Text

theses

Understanding the habitat preferences of large mammals is critical for their conservation and management. Resource selection functions (RSFs) can be used to assess these preferences, but they often only incorporate environmental information, such as percent tree cover, at a spatial resolution determined by the source of the data, i.e. satellite imagery. Organisms may respond to their surroundings at larger or smaller spatial scales, and thus the spatial scale of the data may be biologically irrelevant for the species in question. Instead, habitat selection should be assessed on a continuum of spatial scales to identify the ones that are most relevant to the organism. This can be accomplished by locally averaging, or smoothing, layers of environmental information to generate coarser representations of the organism’s surroundings. In Chapter 2, I model habitat preferences of savannah elephants with and without multiple spatial scales. Models that incorporated multiple spatial scales performed better and made different predictions regarding the spatial distribution of high-quality habitat throughout a landscape. This chapter has been published in PeerJ (https://peerj.com/articles/504/). The inclusion of multiple spatial scales for numerous environmental variables can lead to problems in model choice, as not all combinations of variables can be evaluated. In Chapter 3, I model habitat preferences of brown bears by first using the least absolute shrinkage and selection operator (lasso) to order the variables by their importance. I then fit models of increasing complexity by adding one variable at a time in reverse order of importance. I also incorporate the presence of neighboring individuals to account for the possible competitive exclusion of optimal habitat, but this was found not to affect the habitat chosen. In Chapter 4, I determine whether individual desert bighorn sheep have different habitat preferences when they inhabit two mountain ranges with differing availability of freestanding water. For each environmental variable, both a full parameter and a ‘difference’ parameter are estimated, depending on where the sheep movement occurs. Different preferences were found for vegetation at multiple spatial scales, implying that bighorn sheep can utilize the moisture found within vegetation to survive when freestanding water is not available.

ETD_6675

Ph.D.

Includes bibliographical references

by Andrew F. Mashintonio

doi:10.7282/T3KK9DSX

ETD doctoral

The author owns the copyright to this work.