Data inputs and results for "Mammal niches are not conserved over continental scales" by Goldstein et al.

This data packet provides inputs and results for "Mammal niches are not conserved over continental scales" by Goldstein et al., currently in the submission process. This repository will eventually be updated to link to the published manuscript.

==============================================================================================================================================================Overview==============================================================================================================================================================

Data and model products associated with the manuscript "Mammal niches are not conserved over continental scales" by Goldstein et al. 

Files are organized into two subdirectories. The first, "model_inputs/", contains 8 data files intended to be used as part of the reproducible code repository at  https://github.com/dochvam/Mammal_SVCs_ISDM_reproducible. The second subdirectory, "model_outputs/", contains modeled products givingestimated spatially varying niche relationships and predictions of relativeabundance.

Below, we describe the contents of each file type. See the main manuscript for full methodology, data sources, and discussions of spatial scales.

NOTE: Version 1 of this dataset contained some errors that have been correctedin Version 2. Version 2 was used as the input dataset for the analyses in theassociated manuscript. Version 1 should not be used.

==============================================================================================================================================================Subdirectory 1: "model_inputs/"==============================================================================================================================================================

Two versions of each of four files are provided, corresponding to analyses that do or do not consider ancient genetic lineages as potential sources of spatial nonstationarity in mammal niches. Each file type is formatted the same,and the versions are differentiated by either the suffix "nolineage" or "lineage" in the filename. 

===============================================================================File 1: gridcell_covars_lineage.csv and gridcell_covars_nolineage.csv===============================================================================These files are .csvs giving spatial covariate data for each scale 2 cellin North America, summarized to 5000 m. All percentage values are given in 10ths of a percent (scale of 0-1000). The following columns are provided:

- grid_cell: Scale 2 cell ID- Arable: pct arable land (Jung et al. 2020)- EVI_mean: mean enhanced vegetation index (Didan 2021)- EVI_Q95: 95th quantile of EVI (Didan 2021)- Forest: Pct forest cover (Jung et al. 2020)- Grassland: pct grassland (Jung et al. 2020)- Pastureland: pct pastureland (Jung et al. 2020)- Pop_den: Human population density, from Gridded Population of the World (CIESIN 2018)- Precipitation: avg annual precip. (Vega et al. 2017)- Shrubland: pct shrubland (Jung et al. 2020)- Temp_max: Average maximum daily temperature (Vega et al. 2017)- Terrain_roughness (Amatulli et al. 2018)- Wetlands: pct wetlands (Jung et al. 2020)- is_land: Whether or not the cell is on land vs. ocean, used for filtering- Agriculture: Pct. agricultural land (Jung et al. 2020)- Pop_den_sqrt: Square root of human population density (CIESIN 2018)- EVI_variability: Distance btw the 95% inner quantiles of EVI (Didan 2021)

===============================================================================File 2: inat_cts_lineage.csv and inat_cts_nolineage.csv===============================================================================

These files give summaries of iNaturalist sampling effort and detectionsfor target species. The following columns are provided:

- grid_cell: Scale 3 cell ID- n: Total iNaturalist effort in the cell (number of obs. of all mammals)- The remaining columns are named for species. Each column gives the count    of observations of the species in the cell.

===============================================================================File 3: ct_datlist_lineage.RDS and ct_datlist_nolineage.RDS===============================================================================

The ct_datlist files contain R objects that are lists of lists. These objects ultimatelycontain all of the camera detection histories and camera-level covariate data usedin modeling. We use the nice data type "unmarkedFrameOccu" from the unmarked R packageto organize these detection data.

Each outer list is of length equal to the number of species. The ith element of eachlist contains the following named slots:

- species: a string giving the name of the ith species- umf: an unmarkedFrameOccu object. This object has three important slots:    - y: a (# deployments) x (max # replicates) matrix giving 1s, 0s, or NAs indicating         whether the target species was observed in that 10-day window;    - siteCovs: a (# deployments) x 2 data frame with the following columns:        - site_ID: A unique ID of the exact location, shared by deployments with the same           coordinates        - subproject_ID: A unique ID indicating which camera array is associated                with this deployment    - obsCovs: a (# deployments * max # replicates) x 6 data frame with the following columns:            - year: the year of survey, relative to 2020 (zero-year is 2020)            - yday_scaled: the (scaled) Julian date of the beginning of the window            - yday_scaled_sq: yday_scaled^2, for use in estimating a quadratic effect            - log_roaddist_scaled: Scaled distance to nearest road (Meijer et al. 2018)            - Canopy_height_scaled: Scaled canopy height (Potapov et al. 2021)            - obs_len_scaled: Scaled duration of window, to account for some windows                               being cut off at < 10 days- coords: a data frame. Originally, this file gave the exact position for each camera,      but these exact locations have been scrubbed for privacy. See the original sources          cited in the manuscript for full details. This data frame contains the following column:    - scale2_grid_ID: the ID of the Scale-2 5000 m grid cell containing the camera

===============================================================================File 4: grid_translator_wspecs_nolineage.csv and grid_translator_wspecs_lineage.csv===============================================================================

These files are used for bookkeeping to track the relationships between the three spatial scales in this study. Each row corresponds to a single "scale 2"cell, giving the ID of the corresponding S3 and S4 grid and also an ID for eachspecies indicating whether and where it is in the species' range. 

Note that the scale names in the code don't match the manuscript. In the code,"scale 1" is the level of an individual camera, "scale 2" is the 5 km intensitygrid, "scale 3" is the 50 km iNaturalist grid, and "scale 4" is the 100 kmSVC grid.

The following columns are provided:- scale2_grid_ID: unique ID for each cell in the 5 km intensity grid- scale3_grid_ID: unique ID for each cell in the 50 km iNaturalist aggregation- scale4_grid_ID: unique ID for each cell in the 100 km SVC grid- GRID_ID_[species]: for each species, a column is provided on the S4 scale                     counting each cell in the species' modeled range. NAs                 indicate that the S2 cell defined in the row is not             included in the species' modeled range.

==============================================================================================================================================================Subdirectory 2: "model_outputs/"==============================================================================================================================================================

===============================================================================File 1: svc_estimates.csv===============================================================================

This file gives an estimate of the effect of each covariate on each species'intensity, and the uncertainty in that estimate, for each species/covariatepair. Results correspond to lineage models for species with phylogeographiesand non-lineage species otherwise. Each row represents the effect of one covariate on one species' relative intensity process within one 100 km cell g. Note that many estimates of beta_g are uncertain even for strong spatial effects---the model is often confident that a spatial process is supported while estimates of the realized process are uncertain.

The following columns are provided:- x: the x-coordinate of the 100 km cell- y: the y-coordinate of the 100 km cell- species- parname: the name of the covariate- mean: the mean of the posterior samples of beta_g- 2.5%: the 2.5th quantile of the posterior samples of beta_g- 50%: the 50th quantile of the posterior samples of beta_g- 97.5%: the 97.5th quantile of the posterior samples of beta_g

The following spatial projection is used to define X/Y coordinates:"+proj=aea +lat_1=20 +lat_2=60 +lat_0=40 +lon_0=-96 +x_0=0 +y_0=0 +ellps=GRS80 +datum=NAD83"

===============================================================================File 2: predicted_intensity.tif===============================================================================

This file contains a raster "brick" giving the predicted intensity surface and uncertainty in this surface for each species. All predictions are generatedusing models that do *not* account for lineage information---this means that predictions for species with lineages are not from the models reported in themain manuscript. The reason for this is that we found that lineages were overall unsupported, so better predictions can be arrived at by excluding thissource of uncertainty in the underlying intensity process.

The raster brick has 66 layers. Each layer provides either the mean predictedlog intensity in each grid cell across the species range or else providesthe standard error of that predicted log intensity. Layer names indicate outputtype and species associated with each layer.

==============================================================================================================================================================References==============================================================================================================================================================

Camera data are obtained from the following sources, which can be consulted toobtain the original raw camera data

- Cove, Michael V., et al. "SNAPSHOT USA 2019: a coordinated national camera trap survey of the United States." (2021): e03353.- Kays, Roland, et al. "SNAPSHOT USA 2020: A second coordinated national camera trap survey of the United States during the COVID‐19 pandemic." (2022): e3775.- Shamon, H., et al. “SNAPSHOT USA 2021: A third coordinated national camera trap survey of the United States.” Ecology, 105.6 (2024): e4318.- Rooney, B., et al. “SNAPSHOT USA 2019–2023: The first five years of data from a coordinated camera trap survey of the United States.” In Press (2024).- Kays, Roland, et al. "Does hunting or hiking affect wildlife communities in protected areas?." Journal of Applied Ecology 54.1 (2017): 242-252.- Roberts, R. California Department of Fish and Wildlife, Bobcat Program Initiative. wildlifeinsights.org (2023).- Lasky, Monica, et al. "CAROLINA CRITTERS: a collection of camera trap data from wildlife surveys across North Carolina." Ecology 102.7 (2021): e03372.- Forrester, T. (2000). Urban to Wild Project. http://n2t.net/ark:/63614/w12004302. Accessed via wildlifeinsights.org on 2024-08-29.- McMurry, S. et al. In review (2024).- Forrester, T. (2011) Okaloosa S.C.I.E.N.C.E. Project. http://n2t.net/ark:/63614/w12004287. - Myers, J. (2014) Tyson Research Center ForestGEO Project. http://n2t.net/ark:/63614/w12004295.- McMurry, S., and Kays, R.(2023). Calloway Forest Preserve. http://n2t.net/ark:/63614/w12006449. Accessed via Wildlife Insights on 2024-08-29.- McMurry, S., Parsons, A., Lasky, M., Luongo, K.,  Clark, J., McShea, W., Scher, L., Kays, R., Spurlin, J., Martin, G., Frech, G., Barajas-Salazar, K.,  Snider, M. (2022). Last updated October 2023. Calloway Forest Preserve. http://n2t.net/ark:/63614/w12004251. Accessed via wildlifeinsights.org on 2024-08-29.- Kays, R.. (2008). Last updated March 2024. Albany Area Camera Trapping Project. http://n2t.net/ark:/63614/w12003860. Accessed via wildlifeinsights.org on 2024-08-29.- Kays, R., Snider, M., McMurry, S., Alyetama, M. (2024). Last updated April 2024. Pilot Mountain Density 2024. http://n2t.net/ark:/63614/w12007160. Accessed via wildlifeinsights.org on 2024-08-29.- Malleshappa, V., Smithsonian, E., Kays, R., Schuttler, S. (2015). Last updated December 2022. Museums Connect Mexico. http://n2t.net/ark:/63614/w12004298. Accessed via wildlifeinsights.org on 2024-08-29.

Covariate data are obtained from the following sources:- Vega, G. C., Pertierra, L. R. & Olalla-Tárraga, M. Á. MERRAclim, a high-resolution global dataset of remotely sensed bioclimatic variables for ecological modelling. Sci. Data 4, 170078 (2017).- Jung, M. et al. A global map of terrestrial habitat types. Sci. Data 7, 256 (2020).- Amatulli, G. et al. A suite of global, cross-scale topographic variables for environmental and biodiversity modeling. Sci. Data 5, 180040 (2018).- Center For International Earth Science Information Network-CIESIN-Columbia University. Documentation for the Gridded Population of the World, Version 4 (GPWv4), Revision 11 Data Sets. (2018) doi:10.7927/H45Q4T5F.- Didan, K. MODIS/Terra Vegetation Indices 16-Day L3 Global 1km SIN Grid V061. NASA EOSDIS Land Processes Distributed Active Archive Center https://doi.org/10.5067/MODIS/MOD13A2.061 (2021).- Meijer, J. R., Huijbregts, M. A. J., Schotten, K. C. G. J. & Schipper, A. M. Global patterns of current and future road infrastructure. Environ. Res. Lett. 13, 064006 (2018).- Potapov, P. et al. Mapping global forest canopy height through integration of GEDI and Landsat data. Remote Sens. Environ. 253, 112165 (2021).- Jensen, A. J. et al. Geographic barriers but not life history traits shape the phylogeography of North American mammals. Glob. Ecol. Biogeogr. e13875 (2024).

iNaturalist data are obtained from inaturalist.org via the data exporter (see manuscript for details).