This Readme.txt file was generated on 20161117 by Althea A. ArchMiller ------------------- GENERAL INFORMATION ------------------- 1. R Code and Output Supporting: Used-habitat calibration plots: A new procedure for validating species distribution, resource selection, and step-selection models. 2. Author Information Principal Investigator Contact Information Name: John R. Fieberg Institution: Department of Fisheries, Wildlife, and Conservation Biology Address: University of Minnesota-Twin Cities Email: jfieberg [at] umn.edu Co-investigator Contact Information Name: James D. Forester Institution: Department of Fisheries, Wildlife, and Conservation Biology Address: University of Minnesota-Twin Cities Email: jdforest [at] umn.edu Co-investigator Contact Information Name: Garrett M. Street Institution: Department of Wildlife, Fisheries, and Aquaculture Address: Mississippi State University Email: gms246 [at] msstate.edu Co-investigator Contact Information Name: Douglas H. Johnson Institution: US Geological Survey Address: Northern Prairie Wildlife Research Center Email: douglas_h_johnson [at] usgs.gov Co-investigator Contact Information Name: Althea A. ArchMiller Institution: Department of Fisheries, Wildlife, and Conservation Biology Address: University of Minnesota-Twin Cities Email: althea.archmiller [at] gmail.com Co-investigator Contact Information Name: Jason Matthiopoulos Institution: Institute of Biodiversity, Animal Health and Comparative Medicine Address: University of Glasgow Email: Jason.Matthiopoulos [at] glasgow.ac.uk 3. Date of data collection (single date, range, approximate date) 2013 and 2014 4. Geographic location of data collection (where was data collected?): Minnesota 5. Information about funding sources that supported the collection of the data: This work was funded by the University of Minnesota-Twin Cities and the Minnesota Environment and Natural Resources Trust Fund. The Minnesota Department of Natural Resources assisted with collaring and monitoring of the moose. -------------------------- SHARING/ACCESS INFORMATION -------------------------- 1. Licenses/restrictions placed on the data: Attribution-NonCommercial-ShareAlike 3.0 United States 2. Links to publications that cite or use the data: Fieberg, Forester, Street, Johnson, ArchMiller, and Matthiopoulos. Used-habitat calibration plots: A new procedure for validating species distribution, resource selection, and step-selection models. Ecography. DOI:10.1111/ecog.03123. http://doi.org/10.1111/ecog.03123 3. Links to other publicly accessible locations of the data: https://github.com/aaarchmiller/uhcplots 4. Links/relationships to ancillary data sets: Includes a subset of the data used in Street et al. 2016. Habitat functional response mitigates reduced foraging opportunity: implications for animal fitness and space use. Landscape Ecology. doi: 10.1007/s10980-016-0372-z 5. Was data derived from another source? Street et al. 2016. Habitat functional response mitigates reduced foraging opportunity: implications for animal fitness and space use. Landscape Ecology. doi: 10.1007/s10980-016-0372-z 6. Recommended citation for the data: Fieberg, Forester, Street, Johnson, ArchMiller, and Matthiopoulos. R Code and Output Supporting: Used-habitat calibration plots: A new procedure for validating species distribution, resource selection, and step-selection models. Retrieved from the Data Repository for the University of Minnesota, http://doi.org/10.13020/D6T590. --------------------- DATA & FILE OVERVIEW --------------------- 1. File List A. moose12687.csv: Radio telemetry data for one Minnesota moose (ID #12687) from 2013 and 2014. Data was sourced from Street et al. (2016). B. example_1_corr00.html: Complete R code (R Core Team 2015) and associated output for the missing predictor simulation example in the manuscript with correlation(x1,x2) = 0,0 for the training and test data, respectively. Includes code to simulate the datasets, fit generalized linear models (GLM), create calibration plots following Boyce et al. (2002) and Johnson et al. (2006), and create the Used-Habitat Calibration (UHC) plots in the paper (with simulation envelope for f^U(z)) and also an additional plot with simulation envelope for f^U(z) - hat{f}^u(z)i. C. example_1_corrNN.html: Complete R code (R Core Team 2015) and associated output for the missing predictor simulation example in the manuscript with correlation(x1,x2) = -0.3,-0.3 for the training and test data, respectively. Includes code to simulate the datasets, fit GLMs, create calibration plots following Boyce et al. (2002) and Johnson et al. (2006), and create the Used-Habitat Calibration (UHC) plots in the paper (with simulation envelope for f^U(z)) and also an additional plot with simulation envelope for f^U(z) - hat{f}^u(z)i. D. example_1_corrPN.html: Complete R code (R Core Team 2015) and associated output for the missing predictor simulations example in the manuscript with correlation(x1,x2) = 0.3,-0.3 for the training and test data, respectively. Includes code to simulate the datasets, fit GLMs, create calibration plots following Boyce et al. (2002) and Johnson et al. (2006), and create the Used-Habitat Calibration (UHC) plots in the paper (with simulation envelope for f^U(z)) and also an additional plot with simulation envelope for f^U(z) - hat{f}^u(z)i. E. example_2_temp.html: Complete R code (R Core Team 2015) and associated output for the non-linear simulation example in the manuscript with temperature and temperature^2. This program includes code to simulate the datasets, fit GLMs, create calibration plots following Boyce et al. (2002) and Johnson et al. (2006), and create the Used-Habitat Calibration (UHC) plots in the paper (with simulation envelope for f^U(z)) and also an additional plot with simulation envelope for f^U(z) - hat{f}^u(z)i. F. example_3_spatial_split.html: Complete R code (R Core Team 2015) for the spatial example in the manuscript using simulated data with spatial coordinates. This program includes code to simulate the datasets, and create the Used-Habitat Calibration (UHC) plots in the paper. G. example_4_moose.html: Complete R code (R Core Team 2015) for the Step-Selection Function example in the manuscript using moose data from Street et al. (2016). This program includes code to run the SSF and create the Used-Habitat Calibration (UHC) plots in the paper (with simulation envelope for f^U(z)) and also an additional plot with simulation envelope for f^U(z) - hat{f}^u(z)i. H. ecalcrsf.html Complete R code (R Core Team 2015) for the function “ecalcrsf” that creates calibration plots following Boyce et al. (2002) and Johnson et al. (2006). I. tables.html Complete Program R code (R Core Team 2015) to create the tables in the manuscript. J. uhcplots_0.1.0.tar.gz Bundled package for R package uhcplots, with the version used to create the examples used in the manuscript. The package has been compressed into a single file (.tar) and compressed using gzip (.gz). This package is also available in its current form at https://github.com/aaarchmiller/uhcplots; however, the current form may not be backwards compatible with the code in the html examples, which is why this bundled package is included here. K. UHCPlotsPaper_R.zip Complete Program R files (.R extension) for each of the above .html files. 2. Relationship between files: Each example.html file (items B through G) uses various uhcplot functions (available on GitHub https://github.com/aaarchmiller/uhcplots or archived in item J). The example with the MN moose data also uses the dataset (item A). The tables (item I) are created with regression output files created and saved with the codes provided in the example.html files (items B through G). The zipped folder (item K) contains all of the Program R files (.R extension) for each of the html files. 3. Additional related data collected that was not included in the current data package: NA 4. Are there multiple versions of the dataset? The final uhcplots package are available for download and use on GitHub (https://github.com/aaarchmiller/uhcplots). -------------------------- METHODOLOGICAL INFORMATION -------------------------- 1. Description of methods used for collection/generation of data: For complete methodological details, please refer to Fieberg et al. (In Review) and Street et al. (2016). a. Missing predictor simulation examples: Using simulations, we considered a species whose distribution is driven by elevation (x1) and precipitation (x2). Both x1 and x2 are normally distributed with MVN(0, Sigma). We considered 3 different data generating scenarios in which we set var(x1) = var(x2) = 4 but varied the correlation between x1 and x2. In the first example “corr00,” we set cor(x1,x2) = 0 for both training and test data. In the second example “corrNN,” we set cor(x1,x2) = -0.3 for both training and test data. In the final example “corrPN,” we set cor(x1,x2) = 0.3, -0.3 for training, test data. b. Non-linear simulation example: Using simulations, we considered a species whose distribution is driven by a quadratic relationship with temperature (x3). We considered centered values of x3 distributed as N(0,4) and the probability of selecting locations was proportional to exp(2[x3]-[x3]^2). c. Missing predictor simulation example with spatial coordinate: Using simulations, we considered a species whose distribution is driven by elevation (x1) and precipitation (x2). Both x1 and x2 are normally distributed with MVN(0, Sigma). We simulated uniformly distributed x and y spatial coordinates for the presence and background locations associated with two landscapes (a test and a training landscape), allowing the correlation among (x, y) spatial coordinates and the habitat predictors (x1, x2) to differ between the two landscapes. We again fit two models to data collected from the training landscape: the first included only elevation and the second included elevation and precipitation (the correct model). We then evaluated how well these models predicted the spatial distribution of presence points in the test landscape by creating UHC plots for the (x, y) spatial coordinates. d. Step-Selection Function with MN moose data example: Full details about data collection for the MN moose data set are in Street et al. (2016). Briefly, these data were collected from a moose fit with an Iridium GPS radio collar (VECTRONIC Aerospace GmbH, Berlin, Germany) from summer 2013 and summer 2014. The data were spatially joined with land cover data from 2011 NLCD data (Jin et al. 2013), including deciduous forest (decid50), mixed wood forest (mixed50), coniferous forest (conif50), and treed wetlands (treedwet50). 2. Methods for processing the data: Simulated data were created with the R code available in each of the example html files above using the “uhcdatasimulator” function. The MN moose data was processed by generating 10 available locations for each used location by randomly selecting 10 step lengths and turn angles to project the animal forward in time from the previous location. We defined resource availability at the used and available locations as the proportional cover of decid50, mixed50, conif50, and treedwet50. 3. Instrument- or software-specific information needed to interpret the data: Programs were written for Program R (R Core Team 2015) and full session information, including packages used, are provided at the bottom of each example.html file. 4. Standards and calibration information, if appropriate: NA 5. Environmental/experimental conditions: NA 6. Describe any quality-assurance procedures performed on the data: All programs and data have been extensively reviewed for quality assurance and control. 7. People involved with sample collection, processing, analysis and/or submission: John Fieberg, James Forester, Garrett Street, Althea ArchMiller ----------------------------------------- DATA-SPECIFIC INFORMATION FOR: moose12687.csv ----------------------------------------- 1. Number of variables: 13 2. Number of cases/rows: 15158 3. Missing data codes: NA 4. Variable List A. CollarID The unique identifier for the GPS collar. B. datetime The timestamp for each record C. stratum The stratum identifier. One stratum is included for each observed location (presence==1) and its associated available locations generated by the random movement paths (presence==0). D. X X coordinate (meters; NAD1983 UTM Zone 15N) E. Y Y coordinate (meters; NAD1983 UTM Zone 15N) F. step Step length (m) from previous location G. bearing Bearing relative to true north H. presence Binary indicator for each observed location (presence==1) and its associated available locations generated by the random movement paths (presence==0). 0 = Randomly generated available location 1 = Observed location I. decid50 Percent cover of deciduous forest from 2011 NLCD (Jin et al. 2013). J. mixed50 Percent cover of mixed wood forest from 2011 NLCD (Jin et al. 2013). K. conif50 Percent cover of coniferous forest from 2011 NLCD (Jin et al. 2013). L. treedwet50 Percent cover of treed wetland from 2011 NLCD (Jin et al. 2013). M. year Year for each data point ----------------------------------------- References Boyce, MS, Vernier PR, Nielsen SE, and Schmiegelow, FK (2002) Evaluating resource selection functions. Ecological Modeling 157, 281-300. Jin, S, Yang L, Danielson, P, Homer C, Fry J, and Xian G (2013) A comprehensive change detection method for updating the national land cover database to circa 2011. Remote Sensing of Environment 132, 159-175. Johnson, CJ, Nielsen, SE, Merrill, EH, McDonald, TL, and Boyce MS (2006) Resource selection functions based on use-availability data: Theoretical motivation and evaluation methods. Journal of Wildlife Management 70, 347-357. R Core Team. (2015). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Street, GM, Fieberg, J, Rodgers, AR, Carstensen, M, Moen, R, and Moore, SA, et al. (2016) Habitat functional response mitigates reduced foraging opportunity: Implications for animal fitness and space use. Landscape Ecology 1-15. -----------------------------------------