Forecasted greater sage-grouse conservation using novel mitigation

Provencher, Louis 1 ; Badik, Kevin J.1 ; Byer, Sarah1 ; Munn, Liz1 ; Anderson, Tanya1 ; Sedinger, James S.2 ; Campbell, Jeff3 ; Frid, Leonardo 4 ; Low, Greg L.5

Published May 15, 2026 on Dryad. https://doi.org/10.5061/dryad.r7sqv9ssn

Data files

May 15, 2026 version files 6.72 GB

DHSMForPub.zip

5.31 MB
New_GIS_04252023.zip

17.67 MB
NGM_Robertson_ServerRevision.zip

6.70 GB
README.md

22.26 KB
Vegetation_Descriptions.pdf

976.64 KB

Abstract

A common Intermountain West US land management challenge is the conservation of declining greater sage-grouse (GSG) populations amid development pressures in sagebrush shrublands on public lands. High-quality compensatory mitigation (or “offsets”) is an important tool for conservation where impacts are not otherwise being avoided, such as in mine development. We tested whether a compensatory mitigation bank for GSG coupled with public lands policy innovations was feasible in arid shrublands with decadal succession and high failure rates for seedings.

Vegetation change and the associated change of GSG per-capita population growth rate (λ) due to proposed vegetation treatments were modelled on a 218,049-ha central Nevada landscape. Using remote-sensed vegetation rasters and spatial state-and-transition simulation models (STSM) coupled with a demographic GSG model, we compared the effects of proposed vegetation treatments (uplift scenario) to baseline (control) actions on a net gain of λ in sagebrush shrublands and wet meadows across public and private lands. Dominant treatments separated into thinning conifers encroaching shrublands and seeding in shrublands dominated by non-native annual species after past fires. Smaller treatments were used in wet meadows, which are critical GSG late-brood rearing vegetation. The primary data components were, therefore, vegetation remote sensing, spatial state-and-transition simulation models with many input raster that provide initial conditions and spatial constraints imposed on natural and management-caused disturbances, and a greater sage-grouse demographic habitat suitability model with its associated input rasters and lookup tables.

The landscape-wide average λ was 0.74230 at the project’s onset in 2014. The average λ between scenarios was 0.71653 and 0.72289, respectively, for the baseline and uplift scenarios after 35 years of simulations.

The significant λ differences between scenarios and innovative science and policies enabled the parties of the mitigation bank to consider future credit releases based on implemented and monitored aridland restoration actions conducted on private and public aridlands.

There are one simulation database, one supporting simulation folder, Geotiff raster folder, one greater sage-grouse habitat suitability model folder, and one MS Word document of the vegetation description in the zipped archived data:

(A) "NGM_Robertson_ServerRevision.zip" zipped archived folder contains the simulation database "NGM_Robertson_ServerRevision.ssim," which requires the freeware Syncrosim (Syncrosim version 2.4.18 or later; downloaded from www.apexrms.com) and packages ST-Sim (package ST-Sim version 3.3.12 or later; obtained remotely from within Syncrosim) and the input and output data (NGM_Robertson_ServerRevision.ssim.data) folder. The Syncrosim database and data folder are massive (18.7 GB of memory compressed). The data folder was created and managed by Syncrosim during simulations and should not be tampered with as all parameters and data to run the simulations are contained within. The Authors have never worked within the data folder. The NGM_Robertson_ServerRevision.ssim.data contains simulation scenarios. (B) In addition to the simulation database, one folder contain all needed Geotiff rasters (no other GIS files; folder "New_GIS_04252023.zip") that were uploaded in the Syncrosim database (i.e., already uploaded). These Geotiff rasters will be needed for upload if the simulation is conducted on a server with different directory pathway organization than the Cloud server used by this project. (C) The greater sage-grouse habitat suitability model in the folder titled "DHSMForPub.zip" (D)The vegetation description of all ecological systems and their vegetation classes used in remote sensing and models are found in "Provencher_Vegetation_Descriptions.pdf."

Description of the data and file structure

(A) "NGM_Robertson_ServerRevision" Folder. While we cannot describe the content of the Syncrosim database and data folder (again, the "NGM_Robertson_ServerRevision.ssim.data" was entirely created and managed by Syncrosim during simulations and not the authors), we will describe the rasters, which populate some menus of the simulation database "NGM_Robertson_ServerRevision.ssim," within “New GIS 04252023” in part (B) below. The folders and scenarios of "NGM_Robertson_ServerRevision.ssim" are described in the section below "Code/Software."

(B) “New GIS 04252023" contains all the Geotiff rasters used in ST-Sim:

1. CTZ_60m_SYS_042423.tif was an initial condition raster of the ecological system identity per pixel whose numeric values are in the Definitions of ST-Sim/Syncrosim;

2. CTZ_60m_CLA_042423.tif was the initial condition for vegetation classes within ecological systems where each unique vegetation class corresponded to a numerical code found in the Definitions of ST-Sim/Syncrosim;

3. CTZ_60m_LO_042423.tif was an initial condition raster that assigned land ownership values to each pixel, where numeric values are found in the Definitions of ST-Sim/Syncrosim;

4. CTZ_60m_Elevation_042423.tif was used in Slope Multipliers option in Advanced menu to affect the spread of fire on different slopes (already defined in ST-Sim);

5. CTZ_60m_FireStarts_042423.tif was used to control the probability of fire initiation caused by lightning strikes and human ignitions, and used in the menu Spatial Settings->Advanced->Transition Spatial->Transition Spatial Initiation Multipliers of ST-Sim;

6. CTZ_60m_2017Fire_042423.tif was used to force fires that happened in 2017 as the new simulations of this project started in simulated year 2021. The file was used to control fire in the menu Spatial Settings->Advanced->Transition Spatial->Transition Spatial multipliers;

7. CTZ_60m_2018Fire_042423.tif was used to force fires that happened in 2018 as the new simulations of this project started in simulated year 2021. The file was used to control fire in the menu Spatial Settings->Advanced->Transition Spatial->Transition Spatial multipliers;

8. CTZ_60m_2019Fire_042423.tif was used to force fires that happened in 2019 as the new simulations of this project started in simulated year 2021. The file was used to control fire in the menu Spatial Settings->Advanced->Transition Spatial->Transition Spatial multipliers;

9. CTZ_60m_2020Fire_042423.tif was used to force fires that happened in 2020 as the new simulations of this project started in simulated year 2021. The file was used to control fire in the menu Spatial Settings->Advanced->Transition Spatial->Transition Spatial multipliers;

10. CTZ_60m_AllFire_042423.tif was used to allow fires across the entire landscape starting in 2021 as the new simulations of this project started in simulated year 2021 (thus overriding the CTZ_60m_2020Fire_042423.tif file). The file was used to control fire in the menu Spatial Settings->Advanced->Transition Spatial->Transition Spatial multipliers;

11. CTZ_60m_BufferRoads_042423.tif stamped in fuels breaks along busy paved and dirt roads by substantially lowering the fire return interval of the selected pixels. A value of 0.0001/year multiplied a pixel's fire return interval by 0.0001 (i.e., will rarely burn), whereas all other pixels will have unchanged fire return intervals by multiplication by a value = 1. The file was used in the menu Spatial Settings->Advanced->Transition Spatial->Transition Spatial multipliers, but only for the Uplift scenario;

12. CTZ_60m_FenceWaterDelivery_042423.tif was used to constrain where fencing and water delivery systems would be implemented around wet meadows. The file was used in the menu Spatial Settings->Advanced->Transition Spatial->Transition Spatial multipliers, but only for the Uplift scenario;

13. CTZ_60m_FeralHorse_042423.tif as used to spatially constrain where feral horse are found and graze. The file was used in the menu Spatial Settings->Advanced->Transition Spatial->Transition Spatial multipliers;

14. CTZ_60m_GrazeEarlySpr_042423.tif was used to spatially constrain location and intensity of cattle grazing during the early spring season with binary 0 (not grazed) or 1 (grazed) pixel values. The file was used in the menu Spatial Settings->Advanced->Transition Spatial->Transition Spatial multipliers;

15. CTZ_60m_GrazeFall_042423.tif was used to spatially constrain location and intensity of cattle grazing during the fall grazing season with binary 0 (not grazed) or 1 (grazed) pixel values. The file was used the menu Spatial Settings->Advanced->Transition Spatial->Transition Spatial multipliers;

16. CTZ_60m_GrazeLateSpr_042423.tif was used to spatially constrain location and intensity of cattle grazing during the late spring grazing season with binary 0 (not grazed) or 1 (grazed) pixel values. The file was used in the menu Spatial Settings->Advanced->Transition Spatial->Transition Spatial multipliers;

17. CTZ_60m_GrazeSummer_042423.tif was used to spatially constrain location and intensity of cattle grazing during the summer grazing season with binary 0 (not grazed) or 1 (grazed) pixel values. The file was used in the menu Spatial Settings->Advanced->Transition Spatial->Transition Spatial multipliers;

18. CTZ_60m_WildHorseGrazing_042423.tif was used to spatially constrain wild horse grazing to herd Management Areas during the summer grazing season with binary 0 (not grazed) or 1 (grazed) pixel values. The file was used in the menu Spatial Settings->Advanced->Transition Spatial->Transition Spatial multipliers;

19. CTZ_60m_WaterDist_SpringFallCowSlope_042423.tif was used to spatially modify cattle grazing pressure as a function of distance and slope from a water source during the spring season of use. The file was used in the menu Spatial Settings->Advanced->Transition Spatial->Transition Spatial multipliers;

20. CTZ_60m_WaterDist_SummerCowSlope_042423.tif was used to spatially modify cattle grazing pressure as a function of distance and slope from a water source during the spring season of use. The file was used in the menu Spatial Settings->Advanced->Transition Spatial->Transition Spatial multipliers;

21. CTZ_60m_WaterDist_SummerHorse_042423 as used to spatially modify cattle grazing pressure as a function of distance and slope from a water source during the summer season of use. The file was used in the menu Spatial Settings->Advanced->Transition Spatial->Transition Spatial multipliers;

22. CTZ_60m_GoldBarTrees_042423.tif constrains where special small tree lopping could only be implemented in the Gold Bar mine area in the southern Roberts Mountains. The file was used in the menu Spatial Settings->Advanced->Transition Spatial->Transition Spatial multipliers;

23. CTZ_60m_GrassValleyExclusion_042423.tif was used to spatially prevent treatments seeding treatments in Grass Valley where the mine's Plan of Operation applies. The file was used in the menu Spatial Settings->Advanced->Transition Spatial->Transition Spatial multipliers, but only for the Uplift scenario;

24. CTZ_60m_HPSSExclude_042423.tif was used to spatially exclude mixed native and introduced grass species seedings in Frenchie Flat. The file was used in the menu Spatial Settings->Advanced->Transition Spatial->Transition Spatial multipliers, but only for the Uplift scenario;

25. CTZ_60m_SeedingOnly_042423 was used to spatially only allow pure introduced grass species seedings in Frenchie Flat. The file was used in the menu Spatial Settings->Advanced->Transition Spatial->Transition Spatial multipliers, but only for the Uplift scenario;

26. CTZ_60m_JDR_Irrigation_042423.tif was used to spatially limit implementation of irrigation and pasture seeding beneficial to greater sage-grouse brood rearing in basin wildrye bottomland of the JD Ranch Headquarters pastures. The file was used in the menu Spatial Settings->Advanced->Transition Spatial->Transition Spatial multipliers, but only for the Uplift scenario;

27. CTZ_60m_Slope15_042423.tif was used to spatially prevent implementation of treatments that cannot be used on slopes greater than 15%. The file was used in the menu Spatial Settings->Advanced->Transition Spatial->Transition Spatial multipliers, but only for the Uplift scenario;

28. CTZ_60m_Slope30_042423.tif was used to spatially prevent implementation of treatments that cannot be used on slopes greater than 30%. The file was used in the menu Spatial Settings->Advanced->Transition Spatial->Transition Spatial multipliers, but only for the Uplift scenario;

29. CTZ_60m_TreatmentZones_042423.tif was used to prevent management actions in allotments or pastures not controlled by Barrick or in the mine's plan of operations. The file was used in the menu Initial Conditions->Tertiary Stratum for management zones where numerical codes are found in the Definitions folder of ST-Sim/Syncrosim under Strata->Tertiary Stratum;

30. CTZ_60m_NoFire_042423.tif was used to completely suppress fires across the entire landscape during certain intervals. The file was used to control fire in the menu Spatial Settings->Advanced->Transition Spatial->Transition Spatial multipliers;

Static rasters

CTZ_60m_SYSXCLA_120420.tif is the initial ecological systems and classes in 8-digit code
Past_Reclass12122018.tif is for Veg clean-up for pasture classification
CTZ_60m_Authorized_ClearVeg_090419.tif is to identify previously authorized disturbances that cleared vegetation
CTZ_60m_DSEP_ClearVeg_090419.tif is to identify a permitted project within the project area that cleared vegetation
DeepsRushBig.tif is an additional project footprint used for comparison that cleared vegetation and location of noise producing, potential predator perches, or busy roads associated with the DeepsRush project
CTZ_60m_HN_Baseline_060420.tif is to identify location of noise producing, potential predator perches, or busy roads for previously authorized disturbances
CTZ_60m_DSEP_HN_090419.tif is to identify location of noise producing, potential predator perches, or busy roads associated with the DSEP project
CTZ_60m_Elevation_120420.tif is a digital elevation model (DEM), in meters, that matches the resolution of the vegetation data
CTZ_60m_SlopeDegrees_120420.tif is an aspect raster derived from the DEM
JDMask.tif is used to match vegetation with known treatments in the JD allotment
CTZ_60m_Authorized_HN_090419.tif is to identify location of noise producing, potential predator perches, or busy roads associated with previously authorized disturbances

Lookup tables

GSG_habitat11192020.txt identifies potential greater sage-grouse (GSG) habitat from the classified vegetation
nesting_sub11192020.txt", header =TRUE) identifies potential nesting habitat
nesting_ms_sub11192020.txt identifies potential nesting habitat in the mountain shrub ecological system
BG_ActPend_lek.txt coordinates of active and pending leks with 15-km of the project boundary. Due to the GSG status the data cannot be provided here.
nonsage_sub11192020.txt is a dataframe for non-sagebrush shrub cover table used for nesting
nonms_sub11192020.txtis a dataframe for accepted mountain shrub cover table used for nesting
grassland_sub11192020.txt is a dataframe for grassland, includes annual grasses, crested wheatgrass, and early seral native grasses dominated classes
LBH_noWM11192020.txt is a dataframe of accepted late brood habitat outside of wet meadows and below the elevation threshold
LBH_elev11192020.txt is a dataframe for late brood habitat that is above 2133.6 m (7000 ft) in elevation
LBH_hum11192020 is a dataframe to identify hummocked wet meadow vegetation to modify late brood habitat value
SG_WM_sub11192020.txt is a dataframe to identify late brood habitat in wet meadow systems
ForbCov11192020.txt is a dataframe for forb cover proxy variable
SPEI-Sept8moV2.txt imports list of iteration-timesteps that are below 1 standard deviation in SPEI (Sept. 8-m lag) to remove dry meadows as late brood habitat in those years
GSG_HSIForPub.R is the R script to calculate lambda (or the per capita growth rate) for each pixel with the project boundary
DHSM.R is the R script used with SyncroSim to identify priority areas for restoration to increase the value of GSG habitat
LBH_noWM_DryYears11192020.txt modifies late brood habitat categories based on whether the year is classified as a dry year

(D) Vegetation Description file "Provencher_Vegetation_Descriptions.pdf" contains all ecological systems and vegetation classes mapped by remote sensing and models. The 8-digit codes associated with each vegetation classes is exactly used as such in the ST-Sim's definitions. All systems are listed in Table of Contents.

Sharing/Access information

Links to other publicly accessible locations of the data:

Not applicable. None of the original remote sensing from the Spot 6/7 at 1.5m pixel resolution is available because private industry paid for it and is used for on-going mitigation decisions; however, the same imagery can be purchased as archival imagery from Airbus Defence and Space. Also, the standard license agreement for purchase of any imagery with Airbus Defence and Space prevents us from sharing the imagery. The simulation Geotiff rasters for systems and vegetation classes above were resampled from 1.5m to 60m.

Code/Software

Simulations require the freeware Syncrosim (Syncrosim version 2.4.18 or later; downloaded from www.apexrms.com ) and packages ST-Sim (package ST-Sim version 3.3.12 or later; obtained remotely from within Syncrosim; downloaded from www.apexrms.com ). Follow instructions from ApexRMS to install the software. The software will open the simulation database NGM_Robertson_ServerRevision.ssim and associated folders (NGM_Robertson_ServerRevision.ssim.data) or by double-clicking the X.ssim files. Beyond this point, advanced training in the use of Syncrosim and the ST-Sim package will be required. Training can be arranged with ApexRMS at www.apexrms.com.

Assuming training was completed, a primary folder named "Definitions" folder will be found after opening NGM_Robertson_ServerRevision.ssim. Nested in the "Definitions" folder are two folders: "1-Spatial Scenarios" and "2-Subscenarios."

(A) If one right-clicks on "Definitions" and selects Open, the folder contains all the codes, structures, and organizations of data that allows upload of tables and rasters, and production of all results. The reader should not modify any Definitions because it can delete all results or corrupt results. Exploring these folders requires training in the use of ST-Sim. Some sub-menus require advanced training to understand, or even modify. In this Readme file, the reader is sometimes directed to precise subfolders to locate codes for ecological systems and vegetation classes. Most tables in the menus can be exported as Excel tables.

(B) Nested in the "Definitions" folder is the folder "1-Spatial Scenarios" containing the simulation scenarios "Baseline_OptionA_TargetsForCattleGrazing" (= control scenario representing the current baseline) and "ACA_OptionB_TargetsForCattleGrazing" (scenario with added management). Each of the latter two folders contain the folders "Dependencies" and "Results." The Dependencies should not be tampered with and contain all the sub-scenarios (see below) that built each simulation scenario (i.e., each set of different sub-scenarios define a simulation scenario). The "Results" folder contains the results expressed as a scenario we named "PM-v3.1.0" and "PM-v3.2.0", respectively, in the folders "Baseline_OptionA_TargetsForCattleGrazing" and "ACA_OptionB_TargetsForCattleGrazing." The reader will notice unique numerical codes before scenario names that were automatically assigned by ST-SIM and not the authors. The term "PM" stands for Proposed Management in the third version of the project (v3), and the numbers "1" and "2", respectively, stand for the baseline = 1 and added management = 2. The result scenarios are already loaded with results (i.e., no simulations are required to make figures and tables) and can be charted or mapped by first right-clicking on the scenario and selecting "Add to Results." The charting and mapping menus are found at the bottom of the software and requires training to use properly beyond pre-set figures assembled by the authors. Do not use the other scenario without results, but with "Dependencies" folder, named "Custodial-Modern." This represents a true control scenario with any management from the 2014 phase of the project (but not used in version 3 of this project).

(C) All supporting input data are in folder "2-Subscenarios." Should the reader choose to run simulations, the reader may need to upload all rasters (see above in section 1. (B)) in different menus. Each subscenario is shown as a folder that, once opened, contains one or more scenarios that uniquely defines a component of the simulations. The ten subscenario folders and their nested sub-scenarios are:

"0 RunControl & Output_Options"

a) "RunControl & OutputOptions"

"1 Initial Conditions"

a) "Non-Spatial Initial Conditions" used for debugging simulations

b) "Spatial Initial Conditions - Simplified03242023" contains the primary vegetation, land ownership, and treatment zones input data of simulations

"2 Pathways"

a) "Pathways" contains all state-and-transition simulation models

"3 Transition Multipliers" where each scenario uniquely defines simulations scenarios

a) Folder "NRV and Custodial" that was used but sets parameters for the true control

b) "Transition Multipliers_ACA_OptionB_No_C-ACA"

C) "TransitionMultipliers_Baseline05102023"

"4 Attributes" this complex menu defines all the cost of treatments and wild horse grazing regimes per simulation scenarios and requires advanced ST-Sim training to explore

a) "Attribute Values Cust_OptionA"

b) "Attributes Values ACA_OptionB"

c) "Attibute Values_Cust_OptionA_TargetsForGrazing"

d) "Attributes Values_ACA_OptionB_TargetsForGrazing"

e) "Attributes Values ACA_OptionB -- No CattleTargets"

f) "Attributes Values ACA_New-Custodial"

6."5 Targets" define the annual areas of cattle grazing and treatments per simulation scenario

a) "Targets_ACA_OptionB_TargetsForGrazing - NotHorses_No-C-ACA"

b) "Targets_Baseline_OptionA_TargetForGrazing -NotHorses - No_C-ACA"

"6 External_Variables & Distribution" is an advanced menus of ST-Sim

a) "External Variables" contains all climate time series expressed as different Standard Precipitation-Evapotranspiration Index time series

b) "Distribution" is the cross-walking menu that converts climate time series to ecological disturbances

"7 Spatial_Settings"

a) "Custodial" is a folder that contains the unused spatial seetings for the true control (not simulated)

b) "Spatial_Settings_Baseline+ACA - NoAsInSpreadDistribution" is the most complex scenario of ST-Sim and is were all other input rasters of section 1. (B), except those of initial conditions, are uploaded.

"8 Reference Condition & UED"

a) "ReferenceConditions" is the unused scenario containing the percentages defining the reference conditions for all ecological systems and used to calculate Unified Ecological Departure (see Provencher et al. 2021 in publication)

"9 External Program"

a) "10 External Program" is the scenario that call the R script of section 1. (C) to calculate greater sage-grouse habitat suitability. The "10 External Program" subscenario was populated with the R script pathway to use in folder "DHSMForPub" and the timesteps to calculated sage-grouse habitat suitability; however, the user will need to specify the pathway to the R executable file on their machine in the first line titled "External program." We left the line blank because the pathway we used on a AWS Cloud server was just for that server and would be meaningless for other machines.

(D) The simulations required one week of run time use an AWS server with at least 24 processors running in parallel mode and 256 GB of memory.

2.1. Study Area

The 218,049 ha Area of Interest (AOI) includes the Cortez Mountains, Pine Valley, Simpson Park Mountains, and Roberts Mountains, Crescent Valley, the northern Toiyabe Range, and Shoshone Range in central Nevada (hereafter Cortez at 40^o08’49.27” N, 116^o18’52.11” W; Fig. 1). The geology is within the Basin and Range Province with volcanic rocks dominating, although the eastern half of the Roberts Mountains and deeper bedrock are made of carbonate rock. Gold and silver exploration and mining has occurred in the region since 1862’s.

Ecological systems are zonally stratified by precipitation and soils as is typical of the Great Basin ecoregion. The dominant ecological systems are saline shrublands, sagebrush shrublands on semi-desert to mountain soils, and sub-xeric woodlands. Wet meadows, mountain shrubs, aspen woodlands, and subalpine conifers are scattered across these systems at different elevations. GSG populations are primarily found in the Cortez Mountains, Pine Valley, Roberts Mountains, Simpson Park Mountains, and Shoshone Range. From lower to higher elevations, GSG either consume or utilize the vegetation communities dominated by Wyoming big sagebrush (Artemisia tridentata subsp. wyomingensis), black sagebrush (A nova), low sagebrush (A. arbuscula), and mountain big sagebrush (A. tridentata subsp. vaseyana). GSG will also utilize wet meadows where chicks feed on invertebrates during brood rearing.

2.2. Overview of Methodology

Landscape Conservation Forecasting™ helps land managers design cost-effective strategies to restore ecological systems in large landscapes. The method can be summarized by the “3 Ms”: Maps obtained from remote sensing, Models designed in a spatial state-and-transition simulation software, and Metric of condition provided by GSG habitat suitability. Each of the three “Ms” form the next sections of the Methods.

2.3. Maps

STSMs and metric estimation required two vegetation layers from remote sensing: ecological systems and their vegetation classes. Ecological systems are potential vegetation types expected in the physical environment under natural disturbance regimes usually named for the dominant upper-layer vegetation (e.g., Wyoming big sagebrush). Vegetation class is the ecological system’s current identity defined by canopy structure, succession, and whether it is in reference or uncharacteristic condition.

Remote sensing was conducted to map ecological systems and vegetation classes from Spot6/7 1.5-m resolution multi-spectral pan-sharpened satellite imagery over time to create an initial conditions map that is up to date as of 2019. Private inholdings not belonging to Barrick were excluded from field surveys and mapping.

2.4. State-and-Transition Simulation Models

STSMs are stochastic models of landscape change and used to forecast the outcomes of what-if scenarios. The landscape in an STSM consists of a discrete set of simulation cells classified into a discrete set of states. Those discrete sets of states were obtained from the remote-sensed ecological systems and vegetation classes. Simulation cells change over discrete timesteps according to a discrete set of possible transitions that are caused by either natural (e.g., wildfire) or anthropogenic (e.g., seeding) processes operating within temporal or spatial constraints. All simulations were conducted using the ST-Sim package (Version 3.3.12) in SyncroSim (Version 2.3.18; https://docs.stsim.net/). A slightly older but very similar modeling approach presenting the same methodological material used here was published in Provencher et al. (2021; Climate, 9, 79). Below are new modelling details.

2.4.1 Fire

Fire was the most important disturbance in the AOI. Fire ignitions were potentially correlated with areas prone to lightning strikes and areas of high human activity. For natural ignitions, raw point location shape files of observed lightning strikes were obtained from the BLM via a request to the Western Regional Climate Center at Desert Research Institute (https://wrcc.dri.edu/, last accessed 06-10-2023). Strikes were converted to a 60-m density raster (pixel value ranged from 0 to 1) based on a 1.92-km (32 x 32 pixels) moving window. The moving window size was determined by trial-and-error. The highest value was 1, observed in a few pixels with the highest density of lightning strikes. A second layer for anthropogenic ignitions was created by first measuring the distance of each pixel to the nearest busy road (i.e., paved road or gravel road with traffic from mine or ranch equipment) or infrastructure. These distances were converted to a value between 0 and 1 using the equation in Provencher et al. (2021):

The probability of a human ignition decays with distance from roads or infrastructure, with a value of 1 being the greatest probability within half of pixel distance form edge of road. These lightning strike and human initiation layers were combined, such that the maximum value between the two layers for each pixel was retained. After fire initiation, the size distribution and spread of fire activity were as described in Provencher et al. (2021).

2.4.2. Grazing

Grazing was caused by cattle, primarily, wild horses in federal Herd Management Areas, and feral unbranded horses (i.e., outside of feder2al Herd Management Areas and under the responsibility of the State of Nevada’s Department of Agriculture) found from Pine Valley over the Cortez Mountains to Frenchie Flat. Cattle grazing was the most widespread disturbance and as such controlled most the duration of simulations.

Cattle grazing events from all pixels per each of four seasons of use were tallied across all allotments and pastures to estimate the distribution of use. The use by pasture was controlled with area targets (thus, four seasonal targets) matching the cumulative area of pastures grazed. Relative intensity of use for each pasture within allotments was weighted by heads of cattle, proportion of days per year in a pasture, and area of pasture. The annual areas grazed for the early- spring, late spring, summer, and fall, respectively, were 253,046 ha, 259,146 ha, 237,994 ha, and 279,901 ha.

Wild horses were constrained to Herd Management Areas and iteratively controlled by one time-varying target representing the number of horses, which was expressed as the amount of AUM consumed. Manipulating the size of the cattle herd was not a management objective; however, changing the size of the horse populations was a management action that required this approach. Because wild and feral horses occupied small footprints, iteratively simulated horse grazing with AUM consumed (surrogate for herd size) did not lengthen too much computing time, whereas using the same iterative approach with cattle was shown to add two weeks of computation time, which was unacceptable in a mitigation project with mining deadlines. The annual variation in AUMs was the result of periodic BLM herd size management and doubling of population size every four years. To reflect the field observations for feral unbranded horses, target AUM time series were revised such that AUM consumed dropped from 20,073 in 2016 to 300 AUMs in 2017 due to county government gathers followed by a rate-of-increase of 20 percent per year (i.e., feral horse numbers doubled every 4-5 years). It was assumed that 20,073 AUM consumed was the carrying capacity based on estimates of feral horse numbers pre-2017. Therefore, for custodial management scenarios, feral horse AUM were allowed to increase until they crossed the carrying capacity. Once the carrying capacity threshold was crossed, AUM consumed were reduced below the threshold until they crossed it again. For the proposed management scenario, if AUMs exceeded 2,880 (200 horses), the model lowered the AUM value to 280, after which horse number could increase again.

More important than the stocking rates or pasture locations, the distance to a water source and slope (slope effect only for cattle as wild horses were found on all slopes) determined intensity of use in any pixel. The distance from water to grazing relationships presented below were developed with local livestock operators and grazing experts during a model building workshop in 2015; the relationships were converted to continuous sigmoid equations to create rasters (Eqs. 1-3). Pixels closest to a water source received the greatest intensity of grazing:

Intensity of summer cattle grazing

= 1 - e^{5.1 ∙ distance – 11} / (1 + e^{5.1 ∙ distance – 11}) (Eq. 1)

Intensity of spring or fall cattle grazing

= 1 - e^{3.15 ∙ distance - 4.5} / (1 + e^{3.15 ∙ distance - 4.5}) (Eq. 2)

Intensity of horse grazing

= 1 - e^{2.1 ∙ distance – 4} / (1 + e^{2.1 ∙ distance - 4}) + 0.2. (Eq. 3)

Grazing intensity decreased as slope (%slope is the slope percentage) increased. We converted Ganskopp and Vavra’s (1987) categorical values to a continuous sigmoid equation (Eq. 4) to create a raster (each pixel value in this raster multiplied the value of the same pixel from the appropriate cattle grazing season):

Intensity of cattle grazing a pixel based on slope

= 0.75 · (1 - e^{25 · (0.013 · %slope) - 3.5} / (1 + e^{25 · (0.013 · %slope) - 3.5}) (Eq. 4)

2.4.3.Temporal Variability Effects on Disturbances

Transition pathways among vegetation classes were governed by fixed rates (e.g., fire’s rate could be 0.01/year = 1/100 years, the inverse of a 100-year mean Fire Return Interval). Annual and monthly climate variability have profound effects on the annual variation of disturbances that cause transitions. Historic climate time series from PRISM and USGS flow gage data were obtained to model variation around the fixed value of disturbance rates. A stochastic weather generator (SWG) was used to replicate each PRISM climate time series 20 times over 35 years – the duration of all simulations. The following disturbances were affected by climate variability: fire, tree encroachment in shrublands, wet years, very wet years, snow deposition, severe drought, tree invasion into shrublands, insect and disease outbreaks, exotic forb or tree invasion, non-native annual species invasion, and annual peak flow events. The methodology of temporal variability was explained in Provencher et al. (2021).

2.4.4. Scenarios

All scenarios were simulated for 35 years. Two underlying scenarios formed the basis of our analysis: “Baseline” and “Uplift” scenarios. The Baseline scenario included a series of treatments that had already been agreed-to in a geography called “Initial Conservation Area”. The uplift scenario included additional vegetation treatments both in time and in space in a geography encompassing the Initial Conservation Area and an “Additional Conservation Area." In both cases, vegetation treatments were designed to fall outside of a designated mine development area to minimize the potential to devalue treatment outcomes because of future development. Likewise, proposed development could be “stamped” into the baseline scenario to evaluate the loss of GSG habitat due to mining activity over time.

The Baseline and Uplift scenarios implemented most treatments in sagebrush systems and wet meadows. The same set of vegetation treatments were used in both scenarios, except spraying of indaziflam to control annual species, and the treatment of riparian systems using low-technology process-based riparian restoration techniques were only used in the Uplift scenario. In the baseline scenario, the bulk of vegetation treatments occurred between timestep 3 and timestep 10. The uplift scenario included all these vegetation treatments and added additional vegetation treatments between timesteps 10 and 20.

2.5. GSG Habitat Suitability Model

A statistical and spatial demographic habitat suitability model was developed from a 2003-2012 field study in response to mitigation of the Falcon-to-Gonder transmission line passing over the northern Cortez Mountains and through Pine Valley in Eureka County. Habitat suitability was the metric used to spatially prioritize vegetation treatments compared to maintaining Baseline management. GSG habitat suitability was estimated from current and forecasted vegetation rasters with an R script that was coupled to the SyncroSim software. ST-Sim supplied forecasted vegetation rasters to the R script every 5 years, which subsequently returned an implementation likelihood raster to ST-Sim. More specifically, the R script of habitat suitability returned one raster where pixels with intermediate habitat suitability received the highest vegetation treatment implementation likelihood because pixels in good to excellent condition could only contribute minimally or not at all to mitigation uplift and poor areas were invariably poor because they were too far from leks or brood rearing vegetation (Provencher et al., 2021).

The static habitat suitability of GSG and estimation of spatially explicit λ (Lambda or per-capita population growth rate for any pixel) underpinning this project was detailed in Kane et al. (2017;Ecosphere, 8(7). https://doi.org/10.1002/ecs2.1869). When λ = 1, the population was assumed to be stable. Values above 1 indicated population growth and values below 1 indicated population declines.

2.6. Statistics

We used a randomized complete block design to detect the effects of scenarios. Each climate time series was a block with 19 degrees of freedom (df; 20 replicates -1) for the blocking effect because the same climate replicate was applied to both scenarios (Steel and Torrie 1980). The fixed effect had 1 df because there were only two fixed scenarios.