Skip to main content
Dryad

Nimble code and dataset for: Estimating true density in large, alpine herbivores using Google Earth imagery

Data files

Mar 02, 2023 version files 46.84 KB

Abstract

This nimble code will estimate elk density from the count data (DO_data_list.RData) and covariates (DO_constants_list.RData).  Data and covariates include: a 'y' matrix of double observer counts with 3315 rows (1 for each 250 m square plot) and 3 columns for the observer 1's exclusive counts, observer 2's exclusive counts, and the count of elk detected by both observers. The double-observe protocol was only employed in 372 random cells. 'n' is a vector of total counts – the total number of uniquely detected elk in each plot.  'obs1tot' is a vector of the total count of elk by just observer 1, who counted elk in all 3315 cells. There are two dummy indicator variables that are all 1's ('obs1constraint' and 'obs2constraint') to constrain unobserved but estimated counts to sum to expected totals. For example, observer 1 was the only observer in 2943 plots. This single count would be the equivalent of the sum of the exclusive observer 1 counts and the joint observer 1 and 2 counts in a double-observer protocol. Additionally, the sum of observer 2s exclusive counts and observer 1's total counts must equal the total count of unique individuals in a cell. The covariates 'meadow' and 'deadfall' are the additive log-ratio transformations of the proportion of open meadow and burned forest, respectively, in each 250-m plot. The 'nonforest' covariate is the number of hectares in each plot that was not conifer forest.  'M' is the total number of plots. 'firstsearch' is an indicator variable, indicating which plots were part of the first search area in the alpine >2750 m above sea level. 'domethod' is an indicator variable, indicating which plots were searched using the double-observer protocol.