Functional landscape connectivity models in applied conservation: spatio-temporal variability, on-ground validation and application for African elephants (Loxodonta africana)

Electronic versions


  • Liudmila Osipova

    Research areas

  • PhD, School of Natural Sciences, African elephant, conservation planning, fences, human-elephant conflict, landscape connectivity, step-selection function


Habitat destruction and fragmentation are among the main drivers of global biodiversity loss and developing management approaches that effectively maintain or restore landscape connectivity is one of the major roles of systematic conservation planning. Movement data-based functional connectivity models have great potential for applied conservation as they reflect more than just species habitat preferences and can integrate spatial and temporal dynamics. However, there are few challenges for using such dynamic models in wildlife conservation, including a lack of clear up-to-date methodological frameworks and policy guidelines, and insufficient evidence that connectivity-based conservation corridors are effective.
This dissertation aims to demonstrate how resistance-based functional connectivity models accounting for seasonal and individual variability can improve conservation management decisions. Using radio-tracking data from elephants in the Borderland area between Kenya and Tanzania, step-selection functions were applied to create seasonal landscape resistance surfaces. Based on these seasonal models, I predicted movement corridors connecting major protected areas using circuit theory and least-cost path analysis.
In Chapter 2, I demonstrated how incorporating seasonal variability can make a distinct difference in the final outcomes in connectivity modeling, and how disregarding these differences can negatively affect management decisions on the ground, especially in the areas prone to drought. For this, I developed a new analytical framework for incorporating individual and seasonal variability in resistance surface modeling. I compared space-use predictions derived from the novel approach to predictions obtained using a method typically applied in connectivity research. By comparing empirical elephant movements with simulated movements from both approaches, I demonstrated that my novel framework predicts actual space-use patterns significantly better than the commonly applied models.
In Chapter 3, I applied a time series of seasonally distinct landscape connectivity models to assess how a fence built for human-wildlife mitigation in the study area could affect elephant connectivity in the future. Fence integration into the landscape will cause overuse of habitat patches in other agriculture areas, thereby potentially intensifying human-elephant conflict in new areas. This will likely negate the conservation benefits of fencing across the landscape despite local benefits. These results lead to the conclusion that if fencing is employed on a broader scale, then corridors should be integrated within protected area networks to ensure local connectivity of affected species and the implementation of fencing should be incorporated into the impact assessment process.
The final Chapter 4 focused on the validation of the functional landscape connectivity models and movement corridors using two independently collected datasets. I used multiple-year aerial counts of elephants to evaluate the connectivity model, and a field survey to assess the performance of predicted corridors.
The results of this dissertation confirmed that resistance-based connectivity modeling could have a strong predictive power, provided that seasonality and individual variability are accounted for. Analytically considering seasonal effects and individual movement behavior can significantly improve the performance of connectivity models and their effectiveness in conservation planning and wildlife management.


Original languageEnglish
Awarding Institution
  • Matthew Hayward (Supervisor)
  • Niko Balkenhol (External person) (Supervisor)
Award date17 Jun 2019