Forest ecosystems are sensitive to global change; especially at the ecotones we expect high sensitivity to changes in climate, disturbance regimes or land use. For instance, the treeline ecotone is expected to move upward in elevation with global warming. The advent of machine learning, specifically computer vision, provides powerful tools for monitoring ecotones across large spatial scales using remote sensing data. In this study, we focused on the spatiotemporal development of ecotones bracketing the subalpine forest belt (i.e., the upper boundary, formed by the treeline, and the ecotone to montane forests as the lower boundary) in a protected forest landscape in the European Alps. Our objectives were threefold: First, we aimed to identify trees and shrubs on historic and recent orthophotos using deep learning, with special attention to integrating multiple sensor types into one computer vision framework. Second, based on the computer vision inference, we sought to map the a) treeline and b) montane-subalpine ecotone. Third, we aimed to describe the spatiotemporal changes occurring in both ecotones.

We based our analysis on historic and recent aerial images of Berchtesgaden National Park in the Northern Alps, covering roughly 210 km² in nine time steps from 1953 to 2020. The images were captured through both analog (panchromatic and color infrared) and digital (color infrared, RGB) sensors. To generate training data for deep learning, we manually interpreted randomly distributed 0.5 ha segments across all time steps, resulting in over 110,000 annotations of coniferous and broadleaved trees, shrubs, and standing dead trees. We tested different instance segmentation frameworks and selected the best performing model architecture to create wall-to-wall tree maps for each time step. Using structure and composition of the tree maps, we spatially delineated the ecotones and tracked their changes over time.

We did not find a spatially consistent pattern of ecotones shifting upwards, however we were able to identify areas of change and stability linked to climate, topography, disturbances and land use. We observed remarkable local upward shifts in ecotones, particularly of the montane-subalpine ecotone, which shifted up to five times faster than the treeline. In general, we found that subalpine forests, situated between the two ecotones, decreased in area because of an upward shift of its lower boundary, and exhibited an increase in crown cover over time.

Changes in these ecotones and related vegetation zones pose challenges to conservation, restoration and management. Our approach can help to address these challenges, e.g., in the combination with habitat modelling.

How to cite: Maroschek, M., Seidl, R., Senf, C., and Rammer, W.: Tracking shifts of mountain forest ecotones in aerial imagery with deep learning, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-1844, https://doi.org/10.5194/egusphere-egu24-1844, 2024.

Ecotones – transition zones between ecosystems – are sentinels of global change, as they are sensitive to changes in environmental conditions and land use. The alpine treeline ecotone - where the continuous, subalpine forest transitions into the treeless alpine zone – is a characteristic feature of many mountain ecosystems. The transition at the ecotone can be characterized by distinct treeline patterns. Treeline patterns can be simple, such as sharp transitions from forest to alpine vegetation, or complex, e.g., islands of trees and krummholz in a matrix of alpine vegetation. This variation can mediate the impact of global change at the alpine treeline ecotone. However, large-scale attribution, e.g., for an entire mountain range, and spatiotemporal quantification of treeline patterns remain challenging. Automated methods, such as deep learning-based computer vision systems, can help to overcome these challenges. Building on existing definitions of treeline patterns, we aim to characterize the alpine treeline ecotone for the entire mountain range of the European Alps. Our particular objectives are:

• To characterize the patterns of a representative sample of the alpine treeline ecotone of the European Alps based on remote sensing information as training data for deep learning.
• To quantify treeline patterns across the Alps and identify spatial differences in the prevalence of patterns.

In an alpine treeline ecotone, we considered the transition between three vegetation classes: trees (i.e., upright woody plants with a minimum height of 3m), krummholz (i.e., stunted trees and woody shrubs), and treeless alpine vegetation. Three spatial patterns were considered for trees and krummholz describing their state: closed, islands, or isolated individuals. The transitions between these states across elevation, a total of 24 combinations, were used to quantitatively characterize treeline patterns. We selected 1,000 randomly distributed elevational transects between 1,100 and 2,800 m.a.s.l. that include the alpine treeline ecotone across the European Alps. For each transect, we classified treeline patterns for areas of 90m×90m using satellite and orthophoto images. Based on this dataset, we quantified differences in treeline patterns and their distribution in elevation across the European Alps: While in the Prealps, the alpine treeline ecotone is located in lower elevations and treeline patterns tend to be more complex, the ecotone is higher in elevation and less complex in the Central Alps.

The quantification of treeline patterns and their distribution can serve as a basis for further investigations of the alpine treeline ecotone and its spatiotemporal development. We provide an outlook for a deep learning approach that uses the presented dataset combined with a time series of spectrally unmixed satellite data, i.e., fractional abundances of land cover per pixel, as training data. Utilizing satellite data of the past 35 years in annual resolution, we will be able to automatically classify and analyze treeline patterns and their changes across the entire European Alps.

How to cite: Kruse, J.-S., Rammer, W., Mandl, L., Seidl, R., and Maroschek, M.: Characterizing Spatial Patterns of the Alpine Treeline Ecotone Across the European Alps, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-10296, https://doi.org/10.5194/egusphere-egu24-10296, 2024.

Previous research has shown that (1) treelines are shifting upward in elevation on high mountain peaks worldwide, and (2) the rate of the upward shift appears to have increased markedly in recent decades. Because treeline shift is a process manifested over broad scales of space and time, a particular challenge has been that of obtaining a broad-enough view of patterns of treeline shift to permit inferences about geographic and environmental patterns. What is more, intensive studies of treelines have been concentrated in North Temperate regions, such that little information is available about treeline shift patterns in the Tropics. We have attempted to broaden this viewpoint by means of analysis of long  time series of vegetation indices derived from Landsat imagery obtained and analyzed via Google Earth Engine for the 1980s to present. We sampled vegetation indices at points spaced every 100 m along 100 km transects radiating out from 120 high peaks across western North America (Canada to Central America); considerable data preparation was necessary, including ending transects <2 km into closed forest, identifying current treelines via reference to Google Earth imagery, and consideration only of up to <1 km above treeline. Patterns that emerged were—as is well known—that treelines are generally higher at lower latitudes, but also that magnitude of treeline shifts is nonrandomly distributed with respect to latitude, location with respect to coastlines, and size of the mountain mass within which the peak is located. Although analyses are continuing at the time of preparation of this abstract, this analysis offers a broadscale view of treeline shifts over a period of almost 40 years, and over a geographic span of more than 40° of latitude.

How to cite: Corimanya, J., Jimenez-Garcia, D., and Peterson, A. T.: Patterns of Treeline Rise with Climate Change Across Western North America from the 1980s to Present, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-4049, https://doi.org/10.5194/egusphere-egu24-4049, 2024.

Treelines are dynamic ecotones largely influenced by climate and land use changes. The increasing development of remote sensing techniques and the interest on the ecological effects of global warming on forest vegetation have raised the number of treeline studies.

The aims of this study were: i) to define an automatic approach for mapping the current position of the upper forestlines in the Italian Alps and Apennines and ii) to locate hotspots of long-term vegetation dynamics using Landsat-based spectral trend analysis. Hotspots will serve us to analyse the ecological drivers of vegetation change and to predict future vegetation dynamics.

We used the Tree Cover Density (TCD) dataset (Copernicus Land Monitoring Service) and a nationwide digital elevation model to define the polylines representing the forestlines for the reference year 2018. We used the main Italian mountain peaks, extracted from the Global Mountain Biodiversity Assessment (GMBA) dataset polygons, as reference points to detect only the upper forest ecotones based on the elevation difference between peaks and forest pixels. We defined our study areas by applying a positive and negative buffer around the forestlines and we calculated several spectral vegetation indices (e.g. NDVI, EVI, Tasseled Cap Angle) from Landsat timeseries of the last 40 years. In this way, we inferred inter-annual vegetation dynamics, discriminating two sub-areas of interest: the closed forest (below the current forestline) and the upper treeline ecotone (above the current forestline). It should be noted that on the Alps, treelines mainly host conifer species, whereas on the Apennines, broadleaf species (mostly European beach) prevail. We tested the significance of long-term spectral trends through a Mann-Kendall test for monotonicity that accounted for autocorrelation in space and time.

An important outcome of the study was to set up a replicable and unsupervised method to enhance the study of vegetation dynamics at treeline ecotones. This approach will allow the delimitation of the forestlines on a global scale and an ecologically sound comparison between different treeline ecotones. This study is the first step in a nationwide project and will provide the basis for future local-scale investigations of treeline ecotones.

How to cite: Baglioni, L., Morresi, D., Tonelli, E., Lingua, E., Marzano, R., and Urbinati, C.: Forestline detection and treeline ecotone dynamics in the Italian Alps and Apennines by satellite remote sensing, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-15620, https://doi.org/10.5194/egusphere-egu24-15620, 2024.

Treeline ecotones spatial patterns and dynamics are influenced by factors acting at regional, landscape, and local scales. It is widely accepted that treelines change in complex ways depending on their diverse structural features and environmental conditions.
The high variability of environmental conditions and ecological drivers hampers the creation of a general pattern from case studies. A multi-scale approach applied at numerous locations is needed to discriminate between natural and anthropogenic factors that are driving treeline dynamics. Remote sensing techniques are today fundamental tools for a comprehensive assessment of the spatial heterogeneity of treeline patterns and their changes over space and time. Continuous improvements in remote sensing platforms, sensors, and methodologies have considerably increased the quality and reliability of spatial information, such as forest maps, which are essential for monitoring ecotonal dynamics.
In this study, we aimed to comprehensively map individual tree canopies at the treeline ecotone in 10  different sites distributed across the Italian Alps by integrating field and UAV-based data. We first mapped the position of the forestline using the 2018 pan-European Tree Cover Density layer provided by the Copernicus Land Monitoring service. In particular, we considered the pixel line where the tree canopy cover was less than 10% as the forestline. Field data consisted of position, height, and species of 100 trees taller than 50 cm scattered over a 9-hectare area. Each site was also flown over by a multirotor drone to produce an RGB orthomosaic, a digital surface model, and a canopy height model. A total of 1016 individual canopies of different coniferous species were manually classified on the orthomosaics with the aid of semi-automatic annotation software. These data were used to train a deep learning model based on the Mask R-CNN algorithm for object detection and segmentation. The classification masks were lastly combined with a canopy height model providing 3-dimensional information allowing to measure tree height. Preliminary results evidenced that remotely sensed data collected with low-cost equipment such as commercial drones with RGB cameras, coupled with the proposed canopy detection method can be used to produce highly accurate and reliable maps of treeline ecotones. These maps will serve as a starting point to study and monitor the spatio-temporal dynamics of treeline ecotones at the local scale and how they affect biodiversity in high-altitude environments.

How to cite: Carrieri, E., Meloni, F., Urbinati, C., Lingua, E., Marzano, R., and Morresi, D.: Treeline spatial patterns for biodiversity monitoring detected by spectral and 3D information from UAV‐based aerial imagery, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-11942, https://doi.org/10.5194/egusphere-egu24-11942, 2024.

The forest line is a key feature of mountain landscapes around the world. Currently, most forest lines in the Northern Hemisphere are rising due to the combined effects of land-use and climate changes. This upward shift has led to major changes in the functions and services provided by the adjacent socio-ecosystems (e.g. carbon sequestration, biodiversity hosting, services to people…). However, it has not been elucidated how the recent forest-line upward shift fits into the longer context of land abandonment (occurring since the beginning of the 19^th century in France), and how it is currently responding to the accelerating global warming. To answer this question, we assessed the elevation change of the forest line over the French northern Alps since the forest minimum (mid-19^th century in France) using old and current land cover maps.

Three digitalised maps: the État-Major map, BD Forêt® v1.0 and BD Forêt® v2.0 were used to display forest cover at three dates: 1859, 1994 and 2007, respectively. These maps were standardized and combined with a digital elevation model to estimate the average elevation of the subalpine forest lines for 178 municipalities across the French departments of the Northern Alps: Haute-Savoie, Savoie and Isère. We compared forest-line elevations between dates and municipalities to explore temporal and spatial patterns.

The forest line in the French Northern Alps has risen by an average of 152 ± 18 m from its ancient position (1879 ± 21 m a.s.l. in 1859) to its current position (2032 ± 12 m in 1994).  However, no general upward shift was observed during the most recent period from 1994 to 2007, as the forest-line position was 2013 ± 13 m in 2007. In the Haute-Savoie department, a downward shift of 69 ± 12 m was even observed, while forest lines in Isère and Savoie were stable. Forest-lines upward shift in the French Northern Alps has been driven by agricultural abandonment, mountain land restoration and global warming since the period of the forest minimum (around 1860). However, it is noteworthy that forest line dynamics are no longer influenced by these factors nowadays and do not follow the acceleration of temperature increase. The current recession of the forest lines may be attributed to anthropogenic pressure related to the tourism activity. This new anthropogenic pressure corresponds to the development of alpine ski resorts and the increase in the human population in Haute-Savoie since 1925, and later in the other departments.

This large-scale spatial and temporal study shows how global and regional factors interact in the long-term to shape mountain landscapes, in particular the ecotone between subalpine forest and alpine grassland. Today, the dynamics of this ecotone is still linked to the contradictory tensions that divide our societies (conservation vs. exploitation). This is why we advocate the cautious management of alpine forest line ecotones, which could contribute to carbon sequestration and biodiversity conservation, provided they are not subjected to excessive human pressure (tourism and grazing).

How to cite: Delpouve, N., Rathgeber, C., Bergès, L., Dupouey, J.-L., Chauchard, S., and Leroy, N.: Is anthropogenic pressure limiting the climate-induced upward shift of the subalpine forest line in the French Northern Alps?, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-15287, https://doi.org/10.5194/egusphere-egu24-15287, 2024.

 Despite the importance of species interaction in modulating the range shifts of plants, little is known about the responses of coexisting life forms to a warmer climate. Here, we combine long-term monitoring of cambial phenology in sympatric trees and shrubs at two treelines of the Tibetan Plateau, with a meta-analysis of ring-width series from 344 shrubs and 575 trees paired across 11 alpine treelines in the Northern Hemisphere. Under a spring warming of + 1°C, xylem resumption advances by 2–4 days in trees, but delays by 3–8 days in shrubs. The divergent phenological response to warming was due to shrubs being 3.2 times more sensitive than trees to chilling accumulation. Warmer winters increased the thermal requirement for cambial reactivation in shrubs, leading to a delayed response to warmer springs. Our meta-analysis confirmed such a mechanism across continental scales. The warming-induced phenological mismatch may give a competitive advantage to trees over shrubs, which would provide a new explanation for increasing alpine treeline shifts under the context of climate change.

How to cite: Liang, E., Li, X., Camarero, J. J., Rossi, S., Zhang, J., Zhu, H., Fu, Y. H., Sun, J., Wang, T., Piao, S., and Peñuelas, J.: Warming-induced phenological mismatch between trees and shrubs explains high-elevation forest expansion, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-10161, https://doi.org/10.5194/egusphere-egu24-10161, 2024.

Since climatic treelines track the elevational position of isotherms across the globe, it is not the question if, but when and how they will arrive at a novel steady-state position. After briefly recalling the essential difference between the edge of the realized and fundamental niche of the life form tree (not to be confused with species’ range limits), I will present data on recent climatic trends in the Alps based on long term meteorological records. Two years of in-situ temperature records from Pinus cembra trees growing right at the current upper edge of tree size individuals in Eastern Tyrol (supplemented with data from the Swiss Engadin region), make it obvious that the current high elevation record positions around 2500 m elevation are lagging substantially behind the upslope shift of the isotherm. This explains, why these trees grew so exceptionally rapid over the past 10 years, partly growing a meter in height in only 6-8 years. The locations with rapid tree radiation are all under nutcracker control. These data permit projections on forthcoming treeline shifts. For the Austrian Alps, the current uppermost trees represent all-time elevation records, and will soon out-range the uppermost fossil elevation records of trees that date back to the warmest period of the Holocene. Suggested reading: Körner C, Hoch G (2023) Not every high-elevation or high-latitude forest edge is a treeline. J Biogeography, open access.

How to cite: Körner, C.: Treeline resposes to 2K warming in the Alps in half a century, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-20484, https://doi.org/10.5194/egusphere-egu24-20484, 2024.