Dispersal and Connectivity in Physical, Seasonal, and Extreme Landscapes

Abstract

Connectivity is a crucial prerequisite for animals to move and disperse, which in turn promotes genetic exchange, facilitates range-shifts, and enables the recolonization of vacant habitats. Preserving and establishing movement corridors that enhance connectivity and facilitate dispersal have therefore become tasks of utmost importance in conservation science. Nevertheless, many of the methods employed to study dispersal and connectivity suffer from important limitations that inhibit a more holistic understanding of landscape connectivity. In this doctoral thesis, I investigate various aspects of animal dispersal and landscape connectivity, focusing on several of these limitations. Specifically, I propose a novel method for quantifying landscape connectivity, examine the impact of changing environmental conditions due to climate change, and investigate the role of seasonal factors when modeling connectivity. I also present and compare multiple methods for dealing with temporally irregular data when estimating habitat and movement preferences of dispersers from GPS data. These aspects are researched primarily using dispersing African wild dogs from northern Botswana as a study system. Due to the study species' wide-ranging behavior and the study areas' highly seasonal habitats, the study system offers a unique opportunity to investigate patterns of dispersal and connectivity in light of the outlined considerations.

Chapter 1 introduces the general background and research topics of this thesis. In particular, I discuss the importance of dispersal and connectivity in the context of conservation science and highlight multiple limitations in current connectivity research. I also describe the study system and give a brief overview of the data collection and general research methods.

Chapter 2 introduces a novel and simple workflow for examining landscape connectivity using simulations from individual-based movement models. To date, connectivity is primarily investigated using least-cost path models or circuit theory. However, both approaches require highly subjective permeability surfaces as inputs and make unreasonable assumptions that are rarely met by dispersing individuals. The proposed simulation workflow overcomes these shortcomings and provides a powerful alternative for studying connectivity via simulated dispersal trajectories and a complementary set of connectivity metrics. I exemplify the application of the proposed workflow and assess connectivity for dispersing wild dogs in the Kavango-Zambezi Transfrontier Conservation Area, revealing major movement corridors and highlighting dispersal hotspots. This chapter was published in Landscape Ecology in 2023 (https://doi.org/10.1007/s10980-023-01602-4).

Chapter 3 investigates how changing environmental conditions due to climate change affect on-the- ground conditions and, ultimately, species’ ability to disperse. Specifically, I investigate how changes in the flooding regime of the Okavango Delta impact wild dogs' ability to disperse between adjacent regions. For this, I combine the simulation framework from Chapter 2 with long-term remote sensing data of the Delta's flood extent and simulate dispersal under an increased and a reduced flood scenario. Both scenarios represent conceivable outcomes under the impact of climate change. I show that an increased flood reduces connectivity and prolongs dispersal durations, yet that the opposite is true under a reduced flood. Importantly, I highlight that the likely hotspots for human-wildlife conflict shift depending on future flood conditions. This chapter was published in Global Change Biology in 2024 (https://doi.org/10.1111/gcb.17299).

Chapter 4 conceptually motivates that seasonality can enter connectivity analyses at three distinct modeling stages, resulting in multiple configurations that greatly differ in terms of the dynamism they encapsulate. To test whether the incorporation of seasonality into connectivity analyses improves their predictive ability, I fit the models associated with each configuration and employ a rigorous validation procedure that compares predicted with observed movement. I also demonstrate that the simulation workflow presented in previous chapters can accommodate for a previously unseen degree of seasonality by allowing the landscape to change as the simulated dispersers move. Surprisingly, I find that predictions from the fitted models only marginally improve upon increasing the degree of seasonality. Nevertheless, inferred patterns of connectivity differ, with connectivity being more evenly distributed when seasonality is accounted for.

Chapter 5 revisits the statistical framework of integrated step-selection functions, which are frequently employed in connectivity studies to assess habitat preferences from GPS data. Currently, step-selection functions require temporally regularly spaced GPS data, which implies that irregular data need to be removed. This can result in a substantial loss of data, especially in studies where data are already scarce. Hence, I propose and compare several methods for dealing with irregular data, thereby allowing to retain additional data for modeling. I compare different methods using simulated data with known parameters and show that retaining irregular data can aid to reduce model uncertainty if appropriate methods are employed. Finally, I exemplify the application of the best performing method using GPS data collected on a spotted hyena from northern Botswana. The use of hyena data instead of wild dog data was mainly due to the availability of an extensive dataset on a single spotted hyena. This chapter was published in Movement Ecology in 2024 (https://doi.org/10.1186/s40462-024-00476-8).

Finally, in Chapter 6, I summarize my findings and embed them in a broader context. I also provide an outlook for topics that require further research, and provide conservation insights for the African wild dog.

Overall, this thesis provides a detailed investigation of dispersal and connectivity for the endangered African wild dog and propose several novel methods and approaches. It thereby challenges established practices and suggests new avenues to model dispersal movements. By combining methodological novelty with conservation insights, this thesis not only advances our understanding of connectivity in general, but yields specific insights that facilitate the implementation of targeted conservation measures. The dispersal model refined throughout the chapters furthermore sets the foundation for a comprehensive population viability analysis that realistically captures how dispersers move across the landscape.