Fig 1.
The study site is within a Suisun Bay marsh located in the eastern portion of the San Francisco Bay-Delta estuary, CA.
Fig 2.
Three indicators inspected to determine the proper buffer distance to implement as well as to determine the correct detection function.
A) Salt marsh harvest mouse frequency of distances moved in meters for the entire 18 year full 100 trap dataset; B) Detection probability curves for male and female SMHM related to distance (red line indicates buffer distance used); C) Effective sampling area as a function of increasing buffer width.
Fig 3.
Crescent Unit trapping grid variations for subsampling reduced trapping designs.
From top left to bottom right: Original field data collection layout of 10x10 grid, 9x9 grid, 8x8 grid, 7x7 grid, 6x6 grid, and a half density grid with 50 traps. Photos of small mammal trapping (photo credits: L. Thompson, I. Woo, and W. Thein).
Fig 4.
Absolute difference and absolute percent difference for both density estimates and standard error (SE) across years for the full dataset compared to the subsampled datasets.
(lower black line indicates the minimum SE observed [2017], red line indicates the mean SE, and upper black line indicates the maximum SE observed [2005]).
Table 1.
Salt marsh harvest mouse density estimates (mice/ha) and capture numbers [D(capture)] for the full dataset and subsampled datasets between 2000 and 2017.
Fig 5.
Density estimate, relative bias, and root mean square error for Monte Carlo simulations (n = 100) using the mean density, detection function, and home range as the true values for comparing the number of trapping occasions (3, 4), and the number of traps (100, 81, 64, 49).
Table 2.
Difference of density estimates for the subsampled datasets compared to the full dataset, showing mean, absolute mean, and absolute max summaries.
Fig 6.
Density estimate, relative bias, and root mean square error for Monte Carlo simulations (n = 100) using the mean density, detection funtion, and home range as the true values for comparing the number of trapping occasions (3, 4), and trap spacing (10, 15, 20 meters).
Table 3.
Relative percent time saved with each subsampling scenario compared to the full grid design.
Percent time estimates included pre-survey setup and logistics, time to conduct the surveys, and post-survey trap cleaning and data entry.