, and hydrology.[4,14] However, the impacts of rock glacier solute fluxes on lakes downstream still remain largely obscure[7] even though more than half of the world population relies on fresh water from mountain areas.[17]

Unexpected high n class="Chemical">nickel and n class="Chemical">manganese concentrations exceeding the appropriate EU limits for drinking water by more than an order of magnitude have recently been reported in a study investigating temporal trends in water chemistry of a remote high-mountain lake, Lake Rasass, situated at the toe of an active rock glacier the European Alps, Italy.[18] The high concentrations of Mn and Ni were observed in the lake water enriched in sulfates (more than 4400 μequiv·L–1). In contrast, negligible concentrations of these metals were recorded in the adjacent pond without a rock glacier in the catchment.[18] In an earlier study, application of paleoecological techniques provided strong evidence of acidic conditions in Lake Rasass (pH 5.8–6.5) during at least the past two centuries.[19] The overall results of the previous studies focused on the lake give evidence of naturally occurring ARD generated by the rock glacier in the Lake Rasass catchment.

The present study aims to explore the efn class="Chemical">fects of rock glacier solute fluxes onpan> a high n class="Species">alpine lake within a crystalline-rock watershed. We applied comparative research design in the study of two neighboring water bodies, Lake Rasass and the adjacent pond, with and without a rock glacier in their mineralized catchments having similar soil conditions and underlying geology. In addition, we used chironomid-based paleolimnological techniques to track long-term temporal trends in the ecotoxicological state of both water bodies and to establish their baseline conditions.

Material Ahd Methods

Study Area

Lake Rn class="Chemical">asn class="Chemical">ass (RAS; surface area = 1.5 ha, max. depth = 9.3 m) and the adjacent pond (RPD; surface area = 0.08 ha, max. depth = 1.6 m) are remote, fishless water bodies situated just 50 m away from each other, above the actual and historical timberline in the upper Vinschgau valley (2682 m a.s.l.) in the Central Eastern Alps, Italy (Figure 1). The bedrock geology of the valley belongs to the Austroalpine Ötztal-Stubai crystalline complex,[20] and the average mineral composition of the bedrock is quartz (50%), feldspar (27%), muscovite (15%), chlorite (6%), and dolomite (2%).[21] The chemical composition of the bedrock is characterized by sources of disseminated water-soluble sulfate (∼180 μg g–1 dry weight (DW); total sulfur is ∼590 μg g–1 DW) and metals, including trace elements Mn (∼597 μg g–1 DW), Ni (∼24 μg g–1 DW) and Zn (∼90 μg g–1 DW).[21] The concentrations of Mn and Zn are comparable with their Earth crustal abundances,[22] while Ni content is more than three times lower. Soil cover is sparse and bedrock is exposed in ∼80% of the RAS and RPD catchment areas. The active rock glacier extends on a north-facing slope down to the shore of RAS and occupies ∼18.5% of its catchment area which adjoins the permafrost-free catchment of RPD. Both water bodies lacking inflow streams have well-developed outflows. Near the RPD shoreline, there are visible sources of groundwater that infiltrates onshore discharges. Since the rock glacier is located only a few meters upslope from RAS, its internal drainage system, a common feature of rock glaciers,[4,23,24] is most likely connected to the lake, and meltwater seeps from the rock glacier into RAS through cracks and fractures in the bedrock.

Figure 1

Map showing the location of the study site (a) and photograph of Lake Rasass (RAS) and the adjacent pond (RPD) and their catchments (b). The red arrows indicate the boundary of the active rock glacier. The map of Italy is reproduced from http://d-maps.com/.

Map showing the locan class="Chemical">tion of the study pan> class="Chemical">site (a) and photograph of Lake Rasass (RAS) and the adjacent pond (RPD) and their catchments (b). The red arrows indicate the boundary of the active rock glacier. The map of Italy is reproduced from http://d-maps.com/.

n class="Chemical">Since 1980, summer and n class="Chemical">cold-season (October–May) temperatures in the European Alps show a simultaneous, strong increase, which is unprecedented over the last millennium.[25] A multifold increase in concentrations of the dominant ions (Mg2+, SO42+, and Ca2+) and electrical conductivity in the RAS water over the last two decades (1986–2005), recorded by Thies et al.,[18] may reflect current permafrost degradation in the rock glacier under the recent warming. Monnier and Kinnard[26] assume that indications of degrading permafrost in a rock glacier may also be the near-melting-point (0 °C) thermal regime of permafrost year-round and a high water content (>40%) in deeper parts of a rock glacier. As solutes concentrate and the freezing point decreases during fluid freezing,[27] it is most likely that highly concentrated salt solutions seep from the rock glacier into RAS through the cold season.

Water Chemistry

Both n class="Chemical">water bodies were sampled for limnpan>ochemicn class="Chemical">al analyses at least twice during the open-water seasons of 2010–2012. Water samples were taken with a Patalas-Schindler sampler (UWITEC, Austria) at 2-m intervals along a vertical profile in RAS and from 0.5 m water depth in the shallower RPD. The physicochemical parameters of water, such as conductivity, pH, alkalinity, and concentrations of major ions and nutrients were determined following standard methods and analytical quality control procedures as described in Mosello and Wathne.[28] Metals (Ag, Al, As, Ba, Be, Cd, Co, Cr, Cu, Fe, Mn, Ni, Pb, Sr, Ti, V, Zn) in water samples were assayed by inductively coupled plasma-optical emission spectrometry (ICP-OES) with standard addition methods.[29]

Sediments and Biota: Sampling Strategy and Chemical Analyses

Sampling strategy wn class="Chemical">as den class="Chemical">signed to compare metal concentrations in surface sediments and biota of the water bodies as follows: the RAS and RPD shallow waters against the RAS deep water, and between shallow waters of RAS and RPD. To provide background information for the sampling in the relatively deep RAS, bathymetric distribution and substratum associations of the most abundant macroinvertebrates were studied in the lake along a depth transect (Supporting Information (SI), Figure S1). Twenty macroinvertebrate samples were collected with an Ekman grab sampler (in deep water) and a kick net (in the upper littoral) at 0.5–1.0 m depth intervals along the east–west transect across the deepest point in the lake. Surface sediment (0–1 cm) samples were collected using an Ekman grab sampler with a lining of plastic inside from three random sites in shallow waters (0.5–1.5 m depth) of both water bodies and in deep water (6–8 m depth) of RAS. The sediment and biota sampling for subsequent multielement analysis was conducted in the early open-water season of 2012.

For muln class="Chemical">tielemenpan>t ann class="Chemical">alysis of body tissues of aquatic organisms, larvae of the chironomid midge Pseudodiamesa nivosa (Diptera: Chironomidae), the oligochaete Tubifex tubifex (Oligochaeta: Tubificidae), and the aquatic moss Warnstorfia exannulata (Hypnales: Amblystegiaceae) were collected with a biological bottom dredge (EFE & GB Nets, UK) in the RAS deep-water habitats (6–8 m depth). Larvae of the chironomid P. nivosa were also collected in RPD (0.5–1.5 m depth). The predatory aquatic beetle Agabus bipustulatus (Coleoptera: Dytiscidae) was collected with a kick net in shallow-water habitats (0.5–1.5 m depth) of both water bodies. Invertebrates were held in lake water at field temperatures for transport to the laboratory where they were sorted according to species, cleaned with distilled water, placed in Petri dishes and kept at 4 °C for 24 h in order to purge the guts. Thereafter they, as well as the aquatic moss W. exannulata, were rinsed with distilled water and frozen pending analyses. Only plastic storage vessels and tools (tweezers, scapulae, etc.) were used. All plastic labware was soaked in 1% nitric acid for at least 24 h and rinsed with doubly deionized water to leach out metal ions.[30]

n class="Chemical">All sedimenpan>t and n class="Species">biota samples were dried in an oven at 25 °C to a constant weight (3–4 days) and homogenized with an agate mortar and pestle, and triplicate 0.5 g aliquots of every sample were used for the analysis. The samples were dissolved with concentrated nitric acid[31] using the microwave digestion technique.[32] A total of 42 elements (Ag, Al, As, Ba, Bi, Br, Ca, Cd, Cl, Cr, Cu, Fe, Ga, Ge, Hf, I, K, Mg, Mn, Mo, Na, Nb, Ni, P, Pb, Rb, S, Sb, Se, Si, Sn, Sr, Ta, Th, Ti, Tl, U, V, W, Y, Zn, Zr) were assayed by energy dispersive X-ray fluorescence (EDXRF) analysis using calibration models based on fundamental parameter methods. In the sediment samples, ten major elements, Si, Al, Fe, Mn, Mg, Na, Ca, K, Ti, and P, were measured as their oxides. Analytical accuracy was verified by the use of CCRMP (Canadian Certified Reference Materials Project) reference samples. Recoveries were within 10% of the certified values.

Sediment Coring, Dating, and Chronology

Sediment cores were takenpan> un class="Chemical">sing UWITEC gravity and piston corers (UWITEC, Austria) in the deepest point of both water bodies in the summer of 2011. The 119.2 cm long sediment sequence from RAS was sectioned contiguously into 156 samples: at 0.22–0.50 cm resolution within the top 24 cm and at 1 cm resolution from 24 cm to the bottom. The 15.5 cm long sediment sequence from RPD was sectioned contiguously every 0.5 cm, resulting in a total of 31 samples. Chronological control for the sediment sequences was provided by accelerator mass spectrometry (AMS) radiocarbon dates derived from terrestrial plant macrofossils: six AMS 14C dates from the RAS sequence and four dates from the RPD sequence. The Bayesian statistical method of the OxCal software package version 4.2.1[33] was used to produce age-depth models (SI Figure S2). According to these estimates, the RAS and RPD sediment cores span time periods of ∼10 200 years and ∼3200 years, respectively. The ages are expressed in calibrated years before present (cal yr BP = calibrated years before AD 1950) (see SI 1 for further details).

Chironomid Analysis

Incidence of n class="Disease">mouthpart abnormalities inpan> chn class="Chemical">ironomid larvae was used as measures of toxic stress in the water bodies. Occurrence of mouthpart deformities in chironomids is a sublethal effect of larval exposure to environmental stress factors.[34] Since mouthpart deformities may also be induced by toxic elements and their compounds not included in routine chemical analyses, they may better reflect environmental quality than chemical analysis alone. Assessment of these morphological abnormalities offers an effective bioindication of toxic stress in modern freshwater ecosystems[34,35] and over previous time periods.[36−38] The mentum and other mouthparts of chironomid larvae damaged during the cleaning and mounting process usually have abrupt breaks that are clearly visible and easily distinguishable from deformed structures.[39]

The chn class="Chemical">ironomid-n class="Chemical">based paleolimnological study was designed to track and compare past changes in incidence of larval mouthpart deformities in identical chironomid taxa from RAS and RPD. To provide background information on the structure of the subfossil chironomid assemblages, sediment core samples from both water bodies were analyzed for species composition of the assemblages following standard procedures outlined in Brooks et al.[40] (see SI 2 for more details). Larval head capsules of the two most common chironomid taxa, Pseudodiamesa nivosa and Micropsectra radialis-type, found in both dredge and sediment core samples (live specimens and their subfossil remains, respectively) from RAS and RPD (SI Figures S1 and S3) were inspected for deformities of the mentum, a mouth part which is typically well preserved in subfossil material.

Three biometric variables, head length, head width, and mentum width, were men class="Chemical">asured for each head capsule. These variables were used to n class="Chemical">assign each head capsule to a specific larval stage of development (instar). A minimum of 100 third- and fourth-instar head capsules of each of the two chironomid taxa, P. nivosa- and M. radialis-type, were inspected for mentum deformities in each sediment sample at 250–400× magnification using a compound microscope. The severity of mentum deformities was categorized into three classes according to Lenat:[41] Class I—slight deformities, such as ’fused’ teeth; Class II—more conspicuous, moderate deformities, such as extra teeth, missing teeth, large gaps, and distinct asymmetry; and Class III—severe deformities, including at least two Class II characteristics. Head capsules of live larvae were treated and inspected for mentum deformities in the same manner as their subfossil remains.

Statistical Analyses

One-way ann class="Chemical">alyn class="Chemical">sis of variance (ANOVA), followed by Bonferroni post hoc test, was used to determine whether there are any significant differences in the elemental composition of water, sediment, and biota between the two adjacent water bodies, as well as between the shallow and deep-water habitats of RAS. All data were square root transformed to reduce the biasing effect of extremely low or high element concentrations. The nonparametric Mann–Whitney U-test was used to detect significant differences in the frequency of chironomid deformities between different time intervals in the Holocene. Percentage data were transformed to square roots to stabilize variances. SPSS version 17.0 software (SPSS Inc., Chicago, IL) was used.

Results and Discussion

Chemicn class="Chemical">al ann class="Chemical">alysis of water samples revealed that the ratios of dissolved inorganic nitrogen (DIN = NO3–-N + NH4–-N) to total phosphorus (TP) exceed 100 by mass in RAS and RPD (see SI 3 for more details). This suggests that phytoplankton communities in both water bodies are phosphorus-limited, defined by Bergström[42] as DIN:TP > 3.4 by mass. Physico-chemical analysis of water taken from RAS and RPD in 2010–2012 indicated that the striking differences in limnochemistry of these two neighboring water bodies, recorded in 2005,[18] persisted to the present day (Table 1). Although the water in both water bodies is dominated by the same ions (Ca2+, Mg2+, and SO42–), the total ion concentration and electrical conductivity in RAS (∼10 500 μequiv. l–1 and ∼510 μS cm–1, respectively) are more than an order of magnitude greater than in RPD (∼780 μequiv. l–1 and ∼45 μS cm–1, respectively). The concentrations of many dissolved minor and trace elements (Al, Cu, Mn, Ni, Sr, Zn) are negligibly small (one-way ANOVA, Bonferroni post hoc test, p < 0.05) in the circumneutral RPD (pH 6.19–6.64) compared to the acidic RAS (pH 4.93–5.86) (Table 1). The concentrations of dissolved iron are extremely low (<10 μg l–1) in both water bodies. However, in the acidic RAS, an extensive area of the lake bottom at 5.0–8.5 m water depth is coated with orange-colored iron oxyhydroxides (SI Figures S1). It is known that under mildly acidic conditions in surface waters affected by ARD, the soluble ferrous iron oxidizes to ferric iron, which precipitates as a ferric oxyhydroxide.[43]

Table 1

Physical and Chemical Properties of Water in RAS and RPD in the Early Open-Water Seasona

	RASc		RPD
parameterb	0–5 m (n = 6)	5–9 m (n = 4)	(n = 3)	EU limit valued
conductivity (μS cm–1 at 25 °C)	231 ± 46k	857 ± 117m	47 ± 6n	2500
total ions (μequiv·L–1)	4,185 ± 1169k	18,513 ± 2938m	779 ± 103n	NA
Ca2+ (mg L–1)	18 ± 6k	76 ± 10m	3 ± 1k	NA
Mg2+ (mg L–1)	13 ± 4k	63 ± 10m	0.7 ± 0.1n	NA
Na+ (mg L–1)	0.6 ± 0.1k	4.0 ± 0.4m	0.2 ± 0.1k	200
K+ (mg L–1)	0.3 ± 0.1k	1.1 ± 0.1m	0.2 ± 0.1k	NA
SO42– (mg L–1)	99 ± 28k	310 ± 124m	15 ± 3n	250
TPe (μg L–1)	1.4 ± 0.1k	1.6 ± 0.1k	1.5 ± 0.5k	NA
NO3–-Nf (μg L–1)	191 ± 23k	208 ± 14k	172 ± 22k	NA
NH4–-Ng (μg L–1)	18 ± 5k,m	35 ± 13m	5 ± 1k	NA
Al (μg L–1)	395 ± 120k	1,519 ± 126m	3 ± 1n	200
Cu (μg L–1)	4 ± 1k	9 ± 4k	<2m	2000
Fe (μg L–1)	1 ± 1k,m	7 ± 4m	<1k	200
Mn (μg L–1)	181 ± 45k	1,587 ± 609m	<1n	50
Ni (μg L–1)	107 ± 32k	468 ± 91m	2 ± 2n	20
Sr (μg L–1)	72 ± 19k	277 ± 37m	14 ± 5n	NA
Zn (μg L–1)	85 ± 23k	311 ± 29m	3 ± 2n	NA
pH	5.48 ± 0.11k	5.07 ± 0.08m	6.51 ± 0.14n	≥6.5

RAS data for 2011–2012, and RPD data for 2010–2012.

For each parameter, values (mean ± SE) that share common letters (k, m, or n) do not differ significantly (one-way ANOVA and Bonferroni post hoc test, p < 0.05).

Values are for the upper (0–5 m depth) and lower (5–9 m depth) parts of the RAS water column.

NA: not available, i.e. parameter is not regulated by the European Drinking Water Directive;[47] parameter values exceeding the appropriate European Union limit values for drinking water[47] are shown in bold.

TP = total phosphorus.

NO3–-N = nitrate-nitrogen.

NH4–-N = ammonium-nitrogen.

Rn class="Chemical">AS data for 2011–2012, and RPD data for 2010–2012.

For each parameter, vn class="Chemical">alues (meanpan> ± SE) that share pan> class="Chemical">common letters (k, m, or n) do not differ significantly (one-way ANOVA and Bonferroni post hoc test, p < 0.05).

Vn class="Chemical">alues are for the upper (0–5 m depth) anpan>d lower (5–9 m depth) parts of the Rpan> class="Chemical">AS water column.

n class="Chemical">NA: not available, i.e. parameter is not regulated by the Europeanpan> Drinking pan> class="Chemical">Water Directive;[47] parameter values exceeding the appropriate European Union limit values for drinking water[47] are shown in bold.

In the Rn class="Chemical">AS catchmenpan>t dominpan>ated by crystalline basement rocks with an inherently low acid buffering capacity, enhanced cryogenic weathering attributed to ice of the active rock glacier is the most likely mechanism causing groundwater inputs of acidic, metal- and sulfate-enriched water to the lake. Cryogenic weathering of the local bedrock, which involves a plethora of physical, chemical and biological processes (ice segregation, freeze–thaw cycles, volumetric expansion of water, hydration etc.),[44−46] results in the exposure of fresh bedrock surfaces to air and water in a complex rock glacier system and leads to oxidation of sulfide minerals and the generation of ARD.

A solute-rich inflow n class="Chemical">sinks, n class="Chemical">as a rule, down the water column due to its high density. This effect can result in physical and chemical stratification of the water column which reduces water circulation and aeration of deep-water layers. Our data show that strong stratification develops in RAS during periods of prolonged ice cover (8–9 months). After the ice breakup, the concentrations of major ions (Ca2+, Mg2+, SO42+) and minor and trace elements (Al, Mn, Ni, Sr, Zn) in the upper water layers (0–5 m) are four to seven times smaller than in the deeper waters (5–9 m), and the pH decreases from 5.48 near the surface to 5.07 at depth (Table 1 and Figure 2).

Figure 2

Depth profiles of selected physical and chemical properties of the RAS water. Electrical conductivity (a), pH (b), total ion (c), calcium (d), magnesium (e), and sulfate (f) profiles in the RAS water column on fourth July 2012 (closed circles; the lake was ∼70% ice covered) and 29th August 2012 (open circles; ∼1.5 months prior to lake ice freeze-up).

Depth profiles of selected phyn class="Chemical">sicpan> class="Chemical">al and chemical properties of the RAS water. Electrical conductivity (a), pH (b), total ion (c), calcium (d), magnesium (e), and sulfate (f) profiles in the RAS water column on fourth July 2012 (closed circles; the lake was ∼70% ice covered) and 29th August 2012 (open circles; ∼1.5 months prior to lake ice freeze-up).

Thus, the short growing sen class="Chemical">asonpan> inpan> Rn class="Chemical">AS begins under extremely unfavorable conditions for aquatic life, especially in the deep water, and the lake water may be dangerous for human consumption. In fact, the concentrations of manganese, nickel, aluminum, and sulfate in the deep water (5–9 m) exceed the appropriate EU limits for drinking water[47] 32, 23, 7, and 2 times, respectively (Table 1). Although solutes become more homogeneously distributed in the water column by late August/early September due to destratification by wind-induced mixing and convection (Figure 2), concentrations of manganese, nickel, and aluminum along a water depth gradient remain considerably higher (520–820, 140–220, and 440–620 μg L–1, respectively) than the EU maximum permissible levels in drinking water.

Totn class="Chemical">al n class="Chemical">metal concentration, however, does not necessarily correspond with metal bioavailability and toxicity to aquatic organisms.[48] The proportion of the total metal content that is available for uptake by biota depends greatly on the chemical form in which the metals are present. Metals in aquatic systems can exist in different chemical forms, including the dissolved inorganic complexes with dissolved anions (e.g., hydroxides, sulfates, carbonates) where metals occur as free hydrated metal ions, organic complexes with dissolved organic matter, and variety of particulate forms incorporated into the matrix of solid organic or mineral particles.[48] The free ionic form of metals is the most relevant for uptake by biota and hence the most toxic to aquatic life.[49,50] Hydrogen ion activity (pH) and redox potential greatly affect metal availability because of their strong influences on metal partitioning and speciation. Numerous studies have shown that the free metal ion concentrations increase with decreasing pH as well as upon oxidation of initially reduced environments.[48,51,52]

Thus, both processes, namely winter stran class="Chemical">tification resulting in low pH in the deep waters (pH ∼5.1), and oxygenation of the metal-rich hypolimnion and sediment during the open-water season, may result in greater bioavailability and toxicity of metals to biota in RAS.

Sediments and Biota

Sediments

The n class="Chemical">concenpan>trations of most metals in the sediments of RAS and RPD are comparable with their abundances in the Earth’s continental crust (Figure 3). Rather high concentrations of lead and zinc in sediments of both water bodies may be attributed to the site-specific geologic conditions, whereas elevated nickel and copper concentrations (244 and 179 μg g–1, respectively) in the RAS deep-water sediments are likely to be caused by enhanced accumulation of fine-grained particles in the deep water. These particles provide a large surface area for the sorption of metals at the water-sediment interface.[53]

Figure 3

Concentrations (mean ± SE; n = 3) of copper (a), zinc (b), lead (c), nickel (d), chromium (e), aluminum (f), iron (g), titanium (h), manganese (i), and vanadium (j) in surface sediments and biota of RPD and the RAS shallow and deep waters. For each metal, means followed by the same letter are not significantly different (one-way ANOVA, Bonferroni post hoc test, p < 0.05). The abundances of the chemical elements in the Earth’s continental crust[22] are shown with green lines.

n class="Chemical">Concenpan>tran class="Chemical">tions (mean ± SE; n = 3) of copper (a), zinc (b), lead (c), nickel (d), chromium (e), aluminum (f), iron (g), titanium (h), manganese (i), and vanadium (j) in surface sediments and biota of RPD and the RAS shallow and deep waters. For each metal, means followed by the same letter are not significantly different (one-way ANOVA, Bonferroni post hoc test, p < 0.05). The abundances of the chemical elements in the Earth’s continental crust[22] are shown with green lines.

Chironomids

n class="Chemical">Concenpan>tran class="Chemical">tions of 10 elements (Cu, Zn, Pb, Ni, Cr, Al, Fe, Ti, Mn, and V) were significantly higher (one-way ANOVA, Bonferroni post hoc test, p < 0.05) in the larvae of Pseudodiamesa nivosa, omnivores that feed on a wide variety of foods from detritus to small invertebrates,[54] collected from the RAS deep water than from RPD unaffected by the rock glacier solutes (Figure 3). This suggests a high bioavailability of these metals in the RAS deep-water habitats.

Aquatic Moss

The aquatic moss n class="Species">Warnstorfia exannulata, which prefers mildly acidic conditions,[55] have colonized an extensive area of the RAS bottom at 5–8 m water depth (SI Figures S1). Increased concentrations of nickel and aluminum in the W. exannulata tissues were comparable to those recorded in the chironomid larvae from the RAS deep-water habitats (Figure 3). Aquatic mosses are considered to be good indicators of metal contamination.[56] They are tolerant of metal contamination, long-lived, and have large cationic exchange properties within the cell wall. Owing to the lack of roots, metal uptake occurs straight from the water where they develop.

Oligochaetes

Among n class="Chemical">all n class="Chemical">benthic organisms analyzed from the RAS deep water, the highest concentrations of Cu, Zn, and Pb were recorded in the body tissues of the sediment-ingesting oligochaete Tubifex tubifex (one-way ANOVA, Bonferroni post hoc test, p < 0.05). The Cu concentrations in the T. tubifex tissues were ∼2.5 times greater (one-way ANOVA, Bonferroni post hoc test, p < 0.05) than that in surrounding sediments (Figure 3). Ingestion of contaminated sediment has been shown to be the major source of particle-sorbed metals to sediment-feeding organisms.[57]

Aquatic Beetles

The n class="Chemical">Cu n class="Chemical">concentrations in the predatory beetle Agabus bipustulatus, which is common in shallow-water habitats of both water bodies, were also significantly higher (one-way ANOVA, Bonferroni post hoc test, p < 0.05) in RAS than those from RPD (Figure 3). Results of a recent study[58] demonstrate that predatory aquatic beetles are capable of reflecting trace elements bioaccumulation in habitats with different disturbance levels, although their innate ability to take up/excrete metals may vary widely between species.

Mentum Deformities in Modern Chironomid Larvae

Inspecn class="Chemical">tionpan> of modernpan> Pseudodiamesa nivosa and n class="Species">Micropsectra radialis-type larvae, represented by live specimens and their recent remains in surface sediments, for deformities of mentum revealed the morphological abnormalities in the RAS P. nivosa population. No mentum deformities were noted upon inspection of P. nivosa from RPD and M. radialis-type from both water bodies. Although mentum deformities in Micropsectra larvae have been recorded at some metal-contaminated sites,[37,59] it is likely that the P. nivosa larvae are more susceptible to the harmful effects of acidic, metal-rich environments than the M. radialis-type larvae. Incidence of mentum deformities in the live P. nivosa larvae from RAS varied from 81% to 87% (n = 3) and was comparable with the frequency of mentum deformities among recent remains of P. nivosa in the RAS surface sediments (78–87%, n = 3). The observed deformities were categorized into two classes following Lenat:[41] Class I—fused median tooth (30–34%, n = 6), and Class II—large median gap (48–55%, n = 6) or extra teeth (<2%, n = 6) (Figure 4). Mentum abnormalities are generally related to a physiological disturbance during the molting process, as a somatic response to a wide range of environmental stress factors, but not to a genetic heritability of larvae.[60−62] In many polluted sites, the total incidence of deformities in chironomid populations usually varies between 20% and 50%,[41,63] but the incidence of abnormalities in the RAS P. nivosa population (78–87%) was as high as that recorded in a population of Chironomus spp. (∼83%), another chironomid taxon, in a site heavily polluted by radionuclides and trace metals.[64]

Figure 4

Micrographs (200× magnification) of larval menta of the chironomid Pseudodiamesa nivosa from the RAS modern and subfossil material: normal mentum (a) and deformed menta (b–e). The severity of the observed deformities was categorized into two classes following Lenat:[41] Class I (slight deformities)—fused median tooth (b, c), and Class II (more conspicuous deformities)—large median gap (d) and extra teeth (e). Scale bars = 0.1 mm.

Min class="Chemical">crographs (200× magnpan>ifican class="Chemical">tion) of larval menta of the chironomid Pseudodiamesa nivosa from the RAS modern and subfossil material: normal mentum (a) and deformed menta (b–e). The severity of the observed deformities was categorized into two classes following Lenat:[41] Class I (slight deformities)—fused median tooth (b, c), and Class II (more conspicuous deformities)—large median gap (d) and extra teeth (e). Scale bars = 0.1 mm.

Reconstruction of Temporal Trends in the Ecotoxicological State

The ann class="Chemical">alysis of chironomid remains in core sediment samples showed that a high total incidence of P. nivosa mentum deformities of ∼70% persisted in RAS during the last 2.5 millennia and was significantly lower (Mann–Whitney U-test, p < 0.0004), around 40%, in the early Holocene (∼10 150–9800 cal yr BP) (Figure 5). The incidence of the Class II mentum deformities (median gap or extra teeth) in P. nivosa from RAS was also significantly higher (Mann–Whitney U-test, p < 0.0008) in the late Holocene (∼49%) than in the early Holocene (∼21%). Unfortunately, the cold-stenothermal P. nivosa was absent in RAS, like other high alpine lakes,[65] through a relatively warm interval of the early and mid-Holocene (Figure 5, SI Figure S3), and this species cannot be used for reconstruction of temporal trends in the ecotoxicological state of RAS over the entire Holocene. In RPD, the frequency of P. nivosa mentum abnormalities did not exceed 2% (only the Class I mentum deformities was recorded) through the last 3.2 millennia, or, in other words, it did not exceed the frequency of chironomid deformities found in natural unstressed environments (up to 12–14%).[41,66]

Figure 5

Relative abundance of Pseudodiamesa nivosa in chironomid assemblages and the total incidence of P. nivosa mentum deformities in the sediment records from RPD (a) and RAS (b), and Holocene July air temperatures (TJuly) reconstructed for the Eastern Alps (c).[65]

Relan class="Chemical">tive abundance of Pseudodiamesa nivosa inpan> chn class="Chemical">ironomid assemblages and the total incidence of P. nivosa mentum deformities in the sediment records from RPD (a) and RAS (b), and Holocene July air temperatures (TJuly) reconstructed for the Eastern Alps (c).[65]

The results provide evidence that the present adverse n class="Chemical">conditions in the lake are not unprecedented in its Holocene history and cannot be associated exclusively with global warming and enhanced thawing of the rock glacier in recent years. It is likely that the development of the rock glacier started at the onset of the Holocene, when the area occupied by the modern rock glacier became deglaciated. The formation of RAS at the toe of the rock glacier most probably began at the onset of acidic drainage in the freshly deglaciated area.

Overn class="Chemical">all, our finpan>dinpan>gs illustrate that naturn class="Chemical">al ARD systems generated by active rock glaciers in mineralized areas with sulfide-bearing lithologies may seriously disturb lake ecosystems and endanger the quality of drinking water in pristine alpine environments. More adverse effects may be expected in lakes situated at the toe of greater and more active rock glaciers than the one in the RAS catchment.