WGS Field Sampling Plan 11.4.2021.pdf

WGS Field Sampling Plan– November 2021 Page 1

WGS Field Sampling Plan

Multi-Purpose Machine Gun Range Naval Weapons Systems Training Facility Boardman Oregon

November 2021

Prepared for:

U.S. Department of Navy

Prepared by:

Kern Statistical Services

Native Restoration Consulting

WGS Field Sampling Plan– November 2021 Page 2

Contents

Contents

Acronyms

Background

Objectives

Approach

Results from 2014 and 2015

Power Analysis

Transect and Camera Trap Sample Sizes

Transect Layout

Data Collection

BE Counts

Camera Trap Deployment

Sampling Period

Data Analysis

WGS BE Occupancy Rates and Densities

BE Densities Calculations

Statistical Power Analysis

Power Simulation Procedure

BACI Design

Summary

References Cited

Acronyms

AVD Survey – Audio-visual Survey

BACI - Before After Control Impact

BE Survey – Burrow Entrance Survey

FEIS – Final Environmental Impact Statement

MPMGR – Multipurpose Machine Gun Range

PTA – Potential Training Area

WGS BE – Washington ground squirrel burrow entrance

WGS Field Sampling Plan– November 2021 Page 3

Background

An Installation-wide comprehensive Washington Ground Squirrel (WGS) monitoring project (US Navy 2021), referred to as the “2014 - 2015 WGS Monitoring Project” in this report, was conducted at Naval Weapons Systems Training Facility Boardman (NWSTF Boardman or Installation) in 2014 and 2015 in support of range enhancements addressed in the NSWTF Boardman Final Environmental Impact Statement (FEIS, US Navy 2015). Because the WGS was a candidate species under the Endangered Species Act at the time, the Navy conferenced with the US Fish and Wildlife Service and the Service issued the Conference Opinion for Military Readiness Activities at the NWSTF Boardman (01EOFW00-2013-FC-0017) (USFWS 2013). The Conference Opinion established monitoring and management requirements for the WGS associated with actions covered in the FEIS. The intent of the 2014 – 2015 WGS Monitoring Project was to establish population indices that would serve as baseline data for a Before and After Control Impacts (BACI) study to assess the validity of WGS impacts assumptions made in the FEIS in compliance with the Conference Opinion.

A major finding from the 2014 – 2015 WGS Monitoring Project was that the density of Washington ground squirrel burrow entrances (WGS BEs) on the Installation was significantly lower than had been determined from earlier surveys used in the power analysis that informed the 2014 – 2015 survey design (LimnoTech 2013). As a result, a revised field sampling plan and power analysis are needed to meet the standards for the detection of WGS population changes stipulated in the Conference Opinion. Additionally, the data generated from the 2014 - 2015 WGS Monitoring Project are now six years old and may not accurately reflect current WGS BE densities on the Installation. To maintain compliance with the FEIS and Conference Opinion, new baseline data must be collected prior to the construction of the Multi-Purpose Machine Gun Range (MPMGR), the eastern and western Convoy Live Fire Ranges (CLFRs), and the Demolition Training Range (DTR).

The MPMGR, a 93.6-ha (415-ac) area located in the west central portion of the Installation (Figure 1), is the first of the range enhancements requiring BACI monitoring to be scheduled for implementation. The following sampling plan outlines a methodology for determining WGS indices prior to and after development and use of the MPMGR. Consistent with Section 1.6.4 of the Conference Opinion, surveys will be conducted in a 400-m buffer area around the MPMGR (referred to in this report as MPMGR Buffer). Surveys focus on this 242.5 ha (600 ac) MPMGR Buffer rather than within the MPMGR footprint because WGS impacts were assumed within the footprint and are being mitigated. Conversely, no habitat loss, long-term habitat degradation, or decline in WGS numbers were assumed in the MPMGR Buffer though it will be subject to single-event noise levels less than 140 dBP. Surveys are intended to validate these impacts assumptions.

WGS Field Sampling Plan– November 2021 Page 4

Objectives

The primary objectives of the WGS Field Sampling Plan are to (1) estimate the number of WGS BEs (Washington ground squirrel burrow entrances) within the MPMGR Buffer and reference area (referred to as Central Geographic Stratum) prior to the development and use of the MPMGR, and

2) design the sampling plan to determine whether changes in WGS BE densities in the MPMGR Buffer are attributable to the development and use of the MPMGR or to unrelated causes.

Approach

Two survey methodologies were developed during the 2014 - 2015 WGS Monitoring Project. One survey method, called the AVD (Audio-Visual Detection) Survey, is based on visual and audio detection of WGS by field surveyors. The weakness of this survey is that an understanding of the time that WGS are below ground during the survey period is important to know to accurately report WGS occupancy rates. This would require a separate study, such as a capture-recapture study. For this reason, the BE (Burrow Entrance) Survey methodology was selected for this field sampling plan. The second method is the BE Survey which combines counting burrow entrances along a transect and deploying camera traps at burrow entrances to provide a quantitative estimate WGS BE occupancy rate and density of WGS BEs within the MPMGR Buffer and the Central Geographic Stratum reference area.

The original study design presented in the Washington Ground Squirrel Monitoring Plan (LimnoTech 2013) compared survey results of the 8 Potential Training Areas (PTAs) (Figure 1) with the results of paired reference areas (a 400-m buffer around each PTA). The results of each PTA were also compared to results from surveys taken across the entire Installation and within geographic strata (similar soils and habitat) where each PTA was located. This approach provided a range of scales at which treatment and reference areas were paired and a means to compare treatments of training activities with Installation‐wide and geographic temporal trends.

The purpose of surveying the MPMGR Buffer is should WGS move in or out of the MPMGR Buffer in response to training activities, the movement would be reflected in a density change of WGS BEs in the buffer area. But this cannot be determined without conducting a separate BE Survey within a much larger area – the Central Geographic stratum. If there is no difference between WGS BEs in the MPMGR Buffer and the Central geographic stratum, then we can assume that there has not been a decline or increase in WGS BEs due to training activity in the MPMGR Buffer.

WGS Field Sampling Plan– November 2021 Page 5

Figure 1. Location Maps. The Naval Weapons Systems Training Facility Boardman (NWSTFB) is located several miles south of Boardman OR. The Multi-Purpose Machine Gun Range (MPMGR) is a 93.6 ha area located in the Central Geographic Stratum and one of 8 Potential Training Areas (PTAs) outlined in the FEIS.

NWSTFB

Boardman Conserva�on

Area Arlington

Boardman Washington

Idaho Oregon

Portland

NWSTFB

North

South

Central

Geographic Strata

DZ

MPMGR

CLFR East

UAS

MP East MP West

MP South

CL

FR

W es t

PTAs

MPMGR – Mul�-Purpose Machine Gun Range, CLFRE – Convoy Live Fire Range – East, CLFRW – Convoy Live Fire Range – West, DZ – Drop Zones, MPE – Mortar Pad East, MPW – Mortar Pad West, MPS – Mortar Pad South, UAS – Unmanned Aerial Systems Training Airfield

WGS Field Sampling Plan– November 2021 Page 6

Results from 2014 and 2015

The 2014-2015 WGS Monitoring Project was based on the protocols outlined in the Washington Ground Squirrel Monitoring Plan (LimnoTech 2013). It laid out WGS BE transects at different scales: 1) Installation-wide – which were surveys at the broadest scale, 2) geographic strata scale

– surveys based on similar soils, 3) PTAs – potential trainings areas, and 4) a 400 m buffer around each PTA.

From the 2014-2015 WGS Monitoring Project, we learned that WGS BE/ha densities in the MPMGR and MPMGR Buffer combined were 0.08 in 2014 and 0.13 in 2015. These densities were lower than the Central Stratum (0.25 and 0.23 in 2014 and 2015 respectively) and Installation-wide densities (0.37 and 0.32 in 2014 and 2015 respectively) but these differences were not statistically significant. Based on these densities, the number of WGS BEs occurring in the MPMGR and MPMGR Buffer combined (a 337-ha area) averaged 15 WGS BEs in 2014 and 24 WGS BEs in 2015.

In the 2014-2015 WGS Monitoring Project, BE Surveys were conducted on 40 transects in the MPMGR and 49 transects the MPMGR Buffer. They were surveyed twice, once in 2014 and again in 2015, for a total of 178 BE Surveyed-transects (80 in the MPMGR and 98 the MPMGR Buffer).

The total area surveyed for the MPMGR and MPMGR Buffer was 39.2 ha, or 12% of the 337-ha survey area.

To detect change in WGS BEs based on the very low numbers found in 2014 and 2015 WGS Monitoring Project will require a much greater number of transects and area surveyed as will be presented in this report.

Power Analysis

A power analysis was conducted for the MPMGR Buffer area to determine the number of transects that are needed to detect a meaningful effect in the reduction of WGS BE densities.

The power analysis generates the sample size for the number of transects that will need to be surveyed. It is based on (1) effect size, (2) significance level, and (3) statistical power.

Effect Size. In this sampling plan, the effect size is the difference between WGS BE densities before and after MPMGR development. Two effect sizes were evaluated:

• 67% reduction in WGS BE densities.

• 50% reduction in WGS BE densities.

Significance Level. Significance level is set at 5% providing a 95% confidence that a change in WGS BE densities will be detected.

Statistical Power. Statistical power is the probability that a test correctly rejects the null hypothesis, which is: “there is no decline in WGS BE densities due to range development”.

The higher the statistical power, the lower the probability of making a false negative

WGS Field Sampling Plan– November 2021 Page 7 error, or falsely concluding that there is a change in WGS BE densities due to Navy activity when there is not. Statistical power is only useful when the null hypothesis is rejected. For this survey, statistical powers of 0.8 and 0.7 were evaluated.

Transect and Camera Trap Sample Sizes Transect sample sizes were determined from computer simulations of existing data collected from the 2014 – 2015 Monitoring Project. The simulation varies the following parameters: (1) the assumed WGS BE densities prior to and after development of the MPMGR (effect size) and (2) the number of transects to be surveyed in the MPMGR Buffer and Central Geographic Stratum, which are assumed to be equal for a balanced design (transect sample size). The probability of detecting a specified change in WGS BE occupancy rate is presented as a function of each combination of transect sample size and assumed change in WGS BE densities.

The power analysis is based on deploying a camera trap at every burrow entrance located within the transect strip. The number of camera trap deployments cannot be known in advance because the number of BEs per transect is unknown, but based on the simulation in this plan, it is estimated that for a 0.8 statistical power, 191 camera traps would need to be deployed to detect a 67% reduction in WGS BEs and 331 camera traps to detect a 50% reduction in WGS BEs (Figure 2 and Table 1).

The Central Geographic Stratum will have the same number of transects and camera traps as the MPMGR Buffer depending on the effect size (Table 1). This will double the number of transects and camera traps. For example, at a 67% reduction in WGS BEs at 80% statistical power, the total number of transects is 314 in the MPMGR Buffer and 314 transects in the Central Geographic Stratum, for a total of 628 transects. Camera traps will increase from 191 to 382.

Transect Layout

The simulation curves are based on 400 m long by 4 m wide transect strip dimension. Due to the shape of the MPMGR Buffer, transects may vary in length. The transects are best positioned in a SW to NE direction for most efficient placement in the MPMGR Buffer (Figure 3). The distance that the transect lines are spaced apart depends on the number of transects being surveyed. For example, the spacing between centerlines for the 50% effect size at an 80% statistical power is 11 m, whereas for an effect size of 67% at a 70% statistical power, the distance is 26 m (Table 1).

The same number of transects are evenly spaced in the Central Geographic Stratum to systematically cover the entire geographic area (Figure 4). The distance between transects will be much greater for the Central Geographic Stratum than the distances for the MPMGR Buffer because of the larger size of the Central Geographic Stratum (6,712 ha [16,586 ac]).

WGS Field Sampling Plan– November 2021 Page 8

Figure 2. Power Analysis Curves. These curves are based on 100% of the burrow entrances (BEs) having camera traps. The number of transects does not include surveying the Central Geographic Stratum as a reference area, which will double the number of transects and camera traps.

Table 1. Data collection requirements for 67% and 50% effect size at a 0.7 and 0.8 statistical power. The number of transects does not include surveying the Central Geographic Stratum as a reference area, which will double the number of transects and camera traps.

0 100 200 300 400 500 600

Reduc�on in WGS BEs

67% 50%

Po w er to

D et ec t D iff er en ce s

Number of Transects

100% of Burrow Entrances with Camera Traps

WGS Field Sampling Plan– November 2021 Page 9

Figure 4. 2014-2015 transect layout for the Central Geographic Stratum. For the 2014 – 2015 WGS Monitoring Project (US Navy 2021), 48 transects (yellow lines) were laid out in the 6,112-ha (16,586-ft) Central Geographic Stratum (red area) in the orientation and spacing shown in this map.

The spacing between transects will be closer together for this plan, depending on power and effect size chosen (Table 1). Orientation may also change.

Figure 3. 2014-2015 transect layout for the MPMGR Buffer. For the 2014 – 2015 WGS Monitoring Project (US Navy 2021), transects were laid out in the 400-m MPMGR Buffer in the orientation and spacing shown in this map. The spacing between transects will be closer together for this plan and the number of transects will be greater. The specifics of both will depend on power and effect size chosen (Table 1). Orientation may also change.

Mul�purpose

Machine Gun

Range

400 meter buffer Survey area

2014-2015 Survey

Transects

North Geographic Stratum

Central Geographic Stratum

South Geographic Stratum

MPMGR

WGS Field Sampling Plan– November 2021 Page 10

Data Collection BE Counts Once the beginning of a transect is located using GPS coordinates, the field surveyor records the:

date, time, GPS location, direction of travel, field personnel, and weather conditions (including temperature and wind speed). A digital photograph is taken in the direction of travel. During the survey, one surveyor walks the center of the transect line, maintaining the direction of travel by GPS while the second surveyor identifies the BEs within 2 m either side of the first surveyor. If a BE is found at the edge of the 2 m boundary, a 2 m measuring tape (or rod) is extended from the center line (positioned by the first field surveyor) to determine if the BE is in or out of the transect area.

A unique number is assigned to each BE encountered within the transect area. A GPS location is taken at each BE for relocating the BE for Camera Trap Surveillance. When a transect is finished, the time of day, end GPS coordinates, and the transect length are recorded.

Camera Trap Deployment

At each BE, a motion-activated, infrared video camera is installed. The video camera is positioned so the BE is in the center of the camera screen and activated. It is kept in position for a predetermined period, which, for the 2014 - 2015 WGS Monitoring Project, was 48 hours. There is a possibility that the 48-hour deployment could be shortened but it would require a complete review of the camera trap video files of the 2014-2015 WGS Monitoring Project to determine a minimum deployment period.

Sampling Period

Sampling must be conducted within a two-year period prior to construction (US FWS 2013).

Because WGS BE occupancy rates change from year to year, camera trap surveillance must be conducted in the same season as the BE count. The BE Survey is conducted during the period when WGS are above ground. Surveying at other times of the year will fail to capture WGS in camera traps. Table 2 shows a typical timeline for pup and adult emergence from hibernation in mid-winter and when they return to estivation in late spring. These dates change from year to year and must be determined in the year of sampling.

WGS Field Sampling Plan– November 2021 Page 11

Table 2. Typical Washington ground squirrel activity timeframe.

Data Analysis

WGS BE Occupancy Rates and Densities

From the camera trap surveillance, a customized computer program, such as MATLAB© (The MathWorks Inc.), is used to process the video files. This program searches the TIFF files, using algorithms to identify the sections of video most likely to contain wildlife. Still-frames from these sections are reviewed by personnel to determine the cause of video camera activation. If a still-frame includes an image of a WGS, it is recorded as a WGS BE. The presence of a kangaroo rate in a still-frame is recorded as a KRAT BE.

The WGS BE occupancy rate is the detection of at least one WGS entering or leaving a BE in a specified time (for the 2014 - 2015 WGS Monitoring Project, that period was a 48 hr period). WGS BE occupancy rate is calculated by dividing the number of BEs where WGS were identified by the total number of camera traps that were employed. WGS BE Density is determined by multiplying BE density by WGS BE occupancy:

𝐷𝐷�ℎ,𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂 = 𝑃𝑃� × 𝐷𝐷�ℎ

𝐷𝐷�ℎ,𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂 = estimated WGS BE Density for an area of interest (WGS BEs/ha)

𝑃𝑃� = proportion of camera traps with at least one WGS detected

𝐷𝐷�ℎ = estimated BE Density for area of interest (BEs/ha)

The variance of WGS BE density is:

𝑣𝑣𝑣𝑣𝑣𝑣�𝐷𝐷�ℎ,𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂� = 𝑃𝑃�2𝑣𝑣𝑣𝑣𝑣𝑣�𝐷𝐷�ℎ� + 𝐷𝐷�ℎ

2𝑣𝑣𝑣𝑣𝑣𝑣�𝑃𝑃�� − 𝑣𝑣𝑣𝑣𝑣𝑣�𝐷𝐷�ℎ�𝑣𝑣𝑣𝑣𝑣𝑣�𝑃𝑃��,

WGS Field Sampling Plan– November 2021 Page 12 where the variance of the proportion occupied is:

𝑣𝑣𝑣𝑣𝑣𝑣�𝑃𝑃�� = 𝑃𝑃� × �1 − 𝑃𝑃��/𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶

The total number of WGS BEs is:

𝑁𝑁�ℎ,𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂 = 𝐴𝐴ℎ × 𝐷𝐷�ℎ,𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂

𝑁𝑁�ℎ,𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂 = Total number of WGS BEs in an area of interest

𝐴𝐴ℎ = Area of the area of interest (ha)

BE Densities Calculations

The BE density is determined by first, calculating the BE density of each transect in the area of interest. This is done by dividing the number of BEs recorded in a transect by the transect area:

di = BE density for an individual transect (BEs/ ha)

Xi = Number of BEs counted on transect

L = Length of transect (typically 500 m) w = Width of transect (typically 4 m)

The densities of all transects are averaged using stratified sampling formula:

𝐷𝐷�ℎ =

𝑛𝑛ℎ

� 𝑑𝑑𝑖𝑖 𝑛𝑛ℎ 𝑖𝑖=1

𝐷𝐷�ℎ = BE density for an area of interest nh = index for each transect within the area of interest di = BE density for an individual transect (BEs/ ha)

The variance is estimated from the usual sampling variance of the mean:

�𝑣𝑣𝑣𝑣𝑣𝑣�𝐷𝐷�ℎ� = 𝑠𝑠𝑑𝑑2/𝑛𝑛ℎ�

WGS Field Sampling Plan– November 2021 Page 13 𝑣𝑣𝑣𝑣𝑣𝑣�𝐷𝐷�ℎ� = variance of BE density of area of interest nh = index for each transect within the area of interest 𝑠𝑠𝑑𝑑2 = standard deviation per transect densities

Statistical Power Analysis

The following is a description of the methods used to estimate the number of transects and camera deployments necessary to detect a specified change in WGS BE numbers with high probability (i.e., statistical power1). It is based on findings from BE Surveys conducted in the 2014

– 2015 WGS Monitoring Project (US Navy 2021). Generally, power to detect change is an increasing function of the number of experimental units (i.e., transects) as well as the magnitude of change under consideration. This analysis provides the resource manager options for detecting a range of magnitude of change in WGS BE numbers with various numbers of transects and levels of statistical power. It is anticipated that resource managers will use these results to select the number of transects necessary to detect biologically meaningful changes in WGS BE occupancy rates and densities to make management decisions regarding potential effects from human activities.

Power Simulation Procedure

Statistical power can be calculated directly from equations relating power and sample size when the sample data are from a small number of known statistical distributions and when sampling designs and analysis methods are fairly simple. In this survey however, a two-stage approach is anticipated, which makes power calculations more complicated. For this reason, the Monte Carlo simulation procedure to estimate power was used. This is a standard approach used by statisticians when mathematical theory is difficult or even intractable. The simulation approach follows these basic steps:

1) Find pilot data like what will be collected as a model for data distributions to be simulated.

2) Randomly sample from these pilot data generating facsimile data sets with known statistical properties.

3) Randomly select a second data set from a modified population to emulate the change in distribution of interest. For example, one might shift the mean by 30% to simulate a change in WGS BE numbers.

4) Analyze the facsimile data sets using the methods that will be applied to real data from the actual study, including conducting tests for differences of interest.

5) Repeat steps 2 through 4 many times (1,000 to 10,000) counting the number of times the null hypothesis is rejected and divide by number of simulated repetitions.

1 The power of a statistical test is the probability of correctly rejecting the null hypothesis when it is false.

For two sample tests, this probability increases with the size of the difference.

WGS Field Sampling Plan– November 2021 Page 14

6) The proportion of simulations for which the null hypothesis is rejected is the estimated power of the test for a given sample size and shift in mean (effect size).

This procedure was followed by fitting distributions to the number of BEs counted per 400 m transect in the 2014 and 2015 field seasons for use as pilot data. To simulate camera deployments, a number of BEs to which cameras would be deployed were randomly selected.

Finally, these BEs were randomly assigned a label of occupied or not occupied by randomly selecting a binary variable with binomial probabilities ranging from 4% to 10%, the range of WGS BE occupancy rates observed in the 2014 and 2015 field season. Data were also simulated to represent post construction sampling with a change in WGS BE occupancy rate ranging from -90% to plus 200%. For each repetition of steps 2 through 5 a two sample Student’s T test was applied to test for a difference in mean number of WGS BEs. Power was summarized in tabular and graphical form.

BACI Design

The BACI (“Before After Control Impact”) study design evaluates whether changes in WGS observed in the MPMGR Buffer are attributable to range development as opposed to fluctuations due to naturally occurring stressors such as drought or disease (Smith 2002, Underwood 1992, Underwood 1993).

The BACI approach requires identification and sampling of specific treatment areas, such as PTAs, as well as reasonably matched reference areas. An advantage of the BACI approach is that PTAs and reference areas do not need to be perfectly matched because differences in WGS metrics between PTAs and reference areas observed prior to disturbance activities are compared with subsequent PTAs and reference areas differences observed after disturbance activities. This is shown in Table 3, where a change in PTA metrics (T1 – TO) is compared to the difference in reference area metrics (R1 – R0). Reference and PTAs do not need to have similar baseline conditions, because it is the change in metrics that are compared. Nevertheless, the areas should be similar enough to expect that temporal environmental fluctuations would affect reference areas and PTAs similarly. The underlying hypothesis is one of parallel temporal trends against the alternative that temporal trends are not parallel—potentially indicative of an effect due to the implementation of the FEIS.

Table 3. Pre-and post-treatment metrics structure in a BACI sampling design.

WGS Field Sampling Plan– November 2021 Page 15

Summary

This field sampling plan outlines the survey effort and methodology required to estimate the number of WGS BEs (Washington ground squirrel burrow entrances) within the MPMGR Buffer and Central Geographic Stratum prior to expansion of Navy activities and determine whether changes in WGS BE densities are attributable to these activities or to unrelated causes. We present a power analysis of two effect sizes – a 67% and a 50% reduction in WGS BEs.

For a 67% effect size at a 0.8 statistical power, 314 transects are required in the MPMGR Buffer and 314 in the Central Geographic Stratum, for a total of 628 transects. The transects in the MPMGR Buffer will cover 21% of the area, with placement of transect centerlines 19 m apart. For a 50% effect size at a 0.8 statistical power, 1088 transects (544 in the buffer and 544 in the stratum) will be required.

The BE Survey would be conducted within a two-year period prior to disturbance during the period of the year when WGS are above ground (i.e., mid-winter to late spring). Consistent with requirements in the Conference Opinion, post-disturbance surveys would be conducted once every two years for a period of ten years.

This power analysis assumes a density of 3.8 BEs/ha. If there are substantially more or less BEs per hectare than assumed, the field team should consider reevaluating the power and precision goals early in the field season to determine if the number of camera deployments should be modified.

References Cited

LimnoTech. 2013. WGS Monitoring Plan – NWSTFB. Internal Navy Document. Contract Number N62473-06-D2401 to 0094. 53p.

NWC (Northwest Wildlife Consultants, Inc) and Yensen E. 2007. Washington ground squirrel background information and project mitigation concepts. Prepared for Oregon Military Department, Environmental Branch. Salem, OR: Oregon Military Department.

Smith, E.P. 2002. BACI design. In: Encyclopedia of Environmetrics , El‐Shaarawi, A.H. and W.W.

Piegorsch. John Wiley and Sons, New York, pp 141‐148.

Underwood, A.J. 1992. Beyond BACI: the detection of environmental impacts on populations in the real, but variable, world. Journal of Experimental Marine Biological Ecology. 161: 145‐178.

Underwood, A.J. 1993. The mechanics of spatially replicated sampling programmes to detect environmental impacts in a variable world. Australian Journal of Ecology. 18:99‐116.

U.S. Fish and Wildlife Service. 2013. Conference Opinion Regarding the Effects of Proposed Military Readiness Activities at the Boardman Naval Weapons Systems Training Facility in Morrow County, Oregon on the Washington Ground Squirrel. File No. 01EOFW00-2013-FC-0017. LaGrande Field Office, LaGrande Oregon. 121 p.

WGS Field Sampling Plan– November 2021 Page 16

U.S. Navy. 2015. Final Environmental Impact Statement for Naval Weapons Systems Training Facility. c/o Pacific Fleet Environmental Office, 258 Makalapa Drive, Suite 100, Pearl Harbor, HI 96869-3134, 569p.

U.S. Navy. 2021. 2014 – 2015 WGS Monitoring Project – Naval Weapons Systems Training Facility Boardman Oregon. c/o Pacific Fleet Environmental Office, 258 Makalapa Drive, Suite 100, Pearl Harbor, HI 96869-3134, 118p.