- The research question that you asked (provide one question for each exercise).
Exercise 1:How much area does a ringtail territory occupy, is this consistent between individuals and sexes?
Exercise 2: What environmental variables influence ringtail distribution?
Exercise 3: What forested habitat types are ringtail selecting or avoiding, and do these relationships change at different scales?
Final Project: What does predicted ringtail distribution look like across the Applegate Wildlife Management Unit, and what factors are associated with ringtail presence?
- A description of the dataset you examined, with spatial and temporal resolution and extent.
I captured ringtail from October 2020 to May 2022 in the Applegate Wildlife Management Unit in southwestern Oregon (Figure 1). I deployed both GPS and VHF collars and retrieved 1,462 gps locations from 16 individuals. Collars were scheduled to record locations 3 times per night during foraging hours, and lasted approximately 3.5-5 months per deployment. My GPS locations are clustered on the Eastern portion of the Applegate Wildlife Management Unit (Figure 2).
In conjunction with my GPS data I used remotely derived environmental data including elevation, slope, aspect, and canopy cover at 30m resolution. I used NLCD landcover data types to create buffered habitat layers at the 0.1km and 0.5km scale.
- Hypotheses: predictions of patterns and processes you looked for.
I was particularly interested in mapping my ringtail home ranges to derive estimates for space use, territorial overlap, and utilization distributions. I expected home range sizes to vary by sex, minimal territorial overlap among males or females, and clustered utilization distributions.
I was also interested in habitat associations for ringtail within my study area and the Applegate WMU. Previous data from a California species distribution modelling suggests that ringtail presence is positively influenced by presence of hardwoods, larger hardwoods, canopy closure, steeper slopes, and best modelled at a coarse scale 10km2 (Campbell 2004). These factors likely influence the presence of ringtail in the state of Oregon, but available habitats differ from those available in California.
I expected a positive relationship between ringtail presence and slope, canopy cover, and hardwoods at all ages and scales. I expected a negative relationship between presence and habitats with old growth characteristics, primarily because these are often found at high elevations and do not have a hardwood component. I expected a polynomial relationship between presence and elevation because ringtail are a mid-elevation species (table 1).
- Approaches: analysis approaches you used.
I used program R and the packages adehabitatHR, sp, rgdal, and raster to estimate utilization distributions and create home range polygons. Within the package adehaitatHR, I used the functions KernelUD, getverticeshr, and getvolumeUD to calculate utilization distributions as well as area estimates for each individual.
For habitat associations I used two methods, logistic regression modelling using pseudo absence data, and maximum entropy (MaxEnt) modelling using presence only data. All modelling was done within Program R using the packages raster, reshape2, dismo, maxnet, glmnet, MuMin, presenceAbsence, and ecospat.
- Results: what did you produce — maps? statistical relationships? other? Present the key, important results you created.
For my first exercise I produced home range estimates (Table 1), home range polygons (Figure 3), and utilization distributions for individual ringtail (Figure 4). I was surprised to learn that home range sizes were highly variable between individuals and within sexes (mean 445 ha; range 58-795 ha). I created utilization distributions for each individual ringtail, most were unimodal and a few individuals had bimodal distributions.
For mapping ringtail distribution I produced probability maps using logistic regression modelling and Maxent models in the little Applegate valley where I collected my gps locations (Figure 5). I further expanded the extent of my Maxent model to include the entire Applegate WMU (Figure 6). Using my final model for the Applegate WMU, I produced variable response plots (Figure 6). All habitat types in my final model were buffered to the 0.5km scale.
Table 1. Home range size estimates for 16 ringtail using kernel polygon methods (*Female)
| Individual ID | Home range estimates (ha) | |
| R02 | 546.99 | |
| R05 | 615.08 | |
| R08 | 201.7 | |
| R09* | 641.04 | |
| R10 | 472.95 | |
| R13 | 353.89 | |
| R14 | 232.99 | |
| R15 | 795.26 | |
| R17 | 312.68 | |
| R18 | 310.91 | |
| R19 | 621.33 | |
| R20 | 359.04 | |
| R22 | 430.49 | |
| R23 | 628.86 | |
| R24* | 58.301 | |
| R25 | 539.92 |
- What did you learn from each of the analyses you conducted (i.e., from each exercise)?
Exercise 1: Home range size was highly variable between individuals and within sexes (mean 445 ha; range 58-795 ha). The smallest home range (R24; 58 ha) belongs to a female, but so does one of the larger home ranges (R09; 641 ha). Ringtail do not use their entire home ranges equally and have unimodal or bimodal utilization distributions.
Exercise 2: Steeper slopes, SW aspect, increased canopy cover, and mid-elevations are all important variables when looking at ringtail distribution. Habitat types also influenced ringtail distribution, with an avoidance of grasslands, shrub covered sloped, and old growth stands. Only one habitat type had a significant positive relationship Mixed conifer (white/douglas fir) aged 31-80 years (p-value 0.0002)
Exercise 3: Scale matters when conducting modeling exercises. When habitat variables were buffered to 0.5km, their significance changed. For example at scales of 30-100m hardwoods were negatively associated with ringtail presence, but when buffered to 0.5km scale, hardwoods were positively associated with ringtail presence. I think this result is likely due to the nature of my GPS locations. They were collected during active foraging times and represent a particular behavior. Ringtail are known to use hardwoods as diurnal resting locations, and ringtail can travel >500m in a single evening. It is possible that foraging quality is reduced in hardwood habitats, or prey is not present in sufficient quantities at certain times of the year (ie winter/early spring).
- Significance. How are these results important to science? to resource managers?
Knowing how ringtail use their surrounding environments, including the total space needed to support an individual, and the type and quality of habitat they require all add to our understanding of ringtail ecology. This information can help managers make informed decisions regarding proposed land use changes and their impacts to the species, how populations may react to climate change, and making informed decisions regarding species conservation status.
- Software learning. Your learning: what did you learn about software (a) Arc-Info, (b) GIS programming in Python, (c) programming in R, (d) Modelbuilder in Arc,or (e) other?
I learned many new techniques for manipulating spatial data with program R. I am still more familiar with visualizing data using GIS, but the reproducibility of R code makes it an excellent tool, particularly for modelling.
- Statistics learning. What did you learn about statistics?
I learned there are many ways to quantify the relationships between spatial data, and those methods can be easy to perform using programs such as R and ArcGIS. I found hotspot methods to be very useful for animal movement data, and regression techniques. I used Kernel density methods, utilization distributions (similar to hotspot), and regression methods for my analyses. Hotspot methods are useful for identifying the location and intensity of clustering within a dataset. Spatial autocorrelation can describe how your variable relates to itself, and cross-correlation can describe relationships between two variables. Regression is what I am currently most familiar with, and can be useful for describing how your response variable is influenced by explanatory variables.
Evolving question. How did the results of each analysis lead you to change/refine your question? Write out the original question you stated at the beginning of the class, and restate the question(s) you now plan to address.
Original Question: How is the distribution of ringtail related to the quality of available habitat used via the amount of food, water, and resting structures available?
Exercise 1:How much area does a ringtail territory occupy, is this consistent between individuals and sexes?
Exercise 2: What environmental variables influence ringtail distribution?
Exercise 3: What forested habitat types are ringtail selecting or avoiding, and do these relationships change at different scales?
Final Project: What does predicted ringtail distribution look like across the Applegate Wildlife Management Unit?
Future techniques. What techniques would you like to explore to answer your research questions in the future?
I want to explore my final model set for the Applegate WMU in greater detail, using step-wise selection and AIC and/or AUC to select the best model for my data.
I want to expand my dataset to include historical ringtail location data, and locations collected at diurnal rest site locations. Using rest site and foraging locations will improve the quality of my modelling efforts. I want to expand the extent of my current model to southwestern Oregon, which encompasses the known range of ringtail within the state. At this larger extent I want to try modelling variables at scales up to 10km2, as suggested in Campbell 2004.
Literature Cited
Campbell, L. A. 2004. Distribution and habitat associations of mammalian carnivores in the central and southern Sierra Nevada. Dissertation, University of California Davis, Sacramento, California, USA.
