Dataset of how animal movement paths in a landscape vary due to wildfire
Data files
Feb 08, 2024 version files 26.60 KB
-
fidelity_data.csv
-
README.md
-
script.R
-
shrub_cover_data.csv
Abstract
Animals navigate landscapes based on perceived risks vs. rewards, as inferred from features of the landscape. In the wild, knowing how strongly animal movement is directed by landscape features is difficult to ascertain, but widespread disturbances such as wildfires can serve as natural experiments. We tested the hypothesis that wildfires homogenize the risk/reward landscape, causing movement to become less directed, given that fires reduce landscape complexity as habitat structures (e.g., tree cover, dense brush) are burned. We used satellite imagery of a research reserve in Northern California to count and categorize paths made primarily by mule deer (Odocoileus hemionus) in grasslands. Specifically, we compared pre-wildfire (August 2014) and post-wildfire (September 2018) image history layers among locations that were or were not impacted by wildfire (i.e., a Before/After Control/Impact design). Wildfire significantly altered spatial patterns of deer movement: more new paths were gained and more old paths were lost in areas of the reserve that were impacted by wildfire; movement patterns became less directed in response to fire, suggesting that the risk/reward landscape became more homogenous, as hypothesized. We found evidence to suggest that wildfire affects deer populations at spatial scales beyond their scale of direct impact and raises the interesting possibility that deer perceive risks and rewards at different spatial scales. In conclusion, our study provides an example of how animals integrate spatial information from the environment to make movement decisions, setting the stage for future work on the broader ecological implications for populations, communities, and ecosystems, an emerging interest in ecology.
README: Dataset of how animal movement paths in a landscape vary due to wildfire
This project has one script that performs all analyses and generates all figures. This project has two datasets.
Dataset 1 (fidelity data.csv) is used for the majority of the analyses in the paper.
Site and subplot are self explanatory - type indicates whether the data was collected on a plot that was impacted by wildfire between 2014 and 2018.
The rest of the columns are counts of the number of deer paths of a certain thickness in 2014 vs in 2018. For example, THIN_to_thick are paths that were thin in 2014 and became thick in 2018. NONE_to_(thick/medium/thick) are paths that didn't exist in 2014, whereas (THICK/MEDIUM/THIN)_to_none are paths observed in 2014 that disappeared.
Plots 10 and 20 do not have associated data because originally we were planning to have 10 replicates per burn treatment but could only find 9 that met our criteria for inclusion, and did not want to introduce errors by renumbering plots after the fact.
Dataset 2 (shrub cover.csv) is used for a supplementary analysis of how much shrub cover changed over time and in response to wildfire.
Img_Name is simply an ID used so the data collector could differentiate experimental units. It is not used in the script.
Year indicates whether shrubs are being quantified in 2014 or 2018, whereas Plot is which plot shrubs were measured in. Shrub_cover is a measure of the percent cover of shrubs in a plot.
Notes were notes made by the collector.
Methods
Animal paths were traced from satellite imagery before and after a wildfire, in areas impacted by fire and in unimpacted areas (i.e., a BACI design). Paths were then counted, categorized by size, and assessed to see whether paths were persistent or were formed or lost dynamically over time. This was a pandemic project.