Conference Agenda

The investigation of the settlement structure and its connectivity can provide information about the influence of humanity on its environment. In the study presented here, we examined the percolation of built-up areas worldwide, which was carried out as part of the study “Landscape Criticality in the Anthropocene - Biodiversity, Renewables and Settlements (CriticaL)”. Percolation describes the distance from which larger cluster structures arise and therefore provides information on how close settlement structures are to each other and how permeable they are. Findings on this are of interest not only for economic but also for ecological issues, e.g. how permeable larger settlement areas are for animals. We pursue two measurement approaches with which we can describe for different spatial units how the settlement structures are connected, the so-called critical distance, and how wide open spaces are. Several global datasets were created, which address different levels of spatial resolution. In addition to administrative spatial units, such as country borders and subnational levels, critical distances were also calculated for different grid widths (5 to 0.5 degrees). Furthermore, raster datasets were created in which moving window approaches with different radii were applied in order to be able to make statements at the smallest possible scale. At the raster level, we have also calculated the permeability for different directions (north to south, west to east), which is of particular interest for studies of animal pathways. Besides providing the data, an easy-to-use Docker tool (CriticaL Mapper) has also been developed so that the approaches presented here can be easily applied to custom data. Algorithms specially developed for huge cluster calculations were integrated, which basically run with PostgreSQL and PostGIS in a Python environment. With the help of parallelization techniques, even larger data sets can be processed with high performance.

Urban expansion substantially contributes to global biodiversity loss. As a result, conservationists aim to reconcile biodiversity and urban development and increasingly intend to promote biodiversity within cities. Given the limited space in urban areas, improving ecological connectivity, thus enabling animal movement between remote green spaces, is a key strategy to support urban-dwelling species.

Urban connectivity models are now commonly used to understand animal movement, but translating these models into concrete spatial recommendations, such as where to establish new green spaces or how to shape them in new neighbourhoods, remains challenging. In this study, we present an approach that combines empirically derived connectivity models with reinforcement learning, an emerging field of artificial intelligence, to perform an optimisation. The aim of this optimisation is to identify the locations where greening would most effectively increase the permeability of urban environments for animal movement and hereby support animal establishment in cities.

Our connectivity model used observations of a frequently observed urban bird species, the common blackbird (Turdus merula), to estimate typical movement distances and species-specific landscape resistance. It then utilised these parameters to derive landscape graphs and calculate the Probability of Connectivity, a relative measure of the connectivity of the overall landscape. Our reinforcement learning framework calculated the Probability of Connectivity as a reward function, where a computational agent iteratively placed green patches within the study area, received feedback on the resulting connectivity, and adjusted its strategy accordingly.

Our coupled models identified locations that would most effectively increase ecological connectivity if greened. We found that elongated stepping stones produced the greatest gains in connectivity. Moreover, we determined landscape configurations where small interventions could yield disproportionately high improvements in permeability. Overall, we demonstrated the strength of reinforcement learning to solve spatial optimisation problems in urban ecology and planning. By integrating empirically derived ecological connectivity models with AI-based optimisation, our approach offers a decision-support tool for enhancing biodiversity through targeted urban greening.

Landscape ecology theory has long posited that landscape composition and landscape configuration underlie ecological processes such as animal movement. While clear global-scale evidence exists supporting the key role of landscape composition in mediating animal movement, the effects of landscape configuration remain poorly understood at a global scale.In this study, we utilized a newly developed global settlement percolation dataset to examine the effect of configurations of human settlements on terrestrial mammalian mobility in human-modified landscapes. The critical distances provided by the global settlement percolation dataset represents typical distances among settlement patches (or “porosity”) of the given landscape, and we expect higher settlement porosity translate to greater landscape permeability for animal movement. Using animal tracking data from 8,379 individuals from 66 species, we showed animals in less porous landscapes move one-third less than in more porous landscapes, and this effect is in addition to the amount of human impacts. This research provided clear supports to the classic landscape ecology theory at the global scale that landscape configuration and composition are both key determinants of animal movement.

The aim of an ongoing project by the Federal Environment Agency is to forecast the loss of open space by 2030 and 2050. A three-step process is being used for this purpose. With the help of the QMORE_PR model, future developments in settlement and transport areas in Germany are being projected at the district level up to the year 2050. The model takes into account demand-related factors such as economic growth, demographic changes, and sectoral structural change. By integrating it into the INFORGE macroeconomic model, the effects of a wide variety of developments on land use can be estimated. In addition, it has been expanded to include a module for mapping the expansion of renewable energies. In a second step, the land requirements determined at the district level are modeled in a small-scale grid using the Land Use Scanner model. The small-scale localization allows for an intersection with municipal boundaries, protected areas, and other regional settings in the course of the final evaluation.

Anthropogenic barriers alter and disrupt animal movements and are extremely challenging obstacles for long-ranging ungulate species. Understanding these disruptions and mitigating adverse impacts on wildlife requires knowledge of species-specific habitat needs and behaviors. How different linear features such as roads, railways and fences affect ungulate behaviors may differ greatly in relation to the feature permeability and disturbance, which in turn may be shaped by differences in physical structure or human activity. In the South Gobi region of Mongolia linear infrastructure is expanding at an unprecedented rate, s creating barriers for migratory wildlife. We assessed the interactions of 99 Mongolian wild-ass also known as Khulan ( Equus hemionus hemionus) at different types of linear infrastructures (roads, railways) and anthropogenic presence (mines) in the South Gobi of Mongolia. We calculated the number of interactions between the khulan trajectories and linear features and further categorized them into different behavioral responses (normal vs altered). Using a generalized mixed-effects model, we analyzed the factors influencing infrastructure crossings, including traffic volume, infrastructure type, time of day, and proximity to linear features. Our results revealed significant behavioral changes in response to different linear infrastructures. Khulan took notably longer time to cross these features, with paved roads exhibiting the highest traffic volumes proving to be the least permeable. Moreover, these roads were typically crossed at night during periods of low traffic. Furthermore, frequency of altered movement behavior around linear infrastructure increased three- to fourfold over a ten-year period. Highly mobile species utilize resources across vast landscapes, and we found strong evidence that linear infrastructures are severely impacting khulan movement behaviors in the South Gobi. These altered behaviors, driven by barriers, likely reduce access to quality forage, ultimately jeopardizing fitness and survival. Our findings emphasize key infrastructure and landscape features that, if properly managed, could enhance landscape permeability. This research highlights that the various linear infrastructure being developed in the South Gobi is altering behavior and creating challenges for migratory species whose ability to move across the landscape is crucial to their survival. Mitigation measures of linear infrastructure must be initiated to preserve connectivity to ensure the survival of khulan and other species in the region.