Mapping Elephant Habitat Connectivity

This study aimed to identify potential African elephant (Loxodonta africana africana) wildlife corridors (WC’s) between the South Luangwa National Park (SLNP) and Petuake Protected Area (PPA) in Eastern Zambia. These conservation areas were chosen because the elephant population in the SLNP is reaching its carrying capacity. A WC between the two can enable African elephant dispersal to alternative habitat in the PPA and other protected areas. An area between the SLNP and the PPA has been identified as potential African elephant habitat, however, this area is unprotected. My hypothesis is that this area has greater functional connectivity within potential African elephant habitat than outside it. If this area has greater functional connectivity, then it should be considered being turned into a conservation area and maybe even a WC.

First, I obtained Rapideye data for the 3 study sites in Eastern Zambia from PlanetLabs. Each study site represents a habitat type. Site 1 represents Potential Habitat, Site 2 represents NonHabitat and Site 3 represents Mixed Habitat. I used a Maximum Likelihood Classification in ArcGIS to determine land cover for each habitat type. Then I obtained a roads layer and the Human Influence Index layer from NASA SEDAC. I also incoporated a slope layer derived fro

Then I combined all 4 habitat factor layers using the Weighted Sum tool in ArcGIS to determine elephant habitat suitability for each habitat type.
Then each elephant habitat suitability map was run in CONEFOR to analyze connectivity with a dispersal distance of 11.5 km. I obtained several indicators but focused on using IICConnector, which indicated the connectivity of suitable habitat within each site. Therefore functional connectivity (”connectivity”) was represented by each sites IICConnector values.
I statistically analyzed my results using a regression analysis and an ANOVA.

Potential habitat showed greater areas of high suitability amongst all the sites, and non-habitat had the lowest area of high suitability. This indicates that potential habitat is the most highly suitable elephant habitat amongst the sites. This result also confirms that potential habitat is most likely to be able to host elephant populations, as the patches of high suitability are larger compared to the other sites.

The ANOVA was run using R in RStudio. Connectivity of non habitat was found to be significantly different from other habitat types. The results of the SLR between habitat type and connectivity reveal that non-habitat is significant within the regression. The boxplot shows that non habitat’s connectivity is lower than other habitat types, and that it had the least variation.

Non-habitat is mostly unsuitable to support African elephants, with large patches of low suitability that are close together. Non-habitat had the greatest area of low suitability, while mixed habitat had the smallest area of low habitat suitability.

The Multiple Linear Regression (MLR) was performed using R in RStudio. The results of the MLR showed that Land cover was most significantly correlated with connectivity. Land cover also had a positive correlation with connectivity, showing that as with increased woodland area, connectivity also increased.