top of page

The Living Range Project

Conservation Tech for Working Lands Intelligence

Executive Summary

Project: Holland Ranch Conservation Technology Demonstration
Location: Southwest Montana (private, state, BLM, and U.S. Forest Service grazing lands)
Total Acres: 129,007
Lead Ranch Partner: HRL Inc (Land) and Holland Ranch Company (Livestock)
Technology & Science Partners: Grizzly Systems, Planet, Halter, Lonestar GPS, SPOT, Cultivo
Supporting Organizations: The Nature Conservancy, Southwest Montana Sagebrush Partnership

The Holland Ranch Conservation Technology Project is a real-world demonstration of how modern sensing technology and adaptive grazing tools can be integrated into a working, family-operated ranch to deliver verifiable conservation outcomes while maintaining economic viability. The project explicitly tests what works, what partially works, and what fails when technology is deployed under operational ranching constraints—off-grid, across mixed land tenures, and in landscapes with large carnivores.

​

Core Innovation

Rather than relying on models alone, the project combines:

  • Ground-based sensing (audio, imagery, chemical, livestock telemetry),

  • High-frequency satellite observation, and

  • On-the-ground management decisions, to create a closed-loop system where actions (e.g., grazing moves, fencing decisions, mitigation measures) are continuously evaluated against measurable ecological and operational outcomes.​

​

Priority Outcomes

  1. Wildlife Connectivity: Reduced physical fencing barriers via virtual fencing, corridor mapping from space, and ground verification of wildlife use.

  2. Soil Health & Biodiversity: Adaptive grazing tied to satellite vegetation proxies and bioacoustic indicators (grassland birds).

  3. Water & Climate Resilience: Riparian monitoring, fisheries protection, and beaver-mediated hydrologic restoration.

  4. Carnivore Coexistence: Early detection and proactive mitigation to reduce conflict while supporting lawful population management.

  5. Science-Based Wolf Monitoring: Cost-effective acoustic monitoring to supplement or reduce reliance on aerial surveys and improve census confidence.

  6. Economic Resilience: Lower monitoring costs, reduced livestock loss, and exploration of biodiversity (payment-for-presence) and carbon credit mechanisms.

​​

Halter_logo_landscape_Dark_green.jpg
large_27b71c06-3e86-4716-830f-bb7eec3220da.png
logo.png
images.png
SMSP-Final-Logo-hor-hs.png
LoneStar-Tracking-Logo-2021-1-500x223.png
TNCLogoPrimary_RGB.jpg

Background

Holland Ranch Land & Livestock (HRL) and the Holland Ranch Company are a multi-generational family ranching operation rooted in Southwest Montana (18,937 private acres). The family began ranching in 1949, settling in Grasshopper Valley—near Montana's first territorial capital, Bannack—after relocating from Howe, Idaho. In those early years, the grandparents built the operation through hard, physical work: the grandfather breaking and chasing wild horses in the desert, the grandmother cooking for large haying crews. From the beginning, the ranch has operated as a commercial cow–calf enterprise, a role it has sustained for decades.

​

Today, the ranch runs 1,750 mother cows, depending on year and conditions, as a commercial cow–calf operation supported by approximately 100 bulls. Its grazing footprint spans a complex mix of land tenures, including four U.S. Forest Service allotments, six Bureau of Land Management (BLM) allotments, and three private leases—two of which include associated BLM ground. In addition to livestock production, the ranch supports recreational use through a long-standing partnership with Walker Outfitting, which has conducted controlled elk, deer, and antelope hunts on the property for more than twenty years.

​

The operation relies on both private and public lands, particularly during the summer grazing season. Approximately 1,450 head typically graze public lands across Forest Service (83,672 acres), BLM (23,939 acres), and Montana State DNRC holdings (2,459 acres) during those months. The ranch places high value on its working relationships with public land agencies and views grazing as a management tool that, when applied carefully, can support healthy soils, resilient rangelands, and wildlife habitat alongside livestock production.

​

Range Management Philosophy

The ranch’s approach to land stewardship is grounded in formal training as well as lived experience. Koy Holland grew up immersed in range management, mentored early on by family friend Sam Short—one of the founders of Montana Range Days in 1977. As a youth, Holland traveled across Montana, South Dakota, Wyoming, Nebraska, and Oklahoma participating in range judging, developing both technical skill and a lasting passion for rangeland ecology. He later earned a Bachelor of Science in Range Management (Resource Option), with a minor in Animal Science, from Montana State University, before returning home to apply academic training directly to the realities of a working ranch.

​

That philosophy has translated into concrete infrastructure and management decisions. In 2016, the ranch developed a rotational grazing system for yearling replacement heifers in collaboration with a neighboring operation and several BLM allotments. This effort included installing approximately nine miles of waterline supplied by three wells and nine tanks. Two of the wells are powered by remotely started propane engines driving an 8,000-watt generator mounted on a mobile trailer system with a 100-gallon propane tank. The setup is fully programmable, allowing the system to start and stop up to eight times per day and to shut down automatically via pressure switches, minimizing runtime and fuel use. During a typical three-month grazing season, the propane tank is refilled roughly four times.

​

Wildlife, Conservation, and Monitoring

The ranch occupies a biologically rich landscape that supports trout, pronghorn, elk, moose, wolves, grizzly bears, and some of Southwest Montana’s most important sage-grouse habitat, along with a wide array of breeding grassland birds. The ranch operates with an explicitly wildlife-friendly ethic, recognizing that where water is available, ecological life follows.

​

Looking forward, the operation is pursuing advanced grazing strategies aimed at carbon sequestration and increased biodiversity, with particular attention to grassland bird communities. These efforts are being paired with rigorous monitoring to document outcomes rather than rely on assumptions. The ranch is working with Cultivo to support soil carbon and biodiversity objectives, and plans to integrate ground-based sensors from Grizzly Systems and livestock monitoring platforms such as Halter and Spot, alongside satellite imagery from Planet, combined with on-the-ground verification by Cultivo’s field teams.

​

Recently, the ranch submitted a proposal to the Southwest Montana Sagebrush Partnership, with support from The Nature Conservancy, and was selected for both components of the application: virtual fencing and GPS ear tags. Through these partnerships, the Holland Ranch is positioning itself as a real-world testbed for integrating modern technology into everyday ranch operations—demonstrating not just what is possible, but what is practical, measurable, and transferable to other family-run ranches across the West.

​

Key Partners

 

Grizzly Systems (GrizCam.com)

Grizzly Systems provides the core ground-based sensing and intelligence layer for the Holland Ranch project. GrizCam units are deployed for both real-time operational awareness and long-term evaluation of regenerative grazing outcomes, as well as for predator deterrence trials.

​

In day-to-day operations, GrizCam supports AI-triggered detection of wildlife, livestock, and human activity for surveillance and situational awareness across the ranch’s expansive grazing footprint. Integrated chemical sensors are used to detect early fire indicators, with planned sensor fusion linking these detections to satellite data from Planet for cross-validation from space. When combined with near–real-time livestock location data from Halter collars, GPS ear tags, and SPOT devices, the system provides a unified, landscape-scale view of activity across private and public grazing lands.

​

Longer-term applications focus on bioacoustics monitoring, using AI-driven analysis of soundscapes to quantify biodiversity change over time through established acoustic indices. In parallel, the project will trial tightly controlled, research-grade experiments—under academic oversight—using GrizCam’s smart acoustic and visual triggers to initiate real-time deterrents. These include automated audio playbacks and LED signaling on Halter collars, designed to discourage predators from approaching livestock while minimizing disturbance to non-target wildlife.

​

Planet (Planet.com)

Planet’s participation centers on its Project Centenial and Digital Public Goods initiatives, which provide high-resolution, high-frequency satellite imagery and environmental analytics to support biodiversity and soil-health monitoring in at-risk landscapes.

​

By fusing Planet’s satellite-derived data and Planetary Variables® with ground-level observations from GrizCam, Halter, SPOT, and GPS ear tags, the Holland Ranch project gains a multi-scale view of grazing impacts over time. This integration allows grazing decisions and ecological outcomes—such as vegetation recovery, landscape use, and seasonal change—to be evaluated with a level of accuracy that neither satellites nor ground sensors could achieve alone.

​

Halter (Halterhq.com)

Halter provides the virtual fencing system used to manage livestock distribution and grazing timing without permanent physical infrastructure. During severe drought conditions in 2021, the ranch rested the system entirely. In subsequent years, herd size within the system has been intentionally reduced—from roughly 280 head to approximately 230—to ensure animal welfare and effective learning.

​

Replacement heifers are initially introduced to the system by pairing them with experienced cattle for the first month, accelerating adaptation and reducing stress. Looking ahead, the ranch plans to expand its use of Halter collars as water infrastructure is further developed, enabling a twice-through grazing system with deferred rotation. One pasture will remain excluded during specific periods due to the presence of tall larkspur, which has proven lethal to cattle, underscoring how virtual fencing decisions are informed by both ecological and animal-health constraints.

​

Lonestar GPS Ear Tags (lonestartracking.com)

GPS ear tags form a critical component of livestock tracking and management on the ranch. After extensive evaluation led by Pat Fosse, and guided by clearly defined operational goals, the ranch selected Lonestar GPS ear tags. The initial plan focused on tagging heifer calves grazing the ranch’s largest Forest Service allotment, an approach developed collaboratively with Forest Service managers and supported by the Southwest Montana Sagebrush Partnership.

​

At branding, 250 heifer calves were tagged across two groups—100 in one bunch and 150 in another—allowing the project to evaluate tag performance, retention, and data quality. Heifers were selected because the tags can be removed at weaning during Bangs vaccination, enabling reuse in subsequent years and providing insight into retention rates under real working conditions.

​

The operational value of the tags became immediately clear during a cattle drive spanning roughly ten miles, when a mishap resulted in several calves being lost in dense willow bottoms. With limited labor available, the crew relied on GPS data that evening to identify likely locations, significantly narrowing search efforts and improving recovery efficiency.

​

Livestock Guard Dogs & SPOT GPS Collars (finemespot.com)

Livestock guard dogs equipped with SPOT GPS collars add another layer of spatial awareness, particularly on Forest Service allotments where predator pressure is highest. As cattle are gathered or redistributed, the location data from these collars helps crews anticipate where livestock and guard animals are concentrated, improving response time and reducing uncertainty in rugged terrain.

​

Cultivo (Cultivo.land)

Cultivo leads the project’s soil health, carbon, and biodiversity strategy. Its role centers on combining AI, environmental sensing, and rigorous ground-truthing to identify areas with high potential for ecological regeneration and long-term value creation.

​

Working closely with the Holland Ranch, Cultivo is helping design and implement an adaptive grazing plan that treats livestock as a restoration tool—balancing animal impact with adequate rest and recovery. The primary objectives are to restore soil function, increase resilience and productivity of working lands, and sequester carbon through improved grazing practices. Progress is measured through changes in Soil Organic Carbon (SOC), a key component of Soil Organic Matter that reflects long-term carbon storage.

​

In collaboration with Grizzly Systems, Cultivo will also support biodiversity monitoring on public land grazing allotments, particularly in gravity-fed watering areas on BLM lands that lie within some of Southwest Montana’s premier sage-grouse habitat. Bioacoustic monitoring will be used to track grassland bird responses to grazing management, with the explicit goal of demonstrating measurable biodiversity gains. Together, the partners will model and trial a biodiversity credit framework, exploring its potential as an additional revenue stream grounded in verified ecological outcomes rather than assumptions.

Bioacoustics Protocol for Biodiversity

This below defines a streamlined, scientifically defensible protocol for evaluating whether adaptive cattle grazing under the Living Range Project maintains or improves grassland bird biodiversity in southwest Montana. The protocol prioritizes species-aware AI-based acoustic detection as the primary evidence for biodiversity outcomes. Bioacoustic indices are retained only as secondary diagnostics and covariates, not as direct measures of biodiversity.

​

1. Scientific Objective

The primary objective is to test whether adaptive grazing practices designed to promote healthy grassland structure are at least non-degrading to grassland bird biodiversity. Biodiversity is defined explicitly in ecological terms: species occupancy, species richness, and community stability. The protocol is designed to support defensible claims suitable for scientific review, agency evaluation, and public-lands management decisions.

​

2. Core Study Design

The study follows a Before–After Control–Impact (BACI) or stepped-wedge design. Treatment units are pastures managed under the Living Range adaptive grazing protocol. Control units are comparable pastures managed under delayed, conventional, or temporarily ungrazed conditions. All pastures are monitored simultaneously to separate grazing effects from interannual climate variability.

​

Each pasture is instrumented with multiple fixed autonomous acoustic recorders (GrizCam units). Recorders remain in place year‑round and collect continuous audio (24×7). Analyses focus on biologically meaningful temporal windows tied to avian behavior.

​

3. Primary Data Stream: AI-Based Bird Detection

Passive acoustic recordings are processed using an AI-based bird vocalization detection and classification system. The system produces probabilistic detections with species labels, confidence scores, timestamps, and frequency bounds.

​

3.1 Target Species Definition

Prior to analysis, a target species list is defined. The list emphasizes grassland and sagebrush obligate birds and common associates whose vocalizations are frequent and acoustically distinct. Rare, cryptic, or poorly classifiable species are excluded to avoid biased inference. Sage-grouse detections are treated as a focal, species-specific analysis.

​

3.2 Detection Workflow

Audio is segmented into standardized clips. A two-stage AI pipeline is applied: (1) an event detector identifies candidate bird vocalizations, and (2) a multi-class classifier assigns species probabilities. All detections retain confidence scores and are used probabilistically rather than thresholded into simple presence/absence.

​

4. Validation and Quality Control

A stratified subset of recordings is manually annotated each season to quantify classifier precision, recall, and error modes. Validation sampling is stratified by pasture, treatment status, time of day, and weather conditions. Classifier performance is tracked longitudinally to detect domain shift caused by vegetation change, insect emergence, or grazing activity.

​

If performance degradation is detected, targeted retraining is conducted using newly labeled data, and analyses are rerun.

​

5. Biodiversity Metrics (Primary Endpoints)

Biodiversity outcomes are estimated using hierarchical statistical models that account explicitly for imperfect detection. Primary endpoints include:

- Species occupancy (ψ) for focal species
- Community-level species richness
- Community composition stability over time

Multi-species occupancy models are used where feasible, allowing information sharing across species and improving inference for less frequently detected taxa. Detection probability is modeled as a function of time of day, season, weather, and background noise.

​

6. Inference Framework: Demonstrating Non-Degradation

The protocol adopts a non-inferiority framework. Prior to analysis, acceptable non-inferiority margins (Δ) are defined for key endpoints (e.g., expected richness, focal-species occupancy). Adaptive grazing is considered non-degrading if estimated treatment effects remain above −Δ relative to controls in BACI or stepped-wedge models.

​

7. Secondary Diagnostics: Acoustic Indices

Acoustic indices (e.g., ACI, entropy, NDSI) are computed only as secondary diagnostics. They are used to:

- Identify wind, rain, insect, or anthropogenic noise contamination
- Serve as covariates in detection models
- Flag potential data-quality issues

Indices are not used directly as measures of biodiversity.

​

8. Robustness and Falsification Checks

Analyses must pass the following robustness checks:

- Consistent results under low-wind subsets
- Placebo treatment dates applied to control pastures
- Stability after excluding high-noise periods

​

9. Recommended Acoustic Sampling Protocol

The following recording and analysis settings are recommended to balance ecological relevance, classifier performance, and computational efficiency:

- Recording mode: Continuous (24×7)
- File format: Uncompressed WAV
- Sample rate: 24–48 kHz (minimum 24 kHz)
- Effective analysis band: 0.3–10 kHz
- Clip length for AI analysis: 30–60 seconds
- Primary analysis windows:
  • Dawn chorus: civil twilight to +2 hours (breeding season)
  • Lekking windows for sage-grouse (species-specific seasonal timing)
  • Early morning and late afternoon windows during brood-rearing
- Secondary windows: sparse daytime and nighttime subsampling for disturbance detection

​

10. Interpretation and Claims

Claims are limited to outcomes supported by modeled biodiversity metrics with quantified uncertainty. The strongest allowable claim is that adaptive grazing under the Living Range Project maintains or improves grassland bird biodiversity relative to controls, within predefined uncertainty bounds.

​

References

Bradfer-Lawrence, T., Gardner, N., Bunnefeld, N., Willis, S. G., & Dent, D. H. (2024). The acoustic index user’s guide: A practical manual for defining and generating acoustic indices. Methods in Ecology and Evolution, 15(1), 1–14.

Buxton, R. T., et al. (2018). Efficacy of passive acoustic monitoring as a conservation tool. Biological Conservation, 224, 44–53.

MacKenzie, D. I., Nichols, J. D., Lachman, G. B., Droege, S., Royle, J. A., & Langtimm, C. A. (2002). Estimating site occupancy rates when detection probabilities are less than one. Ecology, 83(8), 2248–2255.

MacKenzie, D. I., et al. (2017). Occupancy estimation and modeling: Inferring patterns and dynamics of species occurrence (2nd ed.). Academic Press.

 

Rhinehart, T. A., Turek, D., & Kitzes, J. (2022). A continuous-score occupancy model that incorporates uncertain machine learning predictions. Methods in Ecology and Evolution, 13(10), 2138–2151.

Stowell, D., et al. (2019). Automatic acoustic detection of birds through deep learning: The first Bird Audio Detection challenge. Methods in Ecology and Evolution, 10(3), 368–380.

Sueur, J., Farina, A., Gasc, A., Pieretti, N., & Pavoine, S. (2014). Acoustic indices for biodiversity assessment and landscape investigation. Acta Acustica United with Acustica, 100(4), 772–781.

bottom of page