Abstract
We report on the key physical features of an active rock glacier that influence the distribution of plants and arthropods. We also perform a comparison with neighboring scree slope and alpine grassland to test whether the environmental features of the rock glacier drive the presence of specific species assemblages. Compared with scree slope and grassland, the studied rock glacier provides particular physical features that determine the presence of unique species. Plant distribution is mainly driven by grain size. Arthropod distribution is linked to grain size, with cold-adapted species found on areas with coarse-grained deep debris, which also shows a distinctive temperature regime with very low values throughout the year. On the basis of these findings, we advance the hypothesis that rock glaciers provide specific ecological conditions creating potential refugia for cold-demanding species during warm climatic periods.
Introduction
The importance of geomorphological heterogeneity for enhancing biodiversity in Alpine ecosystems has often been acknowledged, as well as its important ecological and biogeographical role in response to climatic fluctuations (Baroni et al., 2007; Caccianiga et al., 2011; Gobbi et al., 2011; Matthews, 1992). A specific landform with its distinctive environmental conditions may promote the survival of plant and animal taxa when the surrounding habitats become climatically unfavorable and may thus act as a refugium (Stewart et al., 2010). Therefore, environmental heterogeneity within high-altitude landscape could make the difference between expected survival and extinction of both animal and plant taxa (Birks and Willis, 2008; Essl et al., 2011).
Rock glaciers are a periglacial landform characterized by distinctive environmental conditions because of the occurrence of subsurface ice (permafrost). Previous studies outlined the occurrence of plant cover on their surface (Burga et al., 2004; Rieg et al., 2012); a recent paper reports arthropod assemblages associated with rock glaciers on Californian mountains (Millar et al., 2014). To our knowledge, no data on the presence of arthropod assemblages on Alpine rock glaciers are presently available. Furthermore, so far no study has tried to address the relationship between environmental conditions and plant and arthropod assemblages on these landforms. Rock glaciers, with respect to their surrounding habitats, offer a unique environment, and we test their suitability for the survival of high-altitude plant and arthropod taxa for the first time.
The research aims to describe plant and ground-dwelling arthropod (spiders and ground beetles) assemblages on an active rock glacier of the Alps and to compare them with those of neighboring habitats (scree slope and grassland). We also discuss the role of active rock glaciers as potential refugia for high alpine taxa during warm climatic phases.
Materials and methods
Study area
The Amola Rock Glacier (Central-Eastern Italian Alps, 46°12′–10°42′, Figure 1) is mostly composed of matrix-free, angular boulders up to some meters in diameter, while fine-grained material locally outcrops on the lateral and frontal slopes and on the top of the ridges (Seppi et al., 2011, 2012).
(a) Geographic setting and (b) general view of the Amola Rock Glacier (Central-Eastern Italian Alps). The elevation ranges from 2280 to 2500 m a.s.l., and the area is about 9.7 ha. The bedrock consists of biotitic–amphibolic tonalite. (c) Sampling sites, ground temperature data logger, and surveyed boulders on the rock glacier and in the surrounding area.
Sampling design and environmental variables
To detect the displacement of the rock glacier, topographic surveys were conducted annually since 2001 with a total station (LEICA TCA 2003), measuring a network of 25 large boulders (Figure 1). The maximum measurement uncertainty can be estimated at ±2.5 cm. The velocity of the frontal area of the rock glacier, where the plant and arthropod sampling points are located, was interpolated in a GIS by an inverse distance weighted (IDW) method, obtaining a raster surface. The velocity of the sample points was then estimated by intersecting them with the raster surface.
Near-surface ground temperature was continuously measured at two sites on the rock glacier and on a nearby scree slope (Figures 1 and 2). Data were collected for 3 years (2009–2012) with hourly resolution using data loggers Tinytag TGP-4020 with an external probe (measurement range: −40° to +125°C; accuracy: ±0.35°C) and HOBO H8 Temp (measurement range: −20°C to + 70°C; accuracy: ±0.7°C). The mean annual ground surface temperature (MAGST) was then calculated for each measuring site, and the thermal regime was analyzed.
Thermal regime of the monitoring sites located on the rock glacier (PAT6 and HOBO16) and on a nearby scree slope (HOBO4). The data logger PAT6 was installed on the top of a ridge in fine-grained material, at about 10 cm below the surface to avoid direct solar radiation. The data logger HOBO16 was placed between large boulders few meters below the surface and measured the air temperature in the voids between the boulders. The data logger HOBO4 was installed in fine-grained material about 5 cm below the surface.
For plant and arthropod sampling, two sampling sites were randomly placed on the rock glacier (sites D and E) and two others, as a control, on an undisturbed surface within an alpine grassland community (site F) and on a scree slope (site Fb) (Figure 1). Each sampling site included three randomly placed pitfall traps, the same utilized in similar habitats (see Gobbi et al., 2011), surrounded by four plots for vegetation sampling; each trap was distanced at least 10 m to avoid any spatial autocorrelation. The traps were emptied about every 20 days during the snow-free season (July–September 2010 and 2011).
Two milieu souterrain superficiel (MSS) traps (Juberthie, 1979) were located at different depths on the rock glacier, one within fine grain size at 60 cm (MSS_1) and the other between metric boulders at 6 m of depth (MSS_2). These traps were used to evaluate the presence of endogean fauna and were located in July 2011 and emptied in late August 2011.
Plots for the sampling of plant cover consisted of a metal circle of 40 cm diameter placed at the four opposite sides of the pitfall trap. Vascular plants, bryophytes, and ground lichens occurring within the plot were recorded; the overall vegetation cover and that of every species were visually estimated with a resolution of 5%. The mean values from the four plots were calculated to obtain a single value associated with each trap. Collected data in such a small area may not be representative of the whole plant cover but could be strictly associated with substrate and arthropod sampling.
A substrate sample of 1–2 kg for particle size distribution analysis was taken at every sampling site. An amount of 200 g of substrate was sampled at each pitfall trap for pH (in 1:2.5 soil:water) and organic matter content (Walkley–Black method) analysis. All the soil samples were taken at the surface.
Statistical analyses
The differences in the analyzed variables between each sampling site were evaluated by the analysis of variance (ANOVA) and post hoc Tukey’s HSD test. The relationships between environmental variables and species were outlined through a direct gradient analysis. According to the gradient length (Podani, 2000), a redundancy analysis (RDA) (length <2.5 cm) was performed for plants and a canonical correspondence analysis (CCA) (length >2.5 cm) for arthropods. A forward selection within the CCA and RDA was performed to select the significant explanatory predictor variables. Statistical significance (p < 0.01) of each environmental factor was assessed independently by means of an unrestricted Monte Carlo permutation test (999 permutations). For multiple simultaneous testing, the Bonferroni correction was considered. The significance of the total model, after deleting the non-significant environmental factors, was also tested by Monte Carlo permutation test (999 permutations).
Plants. Four environmental variables (silt–clay, gravel–sand, soil pH, and soil organic matter) and 63 species were included in the analysis. From the analyses, 27 plant species occurring in only one site were eliminated.
Arthropods. We analyzed spider (Arachnida: Araneae) and ground beetle (Coleoptera: Carabidae) assemblages as they are the most abundant ground-dwelling arthropods with well-known ecology and spatial patterns (Brandmayr et al., 2003; Thaler, 2003) and have been found to clearly react to climate warming in the Italian Alps (Brambilla and Gobbi, 2014; Gobbi et al., 2006). Six environmental variables (vegetation cover, vegetation richness, silt–clay, gravel–sand, soil pH, and soil organic matter) and 20 species were included in the analysis. Six rare (number of specimens <2) species were omitted from the analysis.
The statistical analyses were performed using IBM SPSS Statistics v. 20 and CANOCO v. 4.51 for Windows (Ter Braak and Smilauer, 1998).
Results
Environmental variables
From 2001 to 2011, the rock glacier showed mean velocities between 0.1 and 20.7 cm/yr and is therefore active, in spite of an essentially inactive part on its right sector. The velocity of the measuring points located on frontal sector (n = 14) ranged from 1 to 14 cm/yr, with the displacement direction following the maximum slope. The estimated velocities of the sampling sites ranged from more than 2 cm/yr (point D3) to about 10 cm/yr (points E1 and E2; Table 1).
Data of sampling sites. Total displacement and mean velocities of the sampling sites on the rock glacier estimated in the period 2001–2011; vascular plants (pl), bryophytes (br), lichens (li), carabid beetles (be), and spiders (sp) sampled in each site (values represent the average plant cover plots and the number of arthropod individuals). ID: acronyms for plant and arthropod names as reported in Figure 3. Nomenclature follows Pignatti (1982) for vascular plants, Cortini Pedrotti (2001) for bryophytes, and Nimis and Martellos (2008) for lichens.
Soil organic matter content was very low in both rock glacier sites (mean values: 46.03 ± 7.10 and 30.51 ± 12.38 g/kg for sites D and E, respectively) without any significant difference between them (Tukey’s HSD test p = 0.778). Rock glacier sites, scree control sites, and grassland control sites were significantly different from each other (ANOVA test and post hoc Tukey’s HSD test F(3, 10) = 52.851; p < 0.001), with grassland showing the highest values of organic matter (433.74 ± 15.03 and 211.22 ± 44.86 g/kg for grassland and scree, respectively). Soil pH did not show any significant variation among sites (ANOVA test and post hoc Tukey’s HSD test F(3, 10) = 1.612; p = 0.262).
On the rock glacier, remarkably cold conditions were observed at HOBO16 site (Figure 2), where the MAGST was always negative (−1.4°C, −2.3°C, and −1.8°C in 2009, 2010, and 2011, respectively). Winter temperature was affected by wide, short-term fluctuations, suggesting a continuous air circulation in the pore space and an air exchange with the atmosphere due to the low insulating effect of the snow cover. At PAT6 site, the MAGST above zero was recorded in the same years (1.1°C, 0.7°C, and 1.9°C, respectively). Due to the effective thermal insulation of the snow layer, the temperature at this site constantly decreased during winter, reaching a stable period in late winter/early spring with values ranging from −5°C to −3°C (Figure 2). A different thermal behavior characterized the site located outside the rock glacier (HOBO4). MAGST was always above 3°C, and the temperature during the snow cover period was constantly at 0°C (Figure 2), suggesting the absence of permafrost (Hoelzle et al., 1999). The onset of the summer thermal regime (Figure 2) showed that the snow disappeared earlier on the scree slope than on the rock glacier.
Plants
In total, 63 plant species were recorded; total plant cover ranged from 15% (trap E2) to 86% (trap F2) (Table 1). The number of plant species ranged from 4 (trap E3) to 27 (trap F3). Saxifraga bryoides, Oxyria digyna, Poa laxa, Veronica alpina, Adenostyles leucophylla, Cystopteris fragilis, Hedwigia ciliata, and Homalia besseri were exclusive to the rock glacier surface.
Plant cover was correlated with plant species number (r = 0.928; p < 0.001) and was significantly higher on the grassland (site F) (ANOVA test and post hoc Tukey’s HSD test F(3, 10) = 11.518; p = 0.003); plant species number varied significantly among sites (ANOVA test F(3, 10) = 29.610; p < 0.001) and was significantly higher on site F and lower on site E.
After forward selection (RDA), one environmental variable (grain size as gravel–sand) was extracted that significantly correlated with the distribution of the plant species, explaining 54.1% of the total variance (Figure 3a). The first axis was strongly correlated with grain size (r = −0.85 with gravel–sand) and separated the rock glacier sites (left) from control sites. Grain size was significantly correlated with soil organic matter content (r = −0.97 with gravel–sand); therefore, the first axis defines a gradient, with particle size and organic matter amount increasing from right to left. The second axis possibly indicated a moisture gradient probably linked to microtopography, with species of relatively wet sites (Salix herbacea, Soldanella pusilla) at the negative side of the axis and species of dryer sites (e.g. Sedum alpestre) at the positive side.
Direct gradient analysis indicating the relationship between the environmental variables and (a) plant and (b) arthropod species distribution. (a) Plants: four eigenvalues were extracted – first axis: 0.541; second axis: 0.287; third axis: 0.082; fourth axis: 0.032. Significance of all canonical axes was p = 0.001. (b) Arthropods: four eigenvalues were extracted – first axis: 0.923; second axis: 0.922; third axis: 0.532; fourth axis: 0.421. Significance of all canonical axes was p = 0.001.
Arthropods
A total of 394 specimens (391 in the pitfall traps and 3 in the MSS traps) belonging to 28 species were sampled (Table 1). Of them, 13 species were carabid beetles, while 14 species were spiders. The most abundant species were Oreonebria angustata (N = 148, 2 specimens in the MSS traps) and Nebria germari (N = 116, 1 specimen in MSS_2); the latter was found only on the rock glacier together with Trechus tristiculus (N = 19) and Oreoneta montigena (N = 5).
Arthropod species richness and the number of individuals were not correlated to each other (r = 0.40; p = 0.22). Both species richness and abundance did not significantly change between rock glacier (sites D and E) and controls (sites F and Fb) (ANOVA test for species richness: F(3, 10) = 2.77; p = 0.12; ANOVA test for individuals: F(3, 10) = 1.03; p = 0.43).
After forward selection (CCA), one environmental variable (grain size as gravel–sand) was extracted that significantly correlated with the distribution of spiders and ground beetles, explaining 25.9% of the total variance (Figure 3b). The first axis was correlated strongly with grain size (r = −0.99 with gravel–sand). It separated the rock glacier sites (left), characterized by coarse grain size, from the control sites (right). Grain size was significantly correlated with soil organic matter amount (r = −0.97 with gravel–sand); therefore, the first axis defines a gradient, with particle size and organic matter content increasing from right to left.
CCA showed the main gradient evidencing the presence of specific assemblages from sites with the finest grain size and highest organic matter (site F) to sites with large grain size and low organic matter (sites D and E); Fb site showed intermediate position (Figure 3b). Four carabid beetles (Oreonebria angustata, Nebria germari, Trechus tristiculus, and Notiophilus biguttatus) and two spiders (Oreonetides glacialis and Oreoneta montigena) were linked to the sites on the rock glacier (sites D and E).
Discussion
Our data showed the uniqueness of the rock glacier environment, characterized by surface movement and thermal regimes indicating the occurrence of permafrost according to Hoelzle et al. (1999). Temperature regime differentiates rock glacier from the surrounding habitats but also emphasizes its heterogeneity as a function of grain size, with coarse-sized areas showing the ‘negative thermal anomaly’ that characterizes the blocky deposits (Harris and Pedersen, 1998; Juliussen and Humlum, 2008).
Plant cover was not observed on areas consisting of metric boulders or coarse gravel. The overall pattern of plant cover was that of scattered ‘islands’ where plant and moss cover could locally attain 50% cover. The effects of grain size could probably be due to both its direct effect and its role in determining the distinctive temperature regime characterized by very low values even during the growing season, when harsh winter climate and disturbance must be balanced by relatively favorable conditions to allow plant survival (Caccianiga et al., 2006; Körner, 1999).
Substrate velocity is not a limiting factor for plant survival; the maximum value recorded in our sampling sites (10.3 cm/yr) is lower than the limiting threshold for plant survival of 30 cm/yr reported by Burga et al. (2004) and the velocity recorded on debris-covered glaciers (up to c. 15 m/yr; Caccianiga et al., 2011). Our velocity values are lower than 1.5 m/yr, threshold below which grain size becomes the main factor affecting plant cover according to Rieg et al. (2012).
The overall plant species composition of the rock glacier surface could be considered similar to that of scree slopes and moraines (Caccianiga and Andreis, 2004), but lacking the local occurrence of demanding species that are able to colonize favorable microsites on gravitative deposits. Another feature separating rock glacier communities from those of gravitative scree is the higher frequency of some bryophytes, particularly Polytrichastrum alpinum, an indicator of long-lasting snow cover (Odland and Munkejord, 2008); moss abundance has already been indicated by Burga et al. (2004) as a likely indicator of permafrost occurrence.
Concerning the arthropod assemblages, our results indicate the presence of exclusive species to each landform (rock glacier, scree, and grassland). Specifically, the rock glacier hosts large populations of three carabids: Nebria germari, Oreonebria angustata, and Trechus tristiculus. These species are typical of cold and wet high-altitude environments (Ledoux and Roux, 2005): Nebria germari and Oreonebria angustata can be found near glacier fronts, on glacier surface, and on gravitative scree slopes (Brandmayr and Zetto Brandmayr, 1988; Gereben-Krenn et al., 2011). Trechus tristiculus is a cold-adapted species with endogean lifestyle, indicator of deep screes with macroporal system in rocky material (Lompe, 2004). Although these species are known in the literature to be able to live on high-altitude scree slopes, both Nebria germari and Trechus tristiculus could not be found on the studied scree slope. Only a small population of Oreonebria angustata occurred in the studied scree slope compared with that on the rock glacier (5 vs 143 specimens). These species suggest that the rock glacier offers more suitable microhabitat conditions than scree slope because of the distinctive grain size and the presence of subsurface ice. Moreover, the carabids belonging to genus Trechus are able to colonize only habitats with specific microthermal conditions like caves or deep screes (Lompe, 2004). The detritus depth of the rock glacier has been confirmed by the catch of Nebria germari and Oreonebria angustata with the MSS traps, thus suggesting the presence of an architecturally complex macroporal structure where ice ensures the occurrence of continuously low temperature throughout the year.
Rock glacier as potential warm-stage refugia
Concerning the arthropod assemblages, it is possible to define a rock glacier as a superficial subterranean habitat represented by fissure network among boulders, human-sized caves included. Unlike other superficial subterranean habitats like scree slopes, where temperatures could reach relatively high values in summer (this paper; see also Růžička et al., 2013), rock glaciers are selected by cold-adapted species, which avoid scree slopes as they do not offer constantly low temperatures during summer. For example, the recent extinction of Nebria germari and Trechus sp. was demonstrated by Pizzolotto et al. (2014) in some scree slopes of the Dolomites where these species were dominant 30 years before, indicating that on this landform no refugia were available for these cold-demanding species.
It is possible to compare the proposed hypothesis of rock glaciers as refugia during interglacial period with a role similar to that of caves during quaternary climatic fluctuations. Such a role is less apparent for plant species, as the distinctive temperature regime of rock glacier surface could act as a limiting factor for plant survival; however, in a warming scenario, such regime could provide survival opportunities for cold-requiring species, as suggested for debris-covered glaciers (Caccianiga et al., 2011).
Footnotes
Acknowledgements
We thank Jurij Bonomo, Silvia Bussolati, and Gianalberto Losapio for assistance during sampling and the Adamello-Brenta Natural Park for permission for carrying out this research.
Funding
The research reported here is partially funded by the Autonomous Province of Trento (Italy). The field work of R. Seppi was funded by the MIUR Project (PRIN 2010-11): ‘Response of morphoclimatic system dynamics to global changes and related geomorphological hazards’ (coordinators G. Dalla Fontana and C. Baroni). The topographic surveys were supported by the Geological Service of the Autonomous Province of Trento (M. Degasperi, F. Rippa, and M. Zumiani).