Efficiency of frost-cracking processes through space and time: An example from the eastern Italian Alps

Geomorphology 232 (2015) 248–260

Contents lists available at ScienceDirect

Geomorphology journal homepage: www.elsevier.com/locate/geomorph

Efficiency of frost-cracking processes through space and time: An example from the eastern Italian Alps S. Savi a,⁎, R. Delunel b, F. Schlunegger b a b

University of Potsdam, Germany University of Bern, Switzerland

a r t i c l e

i n f o

Article history: Received 15 September 2014 Received in revised form 13 January 2015 Accepted 15 January 2015 Available online 23 January 2015 Keywords: Frost-cracking Sediment production Sediment supply Spatial and temporal denudation rate Climatic variations Holocene

a b s t r a c t It is widely accepted that climate has a strong impact and exerts important feedbacks on erosional processes and sediment transport mechanisms. However, the extent at which climate influences erosion is still a matter of debate. In this paper we test whether frost-cracking processes and related temperature variations can influence the sediment production and surface erosion in a small catchment situated in the eastern Italian Alps. To this extent, we first present a geomorphic map of the region that we complement with published 10Be-based denudation rates. We then apply a preexisting heat-flow model in order to analyze the variations of the frost-cracking intensity (FCI) in the study area, which could have controlled the sediment production in the basin. Finally, we compare the model results with the pattern of denudation rates and Quaternary deposits in the geomorphic map. The model results, combined with field observations, mapping, and quantitative geomorphic analyses, reveal that frost-cracking processes have had a primary role in the production of sediment where the intensity of sediment supply has been dictated and limited by the combined effect of temperature variations and conditions of bedrock preservation. These results highlight the importance of a yet poorly understood process for the production of sediment in mountain areas. © 2015 Elsevier B.V. All rights reserved.

1. Introduction During the past two decades, the extent at which climate influences erosion and limits the height of mountain ranges has been discussed in a controversial way, highlighting the difficulties in discriminating the effects of climate versus tectonic on denudation rates (Molnar and England, 1990; Raymo and Ruddiman, 1992; Egholm et al., 2009; Koppes and Montgomery, 2009; Thomson et al., 2010; Willenbring and von Blanckenburg, 2010; Egholm, 2013; Herman et al., 2013). More recently, the development of in situ produced, cosmogenicnuclide techniques has allowed a substantial enrichment of data quantifying denudation rates in different regions of the world and on multisecular to millennial time scales. Still, denudation rate values measured through cosmogenic nuclides include the effects of a series of processes (e.g., physical and chemical weathering, sediment production, etc.) and conditions (e.g., hillslope-channel connectivity, transport capability of streams and rivers, etc.), which could be difficult to disentangle from each other but which influence the landscape's response to environmental changes (Harvey, 2002). This is particularly the case for cold regions and mountain areas not glaciated nowadays, where periglacial processes including frost-cracking have been ⁎ Corresponding author. Tel.: +49 3319775856. E-mail address: [email protected] (S. Savi).

http://dx.doi.org/10.1016/j.geomorph.2015.01.009 0169-555X/© 2015 Elsevier B.V. All rights reserved.

proposed as a very efficient mechanism of bedrock erosion and sediment production (Anderson, 1998; Hales and Roering, 2005, 2007; Delunel et al., 2010, and references therein). As frost-cracking and related periglacial activities are mostly dependent on temperature-driven ice dynamics and water availability, they are strongly influenced by climatic oscillations. In this context, Anderson (1998) proposed a model in which the relationships between temperature gradients in bedrock and the depth at which frost cracking occurs were addressed. This author proposed that frost-cracking intensities increase at greater depths with decreasing surface temperatures. Hales and Roering (2007) expanded this model by considering hydrologic and heat-flow gradients in order to assess the role of frost-cracking and icesegregation mechanisms for the conditioning of rockfalls. They suggested that not only the intensity of the frost-cracking increases with depth, but also that the efficiency of the frost-cracking mechanism reaches a maximum where positive mean annual air temperatures (MAATs) of ca. 0 °C prevail. More recently, Anderson et al. (2013) explored the role of regolith production as a function of temperature variations and introduced the concept of limited-water circulation that significantly affects the magnitude of frost-cracking prediction in permafrost environments. They also reported that aspect-related microclimates can influence the transport efficiency on hillslopes via creeping, which partially influences the patterns and rates of denudation. Although these studies indicate an important control of

S. Savi et al. / Geomorphology 232 (2015) 248–260

temperature variations and frost-cracking processes on the production of sediments, only little research has been carried out to assess the role of these variables in a natural environment (Hales and Roering, 2005, 2009; Matsuoka, 2008; Delunel et al., 2010). Here, we focus on a formally glaciated catchment, located in the southeastern Alps of Italy, which we use to test whether frost-cracking processes could be invoked to explain the pattern of sediment production and denudation in this basin. To this extent, we present a geomorphic map of the region along with the results of 10Be-based denudation rates that we have published before. We then apply a model that allows us to calculate patterns of frost-cracking intensities in the studied catchment, following the approach proposed by Hales and Roering (2007). In a subsequent step, we reconstruct the temperature variations through the Holocene, based on a compilation of published climatic proxy data (Heiri and Lotter, 2005; Ilyashuk et al., 2011). We use this information to explore how climate changes might have influenced the temporal and spatial patterns of denudation rates in the basin if sediment production would have predominantly been accomplished by frost cracking processes.

249

2. Settings We conducted our study in the Zielbach catchment (33.5 km2) located in the southeastern Alps of Italy (Fig. 1), in which the spatial and temporal distribution of sediment supply was already reconstructed in previous publications (Savi et al., 2014a,b). 2.1. Background The Zielbach catchment can be divided into two distinct subbasins based on the main mechanism of sediment transfer (Savi et al., 2014a). In particular, a debris flow tributary is located in the southeastern margin of the basin and covers an area of ca. 2.3 km2. Basinwide denudation rates reported for this tributary average at ~ 1.6 mm/year, though highly variable in space (variance of 1.3). In contrast, the Zielbach main catchment (25.5 km2), representing N 90% of the whole basin's area, is characterized by the predominant occurrence of alluvial and fluvial processes that have primarily reworked glaciogenic material, which has been previously accumulated on the slopes. Averaged

Fig. 1. Zielbach catchment. (A) Inset showing the location of the study area in the eastern European Alps. (B) Lithological units cropping out in the catchment with indication of the major peaks' name. White dashed line delineates the nunataks and ridges above the maximum ice altitude, based on recognition on the field of trimline and glacial landforms. Dotted black lines delineate the watershed divide of the main Zielbach catchment (MZ-basin) and debris flow tributary (DF-tributary). The color and code on the channels is visualized in Fig. 2. Black crosses along the main channel indicate the location of the major knickzones. (C) Cross section of the A–A′ profile shown in (B) (modified from Bargossi et al., 2010). (D) Picture showing the marble outcrops on the Hochweiße and Lodner peaks.

250

S. Savi et al. / Geomorphology 232 (2015) 248–260

denudation rates for the main catchment are ca. 0.5 mm/year (variance of 0.02). Individual samples show a positive correlation between denudation rates and elevation, which is considered as the effect of the incorporation of soils with higher cosmogenic concentrations (Savi et al., 2014a). A direct influence of past glacial conditions on the denudation rates within the Zielbach catchment was excluded because of the short integration time derived from the 10Be analyses (calculations indicate that all the samples have reached a cosmogenic steady state within 2 ka from deglaciation). The temporal evolution of the catchment's sediment discharge has been documented by Savi et al. (2014b) based on 10Be concentrations and 14C ages of sediments collected from two drillholes located on the terminal fan. According to these authors, sediment discharge from the Zielbach catchment started at the end of the Younger Dryas cold phase (11.7 ka BP), and the supply rates of material appear to have initially followed a classical paraglacial scenario (Church and Ryder, 1972; Ballantyne, 2002) subsequently modulated by external perturbations. During the Holocene, the Zielbach catchment has not been glaciated and, following the production and transport of sediment in this time slice is not related to glacial processes. 2.2. Geology and geomorphology The bedrock geology of the catchment is part of the Texel complex, a subunit of the Austroalpine basement where the main lithologies comprise ortho- and paragneisses, mica schists, quartzites, and marbles (Fig. 1; Bargossi et al., 2010). Paragneisses are present in the lower part of the basin where small outcrops of quartzites can also be found. The central part of the catchment is occupied by a large lens of orthogneiss (Bargossi et al., 2010). In the upper part, paragneisses crop out in alternation with micaschists. In the northeastern corner of the basin, marbles form two high peaks of the study area (e.g., the Lodner peak, 3228 m asl, Fig. 1). Areas in which lithologies have no or little quartz content have been excluded from the calculation of the spatial averaged denudation rates (please refer to Savi et al., 2014a, for details on the calculations). The Zielbach main catchment is composed of several tributaries with U-shaped, cross-sectional geometries and generally low relief. The Zielbach trunk stream originates at ca. 2800 m asl and merges with the Adige River at an elevation of 505 m asl (Fig. 2). At ca. 1000 m asl,

the debris flow torrent that originates on the eastern N45° steep flank discharges into the Zielbach stream. In this tributary, debris flows have been usually triggered by rockfalls occurring on the south face of the Tschigat peak (Fig. 1) and are likely influenced by bedrock landsliding (see Savi et al., 2014a,b). 2.3. Modern and paleoclimate In the main valley bottom, the current climate is temperate with dry winters and rainy summers (Köppen, 1918). In the surrounding mountains, the climate has been classified as ‘Alpine’ (or ‘High-Mountain’ climate; Troll and Paffen, 1964), and mean annual air temperature (MAAT) changes with elevation following a lapse rate of 0.56 °C/ 100 m (Rolland, 2003). For the time-span 1981–2012, the nearby climatic station of Meran (333 m asl, ca. 7 km to the east of the basin) registered a MAAT of 11.6 °C (data from the Hydrographic Office of the Bolzano Province). Temperature variations over one year generally span from ca. 1.2 °C in winter (December–January) to ca. 21.5 °C in summer (July and August); extremes in cold and warm seasons have been registered in January 1985 (− 19 °C) and in July and August (+ 40 °C in 2003 and 2007). Total annual precipitation for the same time span averages to 800 mm. Highest rainfall rates and amounts are usually registered in early summer (June) and autumn (September and October) with cumulative monthly rainfall values of up to 120 mm. Peaks of high intensity, up to 105 mm/day, can also occur randomly during the year with no preferential seasonality. Despite the existence of mountain weather stations, no historical data on snowfall is available for neither the Zielbach catchment nor its direct vicinity. Nevertheless, when scaled to the Zielbach catchment, mean monthly air temperatures measured in Meran indicate that snowfall can occur between the months of October and April; whereas from May to September, the 0 °C isotherm is located above the maximum altitude of the basin (3337 m asl, Rotegg peak, Fig. 1). Additionally, snowfall can occur in the entire basin in the months of December and January only, when the 0 °C isotherm drops down to ca. 550 m asl. During the past glaciation, the region including the Zielbach catchment has been largely covered by ice (Penck and Brückner, 1909; Castiglioni, 1928; Bassetti and Borsato, 2005). At the scale of the Zielbach catchment, the glacial extent can be estimated based on the observations of trimline and glacially scoured outcrops within the catchment. From these observations we estimate that the basin was extensively glaciated up to an elevation of ~2800–3000 m asl, with only the highest summits emerging from the ice surface and thereby forming nunataks (Fig. 1). After the end of the Last Glacial Maximum (LGM), the glaciers receded to high elevation cirques, and the Zielbach catchment most likely became free of ice during the Late Glacial, as reported for several neighboring areas in the European Alps (Ivy-Ochs et al., 2006, 2009; Kelly et al., 2006; Delunel et al., 2014; Dielforder and Hetzel, 2014). Today, only a few remnants of these cirque-glaciers are present above 2700 m asl (Fig. 1). Following glacier fluctuations, the vegetation has colonized the catchment and currently comprises broad-leaved trees in the lower part of the basin and conifer forest at elevations N1000–1200 m asl, while the upper limit of the tree-line is found at ca. 2000 m asl. Small bushes, rhododendron and grassland predominate the landscape at higher altitudes up to ca. 2600 m asl, above which bare bedrock and talus deposits dominate the landscape. 3. Material and methods 3.1. Geomorphological mapping

Fig. 2. (A) Analysis of channels profiles; a and b represent alluvial channels while c, d, and e are debris flow torrents. (B) Gradient vs. drainage area plot. This graph has to be compared with (C) the diagram of Brardinoni and Hassan (2006) that illustrates the nature of the channels depending on the driving transport mechanisms.

We mapped the geomorphological architecture of the landscape using standard techniques including field observations, orthophoto analyses, and information from a 2.5-m digital elevation model (DEM). Field campaigns were made in the years 2009 and 2010. We mapped landforms such as channels, levees, alluvial fans, floodplains,

S. Savi et al. / Geomorphology 232 (2015) 248–260

and moraines and also identified the cover type of the hillslope deposits (such as regolith or forest). High-resolution orthophotos (horizontal ground resolution of 1 m; vertical resolution 4 m) and the 2.5-m DEM are available from the GIS platform of the Bolzano province (http:// gis2.provinz.bz.it/geobrowser/). These data have mainly been used to confirm field observations, to delineate geometries, and to map inaccessible parts of the basin at high elevation. We finally used the geological sheet ‘Merano’ as bases for mapping lithologies and tectonic features in the catchment (Bargossi et al., 2010; Italian Geological Service, Fig. 1). 3.2. Altitude-dependent domains The results of the geomorphological map have subsequently been used as a basis for delineating different altitude-dependent domains based on the nature, type of cover, and the relative areal extent (in percentage) of the Quaternary deposits. The main aim of this task was to explore the possible relationships between the spatial organization of the Quaternary deposits and the denudation rates reported by Savi et al. (2014a). For consistency, we used the same morphological division in subbasins applied in this previous study and then addressed whether the spatial pattern of erosion rates in the catchment could be related to altitudinal-dependent domains, as reported by Hales and Roering (2005) for a study conducted in New Zealand. Accordingly, each tributary was characterized by its mean elevation, which was then used for the presentation of the results (Section 4.2). Finally, we extracted geomorphic attributes of the catchment such as hypsometry, as well as slope and aspect distribution from the DEM to explore a possible control of the basin's morphometry on the denudation rates and the occurrence of Quaternary deposits. 3.3. Temperature reconstruction and frost-cracking modeling Temperature variations have been reconstructed using the proxy data provided by Ilyashuk et al. (2011) and Heiri and Lotter (2005). The first authors reconstructed July air temperatures back to the early Holocene using a chironomid-based study from a lake located ca. 30 km north of the Zielbach catchment. The latter authors provided a summer temperature reconstruction for the Holocene and the Late Glacial times using different aquatic organisms (such as chironomids and diatoms) from small Alpine lakes located ca. 50–100 km to the west. We incorporated the Heiri and Lotter (2005) data in our synthesis to complete our temperature reconstruction for the entire Holocene. We then scaled the results of these studies to the altitude of Meran (333 m asl) where current temperature records are available. To this extent, we used an altitudinal temperature lapse rate of 0.65 °C/100 m for the summer months, which was established by Rolland (2003) for the European Southern Alps. We also estimated uncertainties on the reconstructed temperatures as the ratio between the mean July temperature measured in Meran for the past 30 years and the one reported by Ilyashuk et al. (2011) for the past 100 years, scaled to the altitude of Meran. We then used these July temperature reconstructions to calculate the MAAT in Meran for the Holocene by extrapolating the observed ratio between the mean annual and July temperatures integrated over the past 30 years. Finally we used the reconstructed MAAT and a temperature lapse rate of 0.56 °C/100 m reported by Rolland (2003) to calculate the temperature conditions through time and space in the Zielbach catchment. We additionally reconstructed the distribution of vegetation through space and time following data reported by Tinner and Theurillat (2003), Nicolussi et al. (2005), and Tinner and Vescovi (2005). The scope of this effort was to address the potential effect of vegetation cover within the catchment on the erosion pattern. We used the data of these latter authors as they reconstructed treeline altitudes for the eastern Alps during the time span between the Late Glacial and the Holocene. In a subsequent effort, we used the temperature conditions reconstructed for the Zielbach as inputs for modeling the frost-cracking

251

intensity within the catchment. We based our analyses on the onedimensional heat-flow model proposed by Hales and Roering (2007) that followed the theoretical and experimental studies of Walder and Hallet (1985) and Anderson (1998). This model is based on a mechanism where the growth of segregation-ice lenses within the bedrock leads to the production of cracks at a rate that depends on the temperature gradient in the rock (see Anderson et al., 2013, and reference therein). In particular, this model considers the case where liquid water can migrate through a porous or fractured material toward cold parts where water freezes, allowing ice lenses to grow and eventually the rock to crack. It has been shown that the maximum ice-growth rate is reached within the frost-cracking window (Anderson, 1998; Fig. 3) where rock temperatures are between −3 and −8 °C and where free surface or groundwater is available to contribute to the growth of ice lenses (Hales and Roering, 2007). In order to estimate the frostcracking intensity at the catchment scale for different altitudes and for the present and different periods during the Holocene (Fig. 4), we first modeled the temperature conditions within the rock as a function of time (t) and depth (z) following the equation reported by Hales and Roering (2007) and modified from Anderson (1998): −z

T ðz;t Þ ¼ MAAT þ Ta  e

pffiffiffiffi π αP

rffiffiffiffiffiffiffiffiffiffi! 2π  t π −z sin P αP

where T is the temperature (°C); z is depth below the surface (cm); t is time (days); MAAT (°C) and Ta (°C) are the mean annual air temperature and half the absolute annual air temperature variation (set to 10.5 °C as currently observed in Meran), respectively, and are both extracted from the temperature reconstruction; α is thermal diffusivity (set to 1.5 mm2 s−1; 1–2 mm2 s−1 for most rocks in Anderson, 1998); and P is a one-year period on which the model is run. These data were used to estimate daily rock temperature curves as a function of depth for each catchment cell. It is important to highlight that we applied this model to the bedrock outcrop pixels only, thereby excluding those covered by detrital deposits or vegetation from our subsequent analyses. Thus, following Hales and Roering (2007), we calculated the depthintegrated, frost-cracking intensity (FCI hereafter, in °C/cm/year) as the integral of the absolute temperature gradient within the frost-cracking window (− 3 to − 8 °C), provided that free water is available either from the surface or from the groundwater (T N 0). The FCI were calculated at the catchment scale and also for some particular altitudes (Fig. 4). We acknowledge the recent work of Anderson et al. (2013) in which the authors further improved the frost-cracking model by introducing

Fig. 3. Schematic representation of the temperature distribution as a function of depth in bedrock and seasons (modified from Anderson, 1998). Dark area represents the frostcracking window, between −3 and −8 °C. Note that during a one-year cycle the thermal gradient changes with depth depending on initial MAAT.

252

S. Savi et al. / Geomorphology 232 (2015) 248–260

locations. On these slopes, screes and loose material, in the form of debris-cones or talus, are present at higher altitude below bedrock walls (Fig. 6).

Fig. 4. Variation of frost-cracking intensity (FCI) as a function of elevation at different times in situations of (A) negative and (B) positive MAATs, respectively. Note that in areas characterized by positive MAAT, the FCI is always larger than those exposed to negative MAAT. Note also that the depth of the maximum intensity is a function of temperature.

the effect of distance to unfrozen water, which would probably produce different results as it would not consider free water circulation in the ground but rather a limiting travel distance for it. This point will be discussed in further sections. We note that we only consider the elevation as parameter to calculate the local temperature conditions and the FCI. This simplification could lead to systematic biases in our calculations if the slope angles and aspects of bedrock, which affect local temperature, reveal a complicated pattern. However, at Zielbach this is not the case. In particular, Fig. 5 shows that (i) the 0 °C isotherm is situated at an elevation that has the most frequent occurrence, (ii) the slope angles are independent from the elevation, and (iii) the slope aspects are equally distributed through the basin and thus appear independent from the slope angle. These morphometric properties thus support our basic and simplified assumption of a direct dependency of frost-cracking processes to elevation. 4. Results 4.1. Geomorphological map and morphometric features of the catchment The geomorphological map, as illustrated in Fig. 6, shows the different deposits and landforms that were identified in the field and on orthophotos. This figure shows that in the lower part of the catchment the slopes on the northeastern and southwestern sides are generally steep (40° in average) and have a high relief (up to 2000 m in the debris flow tributary). In this part of the catchment, up to an elevation of ca. 1500 m. asl, Quaternary deposits can be observed only in the areas that have not been influenced by any obvious human activity. At intermediate elevations, boulders and small talus deposits can be identified. On the slope that hosts the debris flow tributary, large-scale bedrock deformation (deep-seated landslide) have produced talus deposits with boulders up to 20 m in diameter that can be found in different Fig. 5. Morphometric analysis of the catchment topography. (A) Altitudinal hypsometry showing the position of the 0 °C isotherm reconstructed for the present, the early, and the Neoglacial, respectively. Note that the shift toward lower or higher temperatures would change the proportion of the basin (%) that can be affected by frost-cracking processes. (B) Elevation frequency (gray line) and slope distribution (black line) as a function of elevation. The error bars represent the standard deviation of slope in 100 m elevation bins. We can observe that the slope does not vary much with altitude. Relative lower slopes (ca. 30°) between 2500 and 3000 m asl probably represent the preservation of glacial morphologies associated to an extensive glacial erosion during the LGM in this part of the basin. (C) Slope distribution as a function of aspect. The slightly elliptical shape of the graph reflects the orientation of the valley, which has a NNW–SSE axis. The absence of a significant spatial relationship between slope angles and aspects indicates that bedrock orientation does not bias the frost-cracking modeling results in this catchment.

S. Savi et al. / Geomorphology 232 (2015) 248–260

253

Fig. 6. Map of the Quaternary deposits observed in the catchment. Quaternary deposits have been classified on the basis of the geomorphology and type of cover.

Further upstream in the Zielbach main catchment, between the first and second knickzones (Fig. 2), the majority of the hillslopes are dominated by bedrock faces formed by the orthogneiss. In this segment, slopes are steep (40° in average) and display a relief of ca. 1300 m. Here, only a few debris cones and polygenetic colluvial fans occupy the basin. Screes are scarce and few rock glaciers can be observed above 2300 m. Most of the deposits are covered by trees or grass. The valley bottom hosts small floodplains that have served as temporary sediment storages. The catchment then widens above the second knick zone, and large tributaries confluence with the main Zielbach torrent. From this point upward, Quaternary deposits are more frequent in the basin, with vegetated slopes and talus covering the lower altitudes and loose material accumulating in the upper part. At an elevation of ca. 2500–2600 m asl, glaciogenic material and rocky deposits (including several rock glaciers) dominate the cover of the main tributaries. Screes, debris cones, and talus slope deposits are restricted to the areas located at the feet of steep bedrock walls (Fig. 6). Bedrock peaks and reaches occupy the highest portions of the basin. Field observations additionally allowed the characterization of different bedrock features. Overall, the basement is made up of metamorphic rocks with different degrees of deformation. Main foliations dip in a NNE direction (Fig. 7), as documented by the dip-direction of the Lodner north face peak. Changes in the dip-direction are related to folds of different generations that are visible at the micro- and macroscale, and they are better preserved in paragneisses and mica schists. Different degrees of bedrock fracturing seem to depend on elevation rather than on lithology. In particular, mountain crests and peaks

present a well-developed pattern of fractures responsible for the formation of rocky ridges (Fig. 7A) that eventually have supplied large rock boulders. Geometries and sizes of boulders mainly depend on lithology, with orthogneisses breaking into large and homogeneous boulders (several meter-scale), whereas paragneisses and micaschists fracture mainly along foliations, thereby forming large and flat blocks. Contrarywise, the bedrock presents well-preserved foliations but lessdeveloped fractures at lower altitudes (Fig. 7B). Roches moutonnées and glacially scoured outcrops are visible up to an elevation of 2800 m and are found on all lithologies. 4.2. Altitude-dependent domains On the basis of the geomorphological data, four distinct altitudedependent domains have been defined, (i) tree-covered, (ii) grasscovered, (iii) loose material, and (iv) bedrock (Fig. 8A). Glaciers, rock glaciers, floodplains, and alluvial or mixed fans have been excluded from the delineation of these domains, as we assume that they form intermediate sinks that have not contributed much to the overall sediment supply during the past years. Except for the data points representing the debris flow affected catchments (Fig. 8, empty symbols) where sediment transport is stochastic, the four domains show distinct altitude-dependent relationships. Above 2400 m asl, the areal extent of bedrock decreases with increasing elevation, while below this altitude any correlations between these variables break apart as implied by the large spread ranging between 40% and 83% (Table 1). A negative altitudinal relationship

254

S. Savi et al. / Geomorphology 232 (2015) 248–260

Fig. 7. Field photographs showing differences in bedrock features within the catchment. (A) Picture of the Tschigat Peak (see Fig. 1) that has been identified as a nunatak and presents highly fractured bedrock outcrops. Drawings highlight the main orientation of the foliations (plane with arrows), the fractures (dashed lines), and an example of a toppling stone (curved arrow in the forefront). The inset shows a closeup of outcrop condition on the crest of the Tschigat Peak (note the goats as scale). (B) Foliated bedrock on the valley bottom; photo taken at the highest knickzone.

can also be identified for the grass-covered areas above an elevation of ca. 2600 m. Likewise, from this same elevation, the spatial extent of tree-covered areas is negatively correlated to altitude. The relative

abundance of loose material, conversely, shows a positive correlation with elevation; i.e., the spatial extent of detrital materials increases as a function of altitude. Overall, the ensemble of the four domains

Fig. 8. Altitude-dependent and cover-type domains. (A) Overview of the distinct domains with uncolored areas delineating bedrock; the division in subcatchments follows Savi et al. (2014a). White circles represent the location of the cosmogenic samples that were used to infer the spatial denudation rate. (B) Denudation rate of Savi et al. (2014a). In (B), (C), and (D) empty symbols represent debris flow affected catchments. (C) Percentage of area covered by each domain in each of the subcatchments. (D) The same domains as in (C) plotted against erosion show the strong trends visible for the debris flow tributary and the main Zielbach catchment. Open symbols represent tributaries affected by debris flow 10Be signal. Note the logarithmic scale of the erosion rate (trend lines are also in the logarithmic form).

S. Savi et al. / Geomorphology 232 (2015) 248–260

255

256

S. Savi et al. / Geomorphology 232 (2015) 248–260

Table 1 Altitude-dependent domains based on cover-type. Tributarya Mean altitude m asl

Total area tributary km2

Bedrock area %

Loose material area %

Grass area %

Tree area %

Erosion ratea mm/year

20 16 15 18 19 11 6 14 8 3 4 1 2 5 7 13 10 9

5.79 6.78 13.57 14.17 15.37 17.91 22.51 21.26 24.54 25.53 0.63 31.29 31.70 0.58 0.37 0.69 0.87 1.54

45 54 51 51 52 51 53 52 54 54 60 50 50 55 83 75 40 50

36 28 29 29 27 26 24 25 22 22 8 19 19 32 4 3 22 7

12 14 14 15 16 18 17 17 16 15 17 15 15 17 3 6 18 12

– – – – – – – – 3 3 7 10 10 – 4 1 15 31

0.84 0.59 0.62 0.67 0.44 0.49 0.31 0.49 0.42 0.41 0.36 1.50 1.30 1.90 0.35 0.15 2.80 2.00

a

2777 2773 2755 2746 2714 2674 2667 2654 2564 2529 2380 2363 2339 2282 2267 2234 2149 1936

From Savi et al. (2014a).

shows a clear spatial pattern at elevations above ca. 2400 m asl, whereas below this elevation, the scatter of the data does not allow us to identify any distinct relationships. 4.3. Temperature variations and frost-cracking modeling The analysis of Holocene temperature variations in different locations of the eastern Alps (Heiri and Lotter, 2005; Ilyashuk et al., 2011) allowed the reconstruction of the temperature anomalies that affected the Zielbach catchment from ca. 13 ka BP until today (Table 2, Fig. 9). The lowest temperature reconstructed for this time span coincides with the Younger Dryas cold phase (~ 11.5 ka BP), with a MAAT in Meran of ca. 9.7 °C, almost 2° lower than at present. The data indicate an increase in MAAT within b2000 years, which leads to a positive temperature anomaly of 1.4 °C in the early Holocene (~ 9.5 ka BP). Subsequently, during the Holocene climatic optimum (at ca. 6 ka BP), temperatures are close to modern values, with a positive thermal anomaly of only 0.1 °C. The beginning of the Neoglacial (at ca. 4.5 ka BP) is then characterized by a drop in temperature of ca. 0.8 °C followed by a further 0.2 °C drop at around 2.5 ka BP. This negative anomaly of 1 °C corresponds to the lowest temperatures that were reached during the Holocene. At around 100 years, during the Little Ice Age, the temperature rose about 0.4 °C (negative anomaly of 0.6 °C), and then finally reached the modern values. The treeline variations (Fig. 9) closely follow the temperature trend, but with larger variations: whereas temperature variations reached a maximum of 3.3 °C, treeline altitudes have differed in the order of 1000 m between the Younger Dryas and the early Holocene. However, this corresponds to a negative altitudinal anomaly of only 500 m with respect to the present. The results of the temperature reconstruction through the Holocene have been subsequently used in the heat-flow model to assess the variation of the frost-cracking intensity (FCI) through space and time. In particular, when applied to the entire catchment, this model illustrates the spatial distribution of the FCI as a function of elevation. At the same time, we can map the FCI variations in space (altitude-dependent) and time (Fig. 10) when applying the model to MAATs reconstructed for the Late Glacial and Holocene time scales, illustrated in Fig. 9. Here we report the results of four distinct time slices: (i) the present; (ii) the Neoglacial (ca. 2 ka BP); (iii) the early Holocene (ca. 9.5 ka BP); and (iv) the Younger Dryas (ca. 11.5 ka BP). Because of its similarity with the modern climatic conditions (positive temperature anomaly of 0.1 °C), we do not report the results for the Holocene climatic

optimum period (ca. 6 ka BP), and we refer to the modern situation for this particular time slice. As already observed by Anderson (1998) and Hales and Roering (2007), the model shows different FCIs depending on the MAAT, where the intensities are higher for slightly positive temperatures (i.e., just above 0 °C) and lower in association with negative MAATs. Accordingly, in three of the four represented time periods (with the exception of the early Holocene), we observe the presence of a double frost-cracking intensity area, where the highest intensity of the lowerelevation area is associated with positive MAATs (referred to as Positive FCI hereafter), while the highest-intensity of the higher-elevation area represents the response to negative MAATs (referred to as negative FCI hereafter; Fig. 10). Similarly, the depths at which the frost-cracking processes operate most efficiently also depend on temperature, with maximum depths around 100 cm for positive MAATs and ca. 600 cm for negative temperatures (Table 2). These values slightly change when considering different initial MAATs. The model results yield nearly constant values for positive temperatures, with an averaged FCI = 602 ± 17 °C/cm/year and an averaged maximum depth of 98 ± 7 cm. In contrast, they show a stronger variability for negative MAATs, with an averaged FCI = 366 ± 151 °C/cm/year and an averaged maximum depth of 528 ± 194 cm. This variability is mainly due to the early Holocene time slice, when high temperatures caused a shift of the negative FCI area toward higher altitude, thereby reducing the area of the basin that could have been affected by stronger FCIs (Fig. 9). It is important to note that for negative MAATs the maximum depth of the FCI does not always correspond to the maximum intensity of the process. This is due to the shape of the FCI gradient that initially increases up to a maximum intensity and then decreases with depth (Fig. 4). Accordingly, for low MAATs the peak of maximum FCI corresponds to variable depths that range between 400 and 600 cm in our analyses. Another important result extracted from the model outputs is the indication of the altitude at which the temperature is constantly within the frost-cracking window (365 days/year within the temperature range of − 3/− 8 °C). Although permafrost forms when the MAAT is below 0 °C for more than two consecutive years (Blanchet and Davison, 2012), this altitude can be taken as a proxy to delineate a conservative elevation at which permafrost could have had a control on sediment production and sediment transfer (Fig. 10). We are aware that the presence of the double FCI area is the result of the model when free circulation of water is inferred, which we justify here through the relatively high porosity of gneiss and schists caused by deep fractures, cracks, and faults. In an alternative model, Anderson et al. (2013) demonstrated that limits for the travel distance of groundwater (possibly more appropriate for slates and shales with low permeability) cause the peak of the FCI to disappear when air temperatures turn positive. Also, they suggest that the intensity and depth at which frost-cracking occurs are controlled by the combined effect of the mean annual air temperature and the annual temperature oscillation. According to their results, in nonpermafrost conditions the most intense frost-cracking activity generally occurs in spring (when liquid water is freshly available from snow melting) and will most likely affect the first 2 m of the surface. 5. Discussion 5.1. Frost-cracking as a mechanism for the production of sediment: relationships with the spatial pattern of erosion rates When considering the present-day situation, the data suggest a strong relationship between altitude-dependent temperature variations and spatial location of screes and loose material on hillslopes. Particularly, Fig. 9 highlights that screes are mostly located below the negative FCI area where the conditions for the generation of sediments are favorable (negative FCI peak and presence of weathered bedrock). Regardless of the positive FCI area present at lower altitudes, the

S. Savi et al. / Geomorphology 232 (2015) 248–260

257

Table 2 Reconstructed paleotemperature and treeline altitudes together with the results of the frost-cracking model for the four chosen time slices.

Present Neoglacial

HCO Early Holocene

YD a

Age ky BP

Reconstructed MAATa °C

Temperature anomaly °C

treeline altitude m asl

treeline anomaly m

Altitude of 0 °C isotherm m asl

Minimum altitude of FCW m asl

Max FCI for positive MAAT °C/cm/year

Depth of max FCI (N0 °C) cm

Altitude of max FCI (N0 °C) m asl

Max FCI for negative MAAT °C/cm/year

Depth of max FCI (b0 °C) cm

0.03–0 0.1–0.03 2.5–0.1 4.5–2.5 8.0–4.5 8.2–8.0 10.0–8.5 10.2–10.0 11.5–10.5 12.5–11.5 13.0–12.5

11.6 11.0 10.6 10.9 11.7 11.1 13.0 11.4 11.2 9.7 10.8

0 −0.6 −1.0 −0.7 0.1 −0.5 1.4 −0.2 −0.4 −1.9 −0.8

2000 2100 2100 2150 2400 2250 2450 2200 2300 1500 1800

0 100 100 150 400 250 450 200 300 −500 −200

2404 2297 2226 2279 2422 2315 2654 2369 2333 2065 2262

3000 – 2800 – – – 3200 – – 2600 –

618 – 607 – – – 593 – – 588 –

105 – 99 – – – 91 – – 98 –

2400 – 2200 – – – 2600 – – 2000 –

360 – 430 – – – 215 – – 461 –

596 – 614 – – – 334 – – 567 –

Temperature reconstructed for Meran (333 m asl). FCW = frost-cracking window; FCI = frost-cracking intensity.

positive correlation between the relative extent of detrital deposits and elevation implies that the mechanisms responsible for the production of sediment are less efficient at lower altitudes where the bedrock conditions limit their activity (e.g., better-preserved bedrock and thus lower porosity and lower erodibility). This could also be compatible with the hypothesis of Anderson et al. (2013), which would imply penned water conditions and therefore the presence of a unique FCI area (the one at higher elevations). Another possible explanation for the lack of unconsolidated material at lower elevations could be related to a faster evacuation of hillslope-derived material. However, we discard this hypothesis in view of the erosion rate pattern (see paragraph below). Furthermore, the occurrence of the largest areas of unconsolidated sediments below the negative FCI area and the permafrost-line provides additional supports for the proposed link between temperature conditions and sediment production, as also suggested by Hales and Roering (2005, 2009). Our data additionally show a strong correspondence between the occurrence of unconsolidated material, the distribution and type of

vegetation cover, and the downstream decreasing trend in the modern erosion rate pattern previously published (Fig. 8; Savi et al., 2014a). In particular, the denudation rates and the spatial extent of unconsolidated material are positively correlated to elevation, whereas the vegetation cover is negatively correlated to this variable. These relationships are well represented in Fig. 8D, where a positive correlation between the pattern of denudation rates, screes, and vegetation is clearly visible. Generally, a high denudation rate, measured with in situ 10Be for stream sediments, requires that sediment production sites (here sites where frost-cracking occurs) and streams are directly connected. Indeed, frost-cracking processes deliver the entire range of grain size from boulders to sands as rock particles continuously disintegrate. This material is then supplied to the channel network through a cascade of various hillslope processes including creep, landslides, and overland flow erosion (Delunel et al., 2010). The 10Be concentration of the sampled bedload material and the calculated denudation rates could then be considered as the combined effect of the ensemble of these processes. Accordingly, while the environmental variables favor the occurrence of high

Fig. 9. Temperature and treeline anomalies reconstructed for the Holocene. The interpretation of the paleo-erosion rates inferred by Savi et al. (2014b) are additionally visualized. While the first phases of rapid paleo-erosion rates can be most likely interpreted through the evacuation of non-consolidated material after the retreat of the LGM glaciers (Savi et al., 2014b), the increase in 10Be-based denudation rates from sub-Atlantic times could potentially be explained by a higher frost-cracking intensity. However, it is important to highlight that, because of the presence of the debris flow tributary, whose activity has been likely promoted by the deep-seated landslide, the sediment supplied to the Zielbach fan could derive from the combined effects of FCI variations and pulses in the landslide movement (Savi et al., 2014b).

258

S. Savi et al. / Geomorphology 232 (2015) 248–260

Fig. 10. Frost-cracking model results in the four different time slices (black line delimitates the Zielbach catchment). Zero °C isotherm is the boundary between lower and higher FCI areas and correspond to the passage between dark and light colors (colored scale is linear). Note, at present and at the end of the Younger Dryas, the spatial distribution of the loose material (either fresh or vegetated) directly below the higher FCI area and the permafrost line (dashed line). The cumulative effect of multiple FCI migrations on bedrock would be stronger at elevations N2600 m asl, where bedrock has been affected by frost-cracking conditions for longer periods.

denudation rates at higher elevations, they deteriorate the erosional conditions moving downstream. We explain this through the presence of larger areas covered by vegetation at lower elevations, which contributes to the stabilization of hillslopes and sediment and thus to the lowering of the soil erosion rate. This finally contributes to decrease the pattern of the erosion rate in the downstream direction, as observed in the Zielbach main catchment. Finally, acknowledging that the influence of glacial inheritance for the explanation of this trend has already been excluded (Savi et al., 2014a) and that vegetation also plays a role in decreasing erosion, we discard the hypothesis of a more efficient erosional mechanism to explain the lack of loose material at lower altitudes in this portion of the basin. Accordingly, the increase in the relative abundance of loose material toward higher elevations most likely reflects a higher efficiency of sediment production at higher altitudes, thus confirming the importance of frost-cracking processes in the initiation of the sediment cascade.

characterized by steep slopes and a high relief, promotes a stronger connectivity between hillslopes and channels and implies a faster mobilization of the detrital material through debris flow processes (Bracken et al., 2014). This particular situation of high connectivity between sources and sinks could explain the lack of large areas of unconsolidated material on the hillslopes and the larger scatters in the data (Fig. 8D). This does not imply that sediment is not produced or has a lower production rate, but rather that all the material that forms on the slopes is mobilized in a relatively short time. In addition, it is important to note the potential role of the deep-seated landslide that affects the slope of this tributary, whose movement, even if slow, could have increased the amount of mobile material and created a preexisting condition (steep bedrock sectors) favorable to rockfalls (which could, in turn, facilitate the action of frost-cracking).

5.2. Differences between the debris flow tributary and the Zielbach main catchment

Accordingly to what has been discussed in the previous sections, the vegetation cover also plays a role influencing the pattern of denudation rates through space and time. Vegetation stabilizes hillslopes, therefore reducing the amount of material that can be produced and incorporated into the main channel network. Because vegetation develops as a function of temperature and altitude, it can influence the denudation rate pattern only at low elevations. In the Zielbach catchment, the areas covered by vegetation (grass, bushes, and trees) are mainly b20% in each of the tributaries. Moreover, Savi et al. (2014a) reported soil erosion rates of one order of magnitude lower than the ones

Following what has been mentioned above, a possible explanation for the scatter between the various variables presented in Fig. 8 is related to the contrasting sediment transport mechanisms between the Zielbach main catchment and the debris-flow tributary basin (alluvial versus debris-flow processes), where a more efficient mechanism of sediment evacuation possibly explains why large areas of scree deposits are rare on the hillslopes. The morphology of this basin,

5.3. Criticality of vegetation and lithology

S. Savi et al. / Geomorphology 232 (2015) 248–260

registered in the channels, and indicate a soil-depth between 30 and 80 cm. For this reason we can assume that the contribution of the sediment production along hillslopes through regolith formation and erosion is low in comparison with the production through frost-cracking processes. However, we cannot directly quantify the contribution of soil formation or erosion during the Holocene, when the treeline oscillated over a space scale of 1000 m. Although this aspect is of broad interest and importance, the elaboration of further relationships in this field goes beyond the scope of this study. The erodibility of the different lithologies and their degree of weathering could also influence the production of sediment through space. As already mentioned in Section 4.1, the frequency and spacing between fractures seems to depend on altitude rather than on lithologies. Also, the surface hardness of the main lithologies present in the catchment has been tested with a Schmidt hammer, with more than 400 measurements (Spada, University of Milan-Bicocca, personal communication, 2011). Averaged R-values are 43, 49, and 52 for mica schists, paragneisses, and orthogneisses, respectively. Although mica schists yield a slightly lower hardness index (Fig. 6), these lithologies could have had an effect on the sediment produced only in tributary 20 (Fig. 8). Paragneisses and orthogneisses show similar rockresistance values and crop out in the majority of the catchment. It thus appears that lithology does not strongly influence the production and release of sediment. 6. Conclusions: variations through time and further research needs Fig. 10 illustrates how the two FCI areas shifted through time depending on the initial differences in MAAT, thereby influencing the proportion of bedrock that can be subjected to frost-cracking processes. The highest intensity of frost-cracking is associated with positive MAATs, with temperatures close to 0 °C. This is also the area where frost-cracking processes operate at their shallower depths (max. ~ 100 cm), making this mechanism extremely efficient in terms of sediment production. However, because positive MAATs are generally found at lower elevations, the process may be limited by vegetation cover and better preserved bedrock. The second peak of FCI is located in the highest portions of the basin, and its efficiency depends on the local MAAT. In particular, during a warmer climate, the shift of this negative FCI peak toward higher elevations results in a lower efficiency of this process because the maximum intensity of the frost-cracking falls beyond the altitudinal range of the basin. Conversely, during a colder climate, the efficiency of frostcracking processes will be higher, especially in peaks and crests areas where the bedrock has a higher degree of fractures. However, under these conditions, permafrost and extremely low MAATs would prevent a fast release and transport of material. Interestingly, we observe that all mountain peak reaches have been constantly above the 0 °C isotherm since the Late Glacial period. These highest elevations have therefore always been affected by negative MAATs. In addition, as these peaks and crests consist in nunataks that were located above the surface of the main local glaciers, we speculate that these high-altitude bedrock segments could have been frequently exposed to frost-cracking processes during most of the Quaternary, which could have had a major impact on preexisting weakness zones. The observation of highly weathered and fractured outcrops on the main peaks and ridges strengthens our speculations. This aspect is important in terms of sediment supply because these bedrock portions could have potentially produced and delivered sediment over long time scales. However, this hypothesis could be challenged because (i) frost conditions (permafrost) and extremely freezing MAATs would have prevented a constant production and release of material (Blanchet and Davison, 2012) and (ii) the possible presence of firn and/or nonerosive icecaps on these landforms (Thomson et al., 2010) would have also strongly limited the efficiency of frost-cracking processes. Nevertheless, the rapid increase in temperature that occurred at the end of the Younger Dryas could have strongly reduced

259

the buffering effect of permafrost, increased the amount of liquid water availability, and promoted a faster release and transport of sediment, as Savi et al. (2014b) have already documented. Indeed, it is possible that the fast denudation rates, identified here for the Early Holocene (Fig. 9), mainly reflects the evacuation of possibly large volumes of glaciogenic material as soon as LGM glaciers retreated from the region (Savi et al., 2014a). After the removal of these sediments, shifts in temperatures and related changes in negative FCI could potentially explain the variations in paleodenudation rates, possibly since Subboraeal times. In support of this hypothesis, data calculated through the Holocene in Europe and North America (Siewert et al., 2012, and references therein) suggested rates of bedrock erosion that are similar to the paleodenudation rates inferred by Savi et al. (2014b). We acknowledge that these interpretations are primarily based on considerations of temperature changes. A complete analysis of the mechanisms proposed here requires the consideration of precipitation variations during the Holocene. However, paleoprecipitation data are among the most difficult information to obtain, and most of the studies conducted in the Alps generally report variable results (Kerschner and Ivy-Ochs, 2008; Ortu et al., 2008; Ilyashuk et al., 2011). Finally, it is important to highlight that, assuming a penned water condition as proposed by Anderson et al. (2013), the dependency of frost-cracking processes on temperature would be strengthened. The positive FCI area would probably disappear, thus giving more importance to the production of sediment in the negative FCI area and less to the conditions of bedrock preservation. However, if this model is considered, major importance needs to be given to the presence and circulation of deep and superficial water and to the seasonality of precipitation. Despite the drawbacks set to our assumptions, our attempt of linking temperature and vegetation variations with paleodenudation rates shows that correlations between these variables are feasible, at least for the past few thousands of years.

Acknowledgments We greatly appreciate the financial support of the Swiss National Science Foundation (project No. PBBEP2_146196 fellowships for prospective researchers) and we would like to acknowledge the efforts made by anonymous reviewers and by the Editor (Richard A. Marston), who helped improving our manuscript.

References Anderson, R.S., 1998. Near-surface thermal profiles in Alpine bedrock: implications for the frost weathering of rock. Arct. Alp. Res. 30 (4), 362–372. Anderson, R.S., Anderson, S.P., Tucker, G.E., 2013. Rock damage and regolith transport by frost: an example of climate modulation of the geomorphology of the critical zone. Earth Surf. Process. Landf. 38, 299–316. Ballantyne, C.K., 2002. A general model of paraglacial landscape response. The Holocene 12 (3), 371–376. Bargossi, G.M., Bove, G., Cucato, M., Gregnanin, A., Morelli, C., Moretti, A., Poli, S., Zanchetta, S., Zanchi, A., 2010. Note Illustrative Della Carta Geologica d'Italia. Foglio 013 — Merano. ISPRA, Servizio Geologico d'Italia. Bassetti, M., Borsato, A., 2005. Evoluzione geomorfologica della Bassa Valle dell‘Adige dall’Ultimo Massimo Glaciale: sintesi delle conoscenze e riferimenti ad aree limitrofe. Studi Trent. Sci. Nat. Acta Geol. 82 (31–42) (ISSN 0392–0534). Blanchet, J., Davison, A.C., 2012. Statistical modelling of ground temperature in mountain permafrost. Proc. R. Soc. A http://dx.doi.org/10.1098/rspa.2011.0615. Bracken, L.J., Turnbull, L., Wainwright, J., Bogaart, P., 2014. Sediment connectivity: a framework for understanding sediment transfer at multiple scales. Earth Surf. Process. Landf. http://dx.doi.org/10.1002/esp.3635. Brardinoni, F., Hassan, M.A., 2006. Glacial erosion, evolution of river long profiles, and the organization of process domains in mountain drainage basins of coastal British Columbia. J. Geophys. Res. 111, F01013. Castiglioni, B., 1928. Ghiacciai delle Venoste Orientali. Boll. Com. Glacio. 8, 91–165. Church, M., Ryder, J.M., 1972. Paraglacial sedimentation: a consideration of fluvial processes conditioned by glaciation. Geol. Soc. Am. Bull. 83, 3059–3071. Delunel, R., van der Beek, P.A., Carcaillet, J., Bourlès, D.L., Valla, P.G., 2010. Frost-cracking control on catchment denudation rates: insights from in situ produced 10Be concentrations in stream sediments (Ecrins–Pelvoux massif, French Western Alps). Earth Planet. Sci. Lett. 293, 72–83. Delunel, R., van der Beek, P., Bourlès, D., Carcaillet, J., Schlunegger, F., 2014. Transient sediment supply in a high-altitude Alpine environment evidenced through a 10Be budget

260

S. Savi et al. / Geomorphology 232 (2015) 248–260

of the Etages catchment (French Western Alps). Earth Surf. Process. Landf. 39, 890–899. Dielforder, A., Hetzel, R., 2014. The deglaciation history of the Simplon region (southern Swiss Alps) constrained by 10Be exposure dating of ice-molded bedrock surfaces. Quat. Sci. Rev. 84, 26–38. http://dx.doi.org/10.1016/j.quascirev.2013.11.008. Egholm, D.L., 2013. Erosion by cooling. Nature 504, 380–381. Egholm, D.L., Nielsen, S.B., Pedersen, V.K., Lesemann, J.E., 2009. Glacial effects limiting mountain height. Nature 460, 884–888. Hales, T.C., Roering, J.J., 2005. Climate-controlled variations in scree production, Southern Alps, New Zealand. Geology 33, 701–704. Hales, T.C., Roering, J.J., 2007. Climatic controls on frost cracking and implications for the evolution of bedrock landscapes. J. Geophys. Res. 112, F02033. http://dx.doi.org/10. 1029/2006JF000616. Hales, T.C., Roering, J.J., 2009. A frost “buzzsaw” mechanism for erosion of the eastern Southern Alps, New Zealand. Geomorphology 107, 241–253. Harvey, A.M., 2002. Effective timescales of coupling within fluvial systems. Geomorphology 44, 175–201. Heiri, O., Lotter, A.F., 2005. Holocene and Lateglacial summer temperature reconstruction in the Swiss Alps based on fossil assemblages of aquatic organisms: a review. Boreas 34, 506–516. Herman, F., Seward, D., Valla, P.G., Carter, A., Kohn, K., Willett, S.D., Ehlers, T.A., 2013. Worldwide acceleration of mountain erosion under a cooling climate. Nature 504, 423. Ilyashuk, E.A., Koinig, K.A., Heiri, O., Ilyashuk, B.P., Psenner, R., 2011. Holocene temperature variations at a high-altitude site in the Eastern Alps: a chironomid record from Schwarzsee ob Sölden, Austria. Quat. Sci. Rev. 30, 176–191. Ivy-Ochs, S., Kerschner, H., Reuther, A., Maisch, M., et al., 2006. The timing of glacier advances in the northern European Alps based on surface exposure dating with cosmogenic 10Be, 26Al, 36Cl, and 21Ne. Geol. Soc. Am. Spec. Pap. 415. Ivy-Ochs, S., Kerschner, H., Maisch, M., Christl, M., Kubik, P.W., Schlüchter, C., 2009. Latest Pleistocene and Holocene glacier variations in the European Alps. Quat. Sci. Rev. 28, 2137–2149. Kelly, M.A., Ivy-Ochs, S., Kubik, P.W., von Blanckenburg, F., Schlüchter, C., 2006. Chronology of deglaciation based on 10Be dates of glacial erosional features in the Grimsel Pass region, central Swiss Alps. Boreas 35, 614–643. Kerschner, H., Ivy-Ochs, S., 2008. Paleoclimate from glaciers: example from the Eastern Alps during the Alpine Lateglacial and early Holocene. Glob. Planet. Chang. 60, 58–71. Köppen, W., 1918. Klassifikation der Klimate nach Temperatur, Niederschlag und Jahresablauf. Petermanns Geogr. Mitt. 64 (193–203), 243–248. Koppes, M.N., Montgomery, D.R., 2009. The relative efficacy of fluvial and glacial erosion over modern to orogenic timescales. Nat. Geosci. 2, 644–647.

Matsuoka, N., 2008. Frost weathering and rockwall erosion in the southeastern Swiss Alps: long-term (1994–2006) observations. Geomorphology 99 (1–4), 353–368. Molnar, P., England, P., 1990. Late Cenozoic uplift of mountain ranges and global climate change: chicken or egg? Nature 346, 29–34. Nicolussi, K., Kaufmann, M., Patzelt, G., van der Plicht, J., Thurner, A., 2005. Holocene treeline variability in the Kauner Valley, Central Eastern Alps, indicated by dendrochronological analysis of living trees and subfossil logs. Veg. Hist. Archaeobot. 221–234. Ortu, E., Peyron, O., Bordon, A., Louis de Beaulieu, J., Siniscalco, C., Caramiello, R., 2008. Lateglacial and Holocene climatic oscillations in the South-western Alps: an attempt at quantitative reconstruction. Quat. Int. 90, 71–88. Penck, A., Brückner, E., 1909. Die Alpen in Eiszeitalter. vol. III. Tauschnitz ed, Lipsia. Raymo, M.E., Ruddiman, W.F., 1992. Tectonic forcing of late Cenozoic climate. Nature 359, 117–122. Rolland, C., 2003. Spatial and seasonal variations of air temperature lapse rates in Alpine Regions. J. Clim. 16, 1032. Savi, S., Norton, K., Picotti, V., Brardinoni, F., Akçar, N., Kubik, P.W., Delunel, R., Schlunegger, F., 2014a. Effects of sediment mixing on 10Be concentrations in the Zielbach catchment, central-eastern Italian Alps. Quat. Geochronol. 19, 148–162. Savi, S., Norton, K., Picotti, V., Akçar, N., Delunel, R., Brardinoni, F., Kubik, P.W., Schlunegger, F., 2014b. Quantifying sediment supply at the end of the last glaciation: dynamic reconstruction of an alpine debris-flow fan. Geol. Soc. Am. Bull. 126 (5–6), 773–790. http://dx.doi.org/10.1130/B30849.1. Siewert, M.B., Krautblatter, M., Hvidtfeldt, Christiansen H., Eckerstorfer, M., 2012. Arctic rockwall retreat rates estimated using laboratory-calibrated ERT measurements of talus cones in Longyeardalen, Svalbard. Earth Surf. Process. Landf. 37, 1542–1555. Thomson, S.N., Brandon, M.T., Tomkin, J.H., Reiners, P.W., Vásquez, C., Wilson, N.J., 2010. Glaciation as a destructive and constructive control on mountain building. Nature 467, 313–317. Tinner, W., Theurillat, J.P., 2003. Uppermost limit, extent, and fluctuations of the timberline and treeline ecocline in the Swiss Central Alps during the past 11,500 years. Arct. Antarct. Alp. Res. 35 (2), 158–169. Tinner, T., Vescovi, E., 2005. Ecologia e oscillazioni del limite degli alberi nelle Alpi dal Pleniglaciale al presente. Studi Trent. Sci. Nat. Acta Geol. 82, 7–15. Troll, C., Paffen, K.H., 1964. Karte der Jahreszeitenklimate der Erde. Erdkunde 18, 5–28. Walder, J., Hallet, B., 1985. A theoretical model of the fracture of rock during freezing. Geol. Soc. Am. Bull. 96, 336–346. Willenbring, J.K., von Blanckenburg, F., 2010. Long-term stability of global erosion rates and weathering during late-Cenozoic cooling. Nature 465, 211–214.