Keywords

1 Introduction

A major key to the understanding of natural hazards and risks is the documentation of past geomorphic process activity (Stoffel and Huggel 2012). As a result of the absence of documentary records, this information must often be developed from natural archives or “silent witnesses” (Aulitzky 1992) that remain visible in the landscape after an event. The significant contribution of tree rings to these endeavors lies in their capacity to preserve evidence of past geomorphic activity, as well as the interactions and linkages of hydrological processes with landforms or the interaction of geomorphic processes with water. In many climates, tree-ring records may thus represent one of the most valuable and precise natural archives for the reconstruction and understanding of past and ongoing processes during the past several hundred years (Stoffel and Bollschweiler 2008; Stoffel et al. 2010; Stoffel and Corona 2014).

The initial employment of tree rings in geomorphic studies was simply as a dating tool (Douglass 1941; Stokes and Smiley 1968)—but only rarely exploited environmental information that could be derived from studies of ring-width variations and records of anatomical anomalies contained within the tree. However, these unique, annually resolved, tree-ring records preserve potentially valuable archives of past geomorphic processes. As many geomorphic processes are also significant natural hazards, understanding their distribution, timing and controls provides valuable information that can assist in the development of mitigation and defense against these hazards and their effects on society (Osterkamp et al. 2012; Stoffel and Wilford 2012).

Apart from the site-specific information common to many trees at any site, individual trees also record the effects of mechanical disturbance caused by external processes. This contribution aims at presenting (1) how dendroecological techniques can help in unraveling and dating past geomorphic disturbance, (2) which growth anomalies can be identified and attributed to past geomorphic process activity, and at providing (3) a suite of examples with selected recent applications and issues in the field of dendrogeomorphology.

2 Dendroecological Consequences of Geomorphic Disturbances

In his seminal work, Alestalo (1971) illustrated that the occurrence of earth-surface processes will typically injure trees, tear off their crown or branches, tilt their stems, partially bury them or expose their roots. Evidence of these events is recorded in growth-ring records of affected trees (Shroder 1978) and can be used for the dating of past geomorphic process activity, provided that tree-ring series are cross-dated properly. Based on the principles presented in these seminal papers (Butler and Stoffel 2013), a set of characteristic growth disturbances has been typically used in dendrogeomorphic studies (Stoffel et al. 2013a), with a clear focus on injuries, reaction wood and growth suppression (Stoffel and Corona 2014). In the following, we provide an overview of how geomorphic process activity usually affects trees and how trees will react to these disturbances.

2.1 Injuries and Callus Tissue

Partial bark removal and wood-penetrating injuries are a common feature in trees affected by geomorphic processes (Trappmann and Stoffel 2013, 2015). Wounds can occur on the tree’s stem, its branches or on roots. If impacts locally destroy the cambium, incremental cell formation will become disrupted and new cell formation will cease in the injured segment of the tree. To minimize rot and the negative effects of insect attacks after damage, the injured tree will compartmentalize the wound (Shigo 1984; Stoffel and Klinkmüller 2013) and start the production of chaotic callus tissue at the edges of the injury so as to continuously close the wound. The extent of wound healing will, however, greatly depend on the annual increment rate, tree age and on scar size (Bollschweiler et al. 2008; Schneuwly et al. 2009a). The presence of injuries and chaotic callus tissue is commonly regarded as a valuable and reliable indicator of past geomorphic process activity. Whereas scars will remain largely visible on the stem surface of tree species with smooth bark (e.g. Abies, Alnus, Betula, Fagus; Trappmann and Stoffel 2013), they may become fully obscured in species with thicker bark structures such as Larix, Picea, Pinus or Quercus (Stoffel and Perret 2006; Trappmann et al. 2013).

2.2 Tangential Rows of Traumatic Resin Ducts

Following cambium disturbance, tangential rows of traumatic resin ducts (TRDs) are produced in the developing secondary xylem of certain conifer species such as Larix, Picea, Pseudotsuga or Abies (Bannan 1936; Jacoby 1997; Stoffel 2008), where they extend both tangentially and axially from the injury (Bollschweiler et al. 2008; Schneuwly et al. 2009a, b). When wounding occurs during the vegetation period of the tree, resin production will start within a few days after the impact and ducts will emerge within 3 weeks after the disturbance). Therefore, when analyzing cross-sections, the intra-seasonal position of the first series of TRDs can be used to reconstruct previous events with monthly precision (Stoffel et al. 2005b, 2008; Stoffel and Beniston 2006; Schneuwly-Bollschweiler and Stoffel 2012), provided that disturbance occurred during the vegetation period. With increasing axial and tangential distance from the impact, however, TRDs tend to migrate to later portions of the tree ring (Bollschweiler et al. 2008; Schneuwly et al. 2009a). The intra-seasonal dating with monthly precision thus has to be based on cross-sections or a large number of increment cores at the same elevation on the stem. This technique cannot be used in Pinus because TRDs do not normally occur in this genus, which produces copious amounts of resin and resin ducts unrelated to mechanical wounding (Blodgett et al. 2007; Ballesteros et al. 2010a).

2.3 Tracheid and Vessel Anomalies

Anomalies in tracheids and vessels have only rarely been used in the past to extract signals of geomorphic activity. Most work has been realized on riparian trees affected by floods and/or debris flows (e.g. St. George et al. 2002; Ballesteros et al. 2010b; Arbellay et al. 2010a, b; Wertz et al. 2013). Other processes, such as snow avalanches, have not been studied frequently in the past (Arbellay et al. 2013). Tree microscopic response to wounding was primarily studied between rings formed in the year of disturbance and subsequent years as well as in uninjured control rings. Injured rings were characterized by much smaller (but more) vessels as compared with uninjured rings, yet fiber and parenchyma cells did not differ significantly in numbers and size between injured and uninjured rings. These results highlight the existence of anatomical tree-ring signatures related to past geomorphic process activity and address an innovative methodological approach to date injuries inflicted on broadleaved trees with minimally destructive techniques. Arbellay et al. (2012a, b) have also expanded their approach to analyze thickness-to-span ratios of vessels, xylem relative conductivity and xylem vulnerability to cavitation, and state that the degree to which the wound-induced anatomical changes in wood structure express the functional need of trees to improve xylem hydraulic safety and mechanical strength at the expense of water transport. They conclude that xylem hydraulic efficiency was restored in 1 year, while xylem mechanical reinforcement and resistance to cavitation and decay lasted over several years.

Research on tracheid changes in conifers has been basically limited to root exposure, with the exception of Stoffel and Hitz (2008) who identified changes in lumina of earlywood tracheids after wounding by snow avalanches and rockfalls or Arbellay et al. (2014a, b) who analyzed ecophysiological traits next to scars in North American conifer trees affected by fire. Exposed roots will continue to grow and fulfill their functions as long as their outer tips remain in the ground. In the exposed portion of the root, anatomical changes will occur (Stoffel et al. 2013b; Ballesteros-Cánovas et al. 2017a) and individual growth rings similar to those in the stem or branches will be formed. The localization of such changes in the tree-ring record allows determination of the moment of exposure (Corona et al. 2010, 2011; Lopez Saez et al. 2011, Stoffel et al. 2012).

2.4 Reaction Wood

Inclination of the stem may result from the sudden pressure induced by hydrogeomorphic processes directly, by the associated deposition of material (e.g. avalanche snow, debris-flow material), or by the slow but ongoing destabilization of a tree through landslide activity or erosion. Tilted trees are common in most areas affected by geomorphic processes and have therefore been used in many publications focusing on the dating of event histories (Braam et al. 1987; Fantucci and Sorriso-Valvo 1999; Lopez Saez et al. 2012a, b, 2013; Sorg et al. 2015).

Subsequent growth in the trunk of a tilted tree will attempt to restore its vertical position, and the reaction will be most clearly visible in that segment of the tree to which the center of gravity has been moved through the inclination of the stem axis (Mattheck 1993). In the tree-ring record, eccentric growth will be visible after a tilting event and thus will allow accurate dating of the disturbance. In conifers, compression wood (also referred to as reaction wood) will be produced on the underside of the trunk. Individual rings will be considerably larger and slightly darker in appearance as compared to the upslope side. The difference in color results from much thicker and rounded cell walls of earlywood and latewood tracheids (Du and Yamamoto 2007). Compression wood also tends to have a higher proportion of latewood, higher lignin content and higher density (Timell 1986). Multiple tilting events in the same stem may be recognized by changes in the amount, color or orientation of reaction wood series in the tree-ring record. By contrast, stem tilting in broadleaved trees leads to the formation of tension wood (Westing 1965) on the upper side facing the tilting agent. Broadleaved trees react upon tilting with ultra-structural modifications (e.g. fewer vessels of smaller diameter, higher cellulose content and a gelatinous layer oriented nearly parallel to the fiber axis) that are typically only visible when studied on micro-sections (Pilate et al. 2004).

2.5 Growth Reduction

Debris flows, floods, or landslides may bury trees by depositing material around their stem base. Growth suppression after burial with debris is caused, on the one hand, by a reduced activity of the roots, and on the other hand, by mechanical effects caused by the enormous weight of debris. The pressure on the cambium exerted by bark and phloem impedes the cell division and leads to a reduced number of cells with narrower lumen (Kny 1877). The supply of water and nutrients will be temporarily disrupted or at least limited (LaMarche 1966; Hupp et al. 1987; Friedman et al. 2005), and the yearly increment will be diminished (Kogelnig-Mayer et al. 2013). Reductions in annual ring widths in tilted trees are thought to be related to the partial destruction of root mass in the case of unstable slopes (Mayer et al. 2010). If stem burial exceeds a certain threshold, trees will die from a shortage of water and nutrient supply.

Bouncing rocks and boulders, debris flows and lahars or the windblast of snow avalanches may cause decapitation of trees or the removal of branches. The loss of the crown or branches is more common in bigger trees, when stems have lost their flexibility. Apex loss has also been observed as a result of rockfall impacts close to the ground level. In such cases, the sinusoidal propagation of shockwaves in the stem results in the break-off of the crown (whiplash; Lundström et al. 2009). Trees react upon decapitation or branch loss with distinct radial growth suppression following the impact. One or several lateral branches will form a “leader” that replaces the broken crown, resulting in the tree morphology called “candelabra” growth (Butler and Malanson 1985; Stoffel et al. 2005a). Leaders may also be formed from prostrated trunks knocked over by geomorphic events.

2.6 Process Dating with Dendrogeomorphic Evidence

The expression of disturbance in the tree-ring record may vary in intensity as well as in the spatial and temporal extent between processes, species and age/size classes of trees. As a consequence, and based on the process and tree species analyzed, different classification systems have been used in the past. In Table 12.1, we present a synthesis of indicators commonly used in dendrogeomorphic research and make a proposal on how to analyze and interpret growth anomalies (Stoffel and Corona 2014). Distinction is made between weak, moderate and strong growth anomalies, and thresholds are typically used to distinguish geomorphic signals from noise (e.g., anthropogenic disturbances, insect attacks). An overview of thresholds and guidelines on how to reconstruct time series of geomorphic events can be found in previous publications (Corona et al. 2012, 2014; Schneuwly-Bollschweiler et al. 2013; Stoffel et al. 2013a; Stoffel and Corona 2014; Trappmann et al. 2014; Chiroiu et al. 2015; Morel et al. 2015); we do not provide further details in this paper but rather present three case studies exemplifying recent progress and remaining questions in the research field.

Table 1 Intensities of growth anomalies (GDs) based on their appearance and/or persistence in the tree-ring series
Full size table

3 Tree Reactions to Floods and Erosion: Two Case Studies

In the following, we present two cases where the application of dendroecological disturbances in trees can be used to infer past mass movement activity. These examples represent the state-of-the-art in the respective research fields, but also point to remaining challenges—thus calling for new research.

3.1 River Floods and Disturbance in Trees

The interaction between trees and fluvial processes allow the dating of past flood activity, but also contribute to an improved understanding of ecological feedbacks in floodplains. The exploratory analysis of tree-ring responses to hydrology initially started in the first half of the twentieth century (Hardman and Reil 1936). Early work primarily focused on streamflow reconstructions by using the relation between tree-ring growth and season and/or annual flow discharge record. This approach has since become a research direction of its own with applications for water resources management (Woodhouse and Lukas 2006). The application of tree-ring records in flood reconstructions was initiated by Robert S. Sigafoos, who described, in his seminal publications in the early 1960s (Sigafoos 1961, 1964), linkages between riparian vegetation and flood frequency in the Potomac River (Washington, USA). He concluded that botanical evidence of past floods events has a clear economic value as it allows improved flood frequency analysis, and consequently risk estimation. Sigafoos (1964) also contributed to the understanding of the dissimilar adaptability of tree species to flood variability, as well as to its influence on seed establishment in floodplains, which has since been considered as a benchmark in subsequent research on riparian ecology (Hupp and Osterkamp 1996; Naiman et al. 2010; Stoffel and Wilford 2012). Tree-ring records have since been used as a valuable source of information in flood hazard assessments, especially in poorly or ungauged mountain catchments (see Ballesteros-Cánovas et al. 2015a, b, c, d, e for a recent review).

Floods can leave a large set of anomalous growth patterns in trees. In floodplains, the distribution, nature and age of riparian species can yield comprehensive information on flow dynamics and competence (Hupp and Osterkamp 1985, 1996), including analyses of changes in fluvial geomorphology (Marston et al. 1995; Garófano-Gómez et al. 2012). At the level of individual trees, disturbances can very frequently be observed in steep, high-gradient streams. The most commonly observed set of botanical evidence in these environments includes abrasion scars caused by sediment, ice or wood transported during floods (Zielonka et al. 2008; Ruiz-Villanueva et al. 2010; Ballesteros-Cánovas et al. 2011a, b; Lagadac et al. 2015), sprouts (new leaders) growing from downed stems (Sigafoos 1964; Gottesfeld and Gottesfeld 1990), trunks tilted by unidirectional flow pressure (Ballesteros-Cánovas et al. 2015d) roots exposed by the erosion of the root-plate system (Malik 2006; Casteller et al. 2015). In low gradient streams, trees growing in the endorheic zone can document past floods as well, even in the absence of other physical evidence, as the transport and production of growth hormones in root systems will suffer from the prolonged influence of water, thus leading to growth abnormalities (St. George and Nielsen 2003; St. George et al. 2002; Wertz et al. 2013; Copini et al. 2016).

The identification and use of the growth abnormalities mentioned above allows reconstructions of multi-centennial flood occurrence records with (sub-)annual resolution, and thereby contributes to an improved understanding of flood variability and flood-climate linkages. Flood reconstructions in low-gradient streams are often based on the detection of anatomical structures linked to anoxic conditions during persistent flood (Yanosky 1983, 1984; Astrade and Bégin 1997; St. George and Nielsen 2003; Ballesteros et al. 2010a, b; Therrell and Bialecki 2015).

These reconstructions have enabled the creation of flood chronologies extending back for several centuries. Recent work in low-gradient river systems, for instance, included the assessment of spring floods in the lower Mississippi River going back to the late eighteenth century (Therrell and Bialecki 2015). In mountain streams, Ballesteros-Cánovas et al. (2015b) used forest management to minimize noise and to maximize flash-flood signals in a set of small catchments in Central Spain. Stoffel et al. (2012) and Casteller et al. (2015), in turn, combined growth anomalies from stems and roots to distinguish erosional from sedimentation processes during flash floods in the Patagonian Andes. Comparable approaches have also been used for regional flood reconstructions in the Polish Tatras (Ballesteros-Cánovas et al. 2015c, 2016), the Flysch Carpathians (Šilhán 2015; Šilhán et al. 2016), Central Spain (Rodriguez-Morata et al. 2016), and in the Indian Himalayas (Ballesteros-Cánovas et al. 2017b) (Fig. 12.1).

Fig. 1
figure 1

Examples of dendroecological responses of conifer trees to geomorphic disturbance (modified from Stoffel et al. 2005a). Numbers 1 and 2 represent tree-ring series from opposite sides of the affected stem

Full size image

In combination with hydraulic models, the height of scars in trees—also referred to as paleostage indicators or PSI—are often used to reconstruct flood discharge and hence flood magnitudes (Fig. 12.2), as they indicate the minimum water level during any past event (Ballesteros-Cánovas et al. 2011a, b, 2015a). Several studies have highlighted good agreement between fresh high water marks (HWM) and scar height; discrepancies have been attributed mostly due to uncertainties in the timing of scar generation within the hydrograph (Smith and Reynolds 1983; Gottesfeld 1996; Yanosky and Jarrett 2002; Ballesteros-Cánovas et al. 2011b). The range of observed differences typically increases in high-gradient streams. Similar ranges were observed by Ballesteros-Cánovas et al. (2011a) who used two-dimensional hydraulic models and took into consideration two different scar populations (i.e. large and small scars) based on classic sediment transport theory assumptions. This approach has been recently used for intra-catchment comparisons in four mountain streams of the Tatra Mountains (Poland, Ballesteros-Cánovas et al. 2016b). Besides scar-based reconstruction, a tree-deformation energy approach has been suggested based on the expected relationship between tree tilting and flood magnitude (Ballesteros-Cánovas et al. 2015c).

Fig. 2
figure 2

Scar-based flow discharge reconstruction based on 2D hydraulic models for an intense flood event that occurred in Central Spain in 1997 (modified from Ballesteros-Cánovas et al. 2011a). Note the agreement between paleostage indicators (scar height) and peak discharge and stream power

Full size image

3.2 Denudation Processes and Erosion Signals in Roots

Erosion is a major threat to soil resources and may impair their ability to deliver a range of ecosystem goods and services in various environments worldwide (Verheijen et al. 2009). It not only leads to a loss of soil fertility, but also causes off-site effects in the form of downstream sedimentation (de Vente and Poesen 2005), reduced hydraulic capacity of rivers and drainage ditches, increased flood risks (Sinnakaudan et al. 2003), the blocking of irrigation channels, as well as a reduction of design life of reservoirs (Romero-Díaz et al. 2012). Soil erosion also leads to the transport of chemicals and thereby alters biogeochemical cycling, which in turn may cause eutrophication of water bodies (Quinton et al. 2010). Rates of soil loss have been measured, modelled or inferred for most types of soil erosion in a variety of landscapes. Yet, traditional field-based methods of monitoring erosion (sediment traps, erosion pins, and bridges) are labor intensive and therefore are generally limited in both time (Cantón et al. 2011) and/or spatial extent (Gillan et al. 2016), thus limiting our knowledge on how erosion processes evolve over time and on what their impacts are on changes in environmental conditions (Poesen et al. 2003; Nadal-Romero et al. 2011). The development of methodologies capable of quantifying erosion rates retrospectively and at different spatio-temporal scales therefore remains a major scientific challenge (Poesen et al. 2003). In this regard, bio-indicators—such as exposed roots—represent an alternative method to determine erosion rates at short to medium term timescales (Stoffel et al. 2013a, b), especially in vegetated, and ungauged regions.

The potential of roots as an indicator of degradation was recognized since the mid-twentieth century. Schulman (1945) was the first to successfully synchronize ring widths from large roots of Pseudotsuga menziesii (Mirb.), thereby obtaining a series of soil moisture changes and a detailed record of year-to-year fluctuations in runoff for the upper Colorado River. In the White Mountains of California, LaMarche (1963) realized that the vertical buttress form observed in exposed roots was the result of (1) bark and cambium stripping after abrasion and weathering from the upper root surface and (2) continued secondary growth on the lower root surface. Based on these observations of shape changes and discontinuity of growth rings, he produced the first estimate of root ages at the time of initial cambium reduction. In the following years, LaMarche (1968) and Carrara and Carroll (1979) used the ratio between the minimum depth of erosion—obtained from the reconstructed root diameter at the moment of denudation—and the time (i.e., number of growth rings) passed since root exposure, to establish total vertical erosional losses from hillslopes over lifetimes of individual, long-lived trees. Since these seminal papers, the dendrogeomorphic method has been widely accepted as a powerful tool in reconstructing rates of soil erosion from analysis of exposed root (e.g., Krause and Eckstein 1993; McAuliffe et al. 2006; Chartier et al. 2009). A similar approach—combined with the sprouting of adventitious roots, scars or growth-ring shape changes from concentric to eccentric following exposure—was also used to quantify gully (e.g., Pérez-Rodríguez et al. 2007), sheet (Bodoque et al. 2005), meander bank (Malik 2006), or shore (Fantucci 2007) erosion. All these studies focused on the variation in tree-ring structure (width, concentricity, scars) at a macroscopic level.

More recently, microscopic approaches and equations have been introduced, determining the year of exposure based on changes in (1) the anatomical structure of tracheids in conifer roots (Gärtner 2007; Rubiales et al. 2008; Corona et al. 2011; Lopez Saez et al. 2011) and in (2) the formation of vessels in roots of broadleaved trees (Hitz et al. 2008a, b). Two criteria are usually retained to assess the first year of exposure, namely the (1) abrupt reduction in tracheid size (Corona et al. 2011); and the (2) abrupt increase in the cell wall thickness (Stoffel and Corona 2014). These changes are interpreted as a response of root cells in order to limit their vulnerability to freezing and water stress-induced embolism caused by higher temperature and moisture variations following exposure (Pitterman and Sperry 2003; Mayr et al. 2007). In the case of sudden erosion processes, these typical signatures coincide with the onset of exposure (Gärtner 2007). Erosion rates are calculated using the rings formed since exposure (NRex) and the thickness of the eroded soil layer (Er), whereby Er is obtained via the height of the exposed part of the root measured with a depth gauge (Ex). These factors, along with root growth (Gr1, Gr2) since exposure and bark thickness on the upper and lower sides of the root (B1, B2) can be related as follows:

$$ Er=\left( Ex-\left( Gr1+ Gr2\right)+\frac{\left(B1+B2\right)}{2}\right)/ NRex $$

In the case of continuous denudation, anatomical changes in roots start to emerge as soon as denudation had reduced soil cover to less than a few centimeters (Corona et al. 2011; Lopez Saez et al. 2011). As a consequence, a bias (ε) related to root response under thin soil layers should be estimated systematically. In addition, high-resolution microtopographic information around roots—derived from e.g. Terrestrial Laser Scanning—are essential in order to reduce uncertainties related to the quantification of eroded soil layers (Ballesteros-Cánovas et al. 2013a, b; Stoffel et al. 2013a, b), which can be as high as 50% of the eroded soil layer in silica badlands (Bodoque et al. 2015). Finally, Corona et al. (2011) points out that reconstructed soil thickness values also depend on the stability of the root axis. If the root axis remains stable over time, a relative uplift of the root center will occur because root increment in its lower part will be balanced by positive adjustment of root curvature (Coutts 1989; Polacek et al. 2006). In this case, an overestimation of Ex will result from the subsequent growth of the upper part of the root and Er has to be calculated as follows:

$$ Er=\left( Ex-(Gr1)+\frac{\left(B1+B2\right)}{2}+\varepsilon \right)/ NRex $$

Based on these anatomical considerations, the reconstruction of erosion rates focused on (1) precise quantification of continuous erosion processes in badlands and on (2) the validation of root-ring measurements by comparison with other records. Corona et al. (2011) thus used exposed roots from Pinus sylvestris to quantify sheet erosion processes on interfluve and gully slopes in marly badlands of the Southern French Alps and obtained bias-corrected (ε) erosion rates of 5.9 ± 2.6 mm year−1. Comparison of reconstructed erosion rates with a series of systematic measurements performed across a network of marking stakes not only shows almost identical rates (5.7 ± 2.3 mm year−1), but also points to the fact that values would have been underestimated on the interfluves and gully slopes without the bias-adjustment. Similarly, in experimental sandy badlands located in Central Spain, Bodoque et al. (2011) and Ballesteros-Cánovas et al. (2015a, b, c, d, e) demonstrated that root-ring records (6.6–8.8 mm year−1) differed by 36% from those obtained through direct observation (60 erosion pins and 12 pedestals; 11.9 mm year−1 for a period of observation of three hydrological years (Lucía et al. 2011), thereby pointing to the strong influence of a high-intensity rainfall event on monitoring data and the smoothing of extremes in medium-term rates of erosion.

In gully and torrential environments, abrupt changes in cell lumina following sudden root exposure have mainly been used to quantify channel incision, headcut retreat and sidewall processes (see e.g. Malik 2008; Lucía et al. 2011; Silhan et al. 2016). In Central Spain, dendrogeomorphic analyses from 120 exposed roots thus enabled the 3D reconstruction of channel incision, widening and retreatment during the last 23 years and with annual precision of a gully developed in silica sand (Ballesteros Canovas et al. n.d.). Annual average gully incision rates were estimated of 0.09 ± 0.1 m year−1 while lateral widening as well as soil erosion rate due to concentrated runoff reached 0.2 m year−1 and 6 cm year−1, respectively. In a same manner, roots from torrent and river systems have been for instance used in Poland, Czech Republic, Switzerland to date the occurrence of bank erosion processes (e.g., Hitz et al. 2008a, b; Malik 2008; Stoffel et al. 2012; Silhan et al. 2016). Based on the evidence conserved in the root-ring record of various species, the authors were also successful in linking the information contained in the roots with the occurrence of extraordinary floods in the wider study region.

With respect to shore erosion, despite the demonstration of the obvious potential of the approach by Bégin et al. (1991) along the St. Lawrence stream (Canada), very limited attention has been paid to the dendrogeomorphic reconstruction of processes in the past. Fantucci (2007) established a map of coastline erosion for the Bolsena Caldera Lake in central Italy based on roots of broadleaved trees and demonstrated that erosion rates were obviously related to the intensity and frequency of winds. In an attempt to quantify shore erosion in the Mediterranean Sea, Rovéra et al. (2013) analyzed anatomical evidence of exposure in Pinus halepensis Mill., roots hanging from decametric, detritic cliffs of Porquerolles Island (Var, France). Reconstructed rates of erosion range from 13 to 34.5 mm year−1 (Fig. 12.3) and are in the same order of magnitude as erosion rates derived from photogrammetric analysis on sandstone cliffs in the Algarve region (Marques 1997), or the Larache region (Morocco, Marques 2003).

Fig. 3
figure 3

Reconstructed rates of cliff erosion derived from exposed Pinus halepensis roots hanging from decametric, detritic cliffs of Porquerolles Island (Var, France). The analysis of oblique roots also allows analysis of temporal changes in cliff profiles, and thus could become a valuable tool for the assessment of the temporal frequency or changes in storminess (Rovéra et al. 2013)

Full size image

4 Recent Progress in Process Dating: Tree Age, Tree Sensitivity and Geomorphic Disturbances

Dendroecological reconstructions of geomorphic processes are based on the identification and dating of growth anomalies to external disturbances. The variability of different types of growth responses is closely related to the nature and extent of the external disturbance (Stoffel and Bollschweiler 2008), with responses occurring with different intensities (Stoffel and Corona 2014).

Several authors have suggested that the nature and intensity of responses will change with tree age (Bollschweiler and Stoffel 2010), as it was described for climate fluctuations (Carrer and Urbinati 2004; Esper et al. 2008; Wu et al. 2013). Quantitative data about possible age-dependent changes in tree sensitivity to geomorphic disturbances is not, however, readily available. The main assumed drivers influencing tree sensitivity are changes in bark thickness, loss of stem elasticity, or changes in stem diameter, with the latter increasing target size. The assumed changes in sensitivity would thus have a direct impact on the ability of a tree to record geomorphic process activity in their ring sequences. As a consequence, the age-dependent tree sensitivity to geomorphic disturbances could significantly affect tree-ring-based chronologies of geomorphic processes.

Some of the most frequently studied geomorphic processes in Alpine environments are debris flows. The dynamic movement of solid blocks, fine materials, water and air significantly influences tree growth and can kill individual trees. Using trees affected by this process, Šilhán et al. (2015) focused on differences in the type and nature of growth disturbances recorded by “young” and “old” trees on the southern slopes of the Crimean Mountains. In total, 1122 tree-ring series from 566 black pines (Pinus nigra ssp. pallasiana) were sampled in 8 debris flow catchments: 361 “young” trees with a post-1930 innermost ring and 205 “old” trees with pre-1930 germination. The year 1930 was selected due to the strong earthquake in 1927 (magnitude M = 6.8), when massive rockfalls reduced the number of trees in the study area, leading to the colonization of the newly created bare surfaces. In total, 1271 growth anomalies could be attributed to debris flow. More than 65% of all growth disturbances were identified between 1939 and 2010, when primarily “young” trees helped the construction of the debris-flow chronology. The dominant growth response of “young” trees was in the form of compression wood (71%). By contrast, the same anomaly represented only 13% of all growth disturbances in the “old” trees during the same time period. “Young” trees did not react by abrupt growth releases to events, whereas this growth response represented 27% of all growth responses in the “old” trees. Considerable differences were also found for growth suppression with 13% in “young” and 53% in “old” trees. Scars and callus tissue represented 16% of all responses in “young” and 6% in “old” trees.

The significant differences in the formation of compression wood can be related to the more flexible stems of “young” trees. Moreover, “old” trees with larger root plates are better anchored on the slopes and can better resist tilting due to the pressure of debris flow material. The next significant difference in the proportion of scars and callus tissue can be attributed to the divergence of the bark thickness of P. nigra, because “young” trees with thinner bark can be easily injured. In the results, less flexible, well-anchored stems with the thick bark of “old” trees react preferentially with abrupt growth changes.

Another common process in mountain environments is rockfall (defined here as a free fall, bouncing or rolling of individual or a few rocks and boulders). Šilhán et al. (2013) analyzed 140 heavily damaged individuals of P. nigra on the Taraktash talus slope, Crimean Mountains, which allowed identification of 977 growth disturbances and the reconstruction of 702 rockfall events. Here, abrupt changes in growth dominated with 69% (abrupt growth suppression, 51%; abrupt growth release, 18%), whereas compression wood (16%) and callus tissue (15%) were much scarcer. Changes in tree sensitivity to rockfall were studied by calculating the number of rockfall events for each decade a tree had at the time of impact (Fig. 12.4a). The first decade of a tree’s life was not taken into account as trees are, in general, very sensitive to non-geomorphic factors (e.g., snow creeping) during their juvenile growth. Results reveal a quite unambiguous pattern of changes in the number of recorded rockfall events per individual tree and for individual decades of its life. In general, all trees revealed an increasing number of recorded rockfalls up to an age of 90 years (an average of 0.43 rockfall events per tree and decade for this time period), with the ninth decade being the one with the largest number of events. Thereafter, the relative number of recorded rockfalls decreases continuously. Although the thinner bark in the youngest trees can be expected to be more sensitive to geomorphic impacts, the sensitivity is not the highest in the early decades of tree life. As rockfall is a discrete process affecting trees with individual falling clasts, tree diameter seems to be the key factor influencing rockfall frequency as increasingly larger trees also represent an increasingly larger target for falling rocks. The stem diameter probably dominates the ability of trees to record impacts up to its ninth decade. As annual increment rates generally decrease with increasing age, the increase in target size will be reduced as well, whereas bark of increasing thickness will gradually provide more protection and thus reduce tree sensitivity to rockfall, at least in the case of P. nigra.

Fig. 4
figure 4

Changes in the number of recorded events of geomorphic processes with tree age: (a) number of rockfall events recorded in individual decades of P. nigra life; (b) number of landslide events recorded in individual decades of P. nigra and F. sylvatica life

Full size image

Landslides represent yet another geomorphic process with a long history of study. Šilhán and Stoffel (2015) evaluated the sensitivity of P. nigra and Fagus sylvatica (L.) to landslides by analyzing one case study each in the Outer Western Carpathians and Crimean Mountains. In total, 39 P. nigra and 119 F. sylvatica were analyzed, with a focus on eccentric growth and compression wood (P. nigra) and eccentric growth (F. sylvatica). Sensitivity was assessed by calculating the number of recorded landslide events per tree and decade of its life (Fig. 12.4b). In both species, the number of landslide signals increased in F. sylvatica up to the sixth decade, when the number of recorded events per tree was highest. Moreover, the intensity of landslide signals (the abruptness of eccentricity changes) also culminated around six decades. The analysis of compression wood, by contrast, yielded somewhat different results. Here, we observed a second peak in the number of recorded landslide events after twelve to thirteen decades. The decrease of tree sensitivity can be attributed to changes in stem elasticity and the decrease in annual increment. Resources are likely allocated to the growing trunk and branches, which in turn limits eccentric growth. This assumed effect would not evidently influences the ability of trees to form compression wood as a reaction to stem tilting. It is also assumed that young P. nigra trees influenced by landslide events would create compression wood to regain the original, vertical growth position, whereas older trees would produce compression wood rather for stabilization than for the straightening of what has become a much larger stem.

5 Concluding Remarks

The original employment of tree rings in natural hazard studies was simply as a dating tool and rarely exploited other environmental and/or ecological information that could be derived from studies of ring-width variations and records of damage contained within the tree. However, these unique, annually resolved, tree-ring records preserve potentially valuable archives of past geomorphic events on timescales of decades to centuries. As many of these processes are significant natural hazards, understanding their distribution, timing and controls provides valuable information that can assist in the prediction, mitigation and defence against these hazards and their effects on society. This chapter has provided some recent examples of dendroecological dating geomorphic process activity, but also pointed to some limitations and knowledge gaps in the field. The age-dependent sensitivity of trees to geomorphic disturbances yet needs more attention, and protocols yet need to be defined to increase confidence in time-series of past events.

As shown with the example on flood reconstructions, the great advantage of tree-based approaches is that they yield reliable evidence of past flood frequency and magnitude that took place prior to the instrumental period. Consequently, this information can be incorporated into flood hazard and risk assessments. By analyzing trees affected by past floods, one can quantify fluvial processes and anticipate potential geomorphic changes in the river channel. This understanding can be incorporated to hazard zonation (Brooks and St. George 2015), and it allows identification of non-stationary catchment behavior. In parallel, the inclusion of reconstructed data in flood-frequency analyses has other considerable advantages (Benito and Thorndycraft 2004; Greenbaum et al. 2014). Similar to historical or sedimentary records, peak discharge reconstructions based on tree-ring records have produced distinct percentile estimations (Ballesteros-Cánovas et al. 2013a, b, 2016b). The advantage of tree-ring records over other paleoflood sources (Baker 2008) is that they provide annually resolved data which can easily be referenced to systematic records. The ubiquity of trees, especially in mountain regions, as well as their high temporal and spatial resolution makes tree-based flood reconstructions highly suitable for hazard assessments. Ballesteros-Cánovas et al. (2016a) recently demonstrated that tree-ring approaches, combined with regional flood-frequency assessments, maximize available information and thus can reduce uncertainties related to flood quartile estimation.

The examples presented on root erosion merely represent snapshots of recent developments in root-based reconstructions of various types of erosion processes in different geographic settings around the globe. They illustrate the substantial progress achieved in the field of dendrogeomorphic erosion analysis. Through the systematic comparison of exposure signals in roots and mean erosion rates with data obtained from monitoring devices, the precision of dendrogeomorphic approaches has been improved substantially. These methods have clear advantages over the shorter time series obtained with repeat monitoring or over longer, but more coarsely resolved records. The resolution of dendrogeomorphic data and the time windows typically covered by roots also facilitate the comparison of averaged erosion rates with meteorological records, and the analyses also exhibit a much better cost–benefit ratio than most other techniques used to infer erosion. Despite limitations primarily related to the presence of trees and shrubs in the study area, to climate seasonality and to the age of roots available for analysis, we believe that root-based erosion assessments therefore constitute a valuable alternative to empirical models, especially in regions where data for calibration and validation are completely missing.

Last but not least, this chapter also clearly calls for more linkages between disciplines and an improved understanding of tree reactions and the formation of growth anomalies with relation to tree age or tree size. A first set of quite preliminary answers has been provided in one of the sections of this chapter, but much more research is still needed in this field.